From boris at codesynthesis.com  Tue Sep  1 07:50:49 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Tue Sep  1 07:50:53 2015
Subject: [odb-users] Segmentation fault while executing odb compiler
	avoided adding a typedef sentence
In-Reply-To: <CA+Jb2D06YrVkCNq1u+MZta=XNrV+PXFr7_Wh2bTBzpnexouWRQ@mail.gmail.com>
References: <CA+Jb2D06YrVkCNq1u+MZta=XNrV+PXFr7_Wh2bTBzpnexouWRQ@mail.gmail.com>
Message-ID: <boris.20150901134920@codesynthesis.com>

MM <finjulhich@gmail.com> writes:

> As I have the odb sources now, i am able to quickly test out any fix you
> may think of.

Before I can start thinking of a fix, I need a test case that reproduces
the problem. Ideally, an archive with all the necessary headers and the
ODB command line that you used.

Thanks,
Boris

From markus at markusklemm.net  Tue Sep  1 11:40:07 2015
From: markus at markusklemm.net (Markus Klemm)
Date: Tue Sep  1 11:40:24 2015
Subject: [odb-users] ORDER BY clause in ODB query language
In-Reply-To: <boris.20150827140741@codesynthesis.com>
References: <B558C998-508E-4F20-BA05-DF6C98A3D1BB@markusklemm.net>
	<boris.20150827140741@codesynthesis.com>
Message-ID: <2FE30C1A-F0F9-4594-99FD-6F454B0D2A5D@markusklemm.net>

I was close to start a new thread, but it basicly the same topic:
My point in the previous question was, that I can't solve this use case below, without choosing the column by a string, rather than by a odb::query_column,yet?!

I didn't minimized it completly because @Boris was wondering about the use cases. So obvouisly I'm not a fan of the one suggested future solution of using operator<, because ordering criteria changes/ depens on the column. This is the real usecase when using the framework Wt (e.g. part of the user supplied model for a table view).

Raw Sourcecode below, gist here: https://gist.github.com/Superlokkus/a994d42534beca9ed5c6
void sort(int column, Wt::SortOrder order = Wt::AscendingOrder){
	odb::query<directory> first_part("ORDER BY"), middle_part,end_part;
	//odb::query_column<directory> middle_type; and decltype(odb::query<directory>::added) middle_type; were my next guesses
	switch (column){
	default:
		middle_part = odb::query<directory>::added;/*Error: no operator "=" matches these operands 
operand types are : odb::query<directory, odb::mssql::query_base> = const odb::mssql::query_column<boost::posix_time::ptime, odb::mssql::id_datetime>*/
		break;
	}
	switch (order){
	case Wt::DescendingOrder:
		end_part = odb::query<directory>("DESC");
		break;
	default:
		end_part = odb::query<directory>("ASC");
		break;
	}
	odb::query<directory> final_query(first_part + middle_part + end_part);
	}

Best Regards

Markus Klemm

> Am 27.08.2015 um 14:14 schrieb Boris Kolpackov <boris@codesynthesis.com>:
> 
> Hi Markus,
> 
> Markus Klemm <markus@markusklemm.net> writes:
> 
>> Is there a way to get a query with an ORDER BY clause with the typed
>> checked, static, non-native ODB query language? The examples in the
>> manual is using the database system-native query language e.g. a
>> concatenated string.
> 
> Currently you have to spell "ORDER BY" as a string, though the column
> can be specified as a C++ name:
> 
> db->query<person> ((query::first == "John") + "ORDER BY" + query::age);
> 
> We do plan to provide some syntactic sugar for this, though I am not
> sure how much static type checking we will be able to do. For example,
> should we assume if the C++ type provides operator<, then we can do
> ORDER BY?
> 
> 
>> That also makes it hard to use the ORDER BY clause because after
>> the odb macro for the column, I need to concatenate the DESC/ASC
>> clause but can?t due type mismatch.
> 
> This should work:
> 
> db->query<person> ((...) + "ORDER BY" + query::age + "ASC");
> 
> BTW, "the odb macro for the column" is not a macro, it is a proper
> C++ variable name.
> 
> Boris

From finjulhich at gmail.com  Tue Sep  1 16:21:04 2015
From: finjulhich at gmail.com (MM)
Date: Tue Sep  1 16:21:11 2015
Subject: [odb-users] Segmentation fault while executing odb compiler
	avoided adding a typedef sentence
Message-ID: <CA+Jb2D2KT6URXB35Q=pAW+hMP4RFtaEZFmMWzGJTjfdYz=jRhQ@mail.gmail.com>

On 1 September 2015 at 12:50, Boris Kolpackov <boris@codesynthesis.com>
 wrote:

> MM <finjulhich@gmail.com> writes:
>
> > As I have the odb sources now, i am able to quickly test out any fix you
> > may think of.
>
> Before I can start thinking of a fix, I need a test case that reproduces
> the problem. Ideally, an archive with all the necessary headers and the
> ODB command line that you used.
>
> Thanks,
> Boris
>

Please find attached 5 c++ files. Donwload them in the same directory and
run:

/usr/local/bin/odb --std c++11 -I/usr/local/include --database sqlite
-I/usr/include derived_odb.hpp

with odb built off master.

This crashes.

Please let me know if I missed to post something, Thanks,
-------------- next part --------------
A non-text attachment was scrubbed...
Name: base.hpp
Type: text/x-c++hdr
Size: 145 bytes
Desc: not available
Url : http://codesynthesis.com/pipermail/odb-users/attachments/20150901/bad07942/base.hpp
-------------- next part --------------
A non-text attachment was scrubbed...
Name: base_odb.hpp
Type: text/x-c++hdr
Size: 119 bytes
Desc: not available
Url : http://codesynthesis.com/pipermail/odb-users/attachments/20150901/bad07942/base_odb.hpp
-------------- next part --------------
A non-text attachment was scrubbed...
Name: cloneable.hpp
Type: text/x-c++hdr
Size: 219 bytes
Desc: not available
Url : http://codesynthesis.com/pipermail/odb-users/attachments/20150901/bad07942/cloneable.hpp
-------------- next part --------------
A non-text attachment was scrubbed...
Name: derived.hpp
Type: text/x-c++hdr
Size: 92 bytes
Desc: not available
Url : http://codesynthesis.com/pipermail/odb-users/attachments/20150901/bad07942/derived.hpp
-------------- next part --------------
A non-text attachment was scrubbed...
Name: derived_odb.hpp
Type: text/x-c++hdr
Size: 206 bytes
Desc: not available
Url : http://codesynthesis.com/pipermail/odb-users/attachments/20150901/bad07942/derived_odb.hpp
From yoav.weiss at intel.com  Wed Sep  2 07:42:52 2015
From: yoav.weiss at intel.com (Weiss, Yoav)
Date: Wed Sep  2 07:43:03 2015
Subject: [odb-users] A view on top of several tables from different sqlite
	files (DB)
Message-ID: <E650D64AEB646045AE83BF4F2E97B8482E5306FC@HASMSX110.ger.corp.intel.com>

Hi,

My question is related to views with multiple tables like:
#pragma db view object (Cell = c) \
                object (Instance = i : c::m_id == i::m_cell_id) \
                query((?) + " GROUP BY " + c::m_name)
struct CellOccCount {
    #pragma db column("c.cell_name")
    std::string m_name;

    #pragma db column ("count(*)")
    long m_count;
};

And the way I query the view is:
odb::result<CellOccCount> results(m_p_db->query<CellOccCount>(query_cond));

Now suppose these 2 tables (Cell and Instance) are located in 2 different sqlite databases. My questions are:

1.       How should I modify the view (if at all) to reflect this?

2.       On which odb::database should I initiate the query (assuming I have 2 instances - one for each database)?

Thanks,

Yoav

---------------------------------------------------------------------
Intel Israel (74) Limited

This e-mail and any attachments may contain confidential material for
the sole use of the intended recipient(s). Any review or distribution
by others is strictly prohibited. If you are not the intended
recipient, please contact the sender and delete all copies.
From boris at codesynthesis.com  Wed Sep  2 11:31:54 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Wed Sep  2 11:31:58 2015
Subject: [odb-users] A view on top of several tables from different
	sqlite files (DB)
In-Reply-To: <E650D64AEB646045AE83BF4F2E97B8482E5306FC@HASMSX110.ger.corp.intel.com>
References: <E650D64AEB646045AE83BF4F2E97B8482E5306FC@HASMSX110.ger.corp.intel.com>
Message-ID: <boris.20150902172345@codesynthesis.com>

Hi Yoav,

Weiss, Yoav <yoav.weiss@intel.com> writes:

> #pragma db view object (Cell = c) \
>                 object (Instance = i : c::m_id == i::m_cell_id) \
>                 query((?) + " GROUP BY " + c::m_name)
> struct CellOccCount {
>     #pragma db column("c.cell_name")
>     std::string m_name;
> 
>     #pragma db column ("count(*)")
>     long m_count;
> };
> 
> And the way I query the view is:
> odb::result<CellOccCount> results(m_p_db->query<CellOccCount>(query_cond));
> 
> Now suppose these 2 tables (Cell and Instance) are located in 2 different
> sqlite databases. My questions are:
> 
> 1. How should I modify the view (if at all) to reflect this?

The view doesn't need to be modified, but see #2 below.


> 2. On which odb::database should I initiate the query (assuming I have
>    2 instances - one for each database)?

No, that's not going to work. You cannot have a query span multiple
SQLite connections. Instead, what you will have to do is attach the
second database file to the existing connection with the SQLite
ATTACH DATABASE statement:

https://www.sqlite.org/lang_attach.html

In ODB you can use execute(), for example:

m_p_db->execute ("ATTACH DATABASE cell.sqlite AS cells");

If you don't have identically names tables in the two databases, then
you should be all set. Otherwise, you would have to qualify the table
names, for example:

#pragma db object table("cells.Cell")
class Cell
{
  ...
};

Boris

From markus at markusklemm.net  Thu Sep  3 09:32:33 2015
From: markus at markusklemm.net (Markus Klemm)
Date: Thu Sep  3 09:32:48 2015
Subject: [odb-users] Query object A by range of objects B, with 1:n relation
Message-ID: <91197426-FA08-473F-89F5-BA264AFC1379@markusklemm.net>

I have a problem using the odb query language with a one to many relationship.
Suppose I have a persistent class file and directory. These have a bidirection one-to-many relationship. To query a set of files with criteria of one directory I successfully used the directory member of the query<file> class. But how do I use it with a range of directories, i.e. all files from this set of directories.

To get more concrete, said I have the definitions:

#pragma db object table("directories") session pointer(std::shared_ptr<directory>)
struct directory{
#pragma db column("directory_id") type("uniqueidentifier") id not_null
	boost::uuids::uuid directory_id;
#pragma db inverse(directory)
	std::vector<std::shared_ptr<file>> files;}};

#pragma db object table("files") session pointer(std::shared_ptr<file>)
struct file{
#pragma db column("file_id") type("uniqueidentifier") id not_null
	boost::uuids::uuid file_id;
#pragma db column("directory_id") not_null
	std::weak_ptr<directory> directory;};

So for one directory the query is obvious: query<file>(query<file>::directory == query<file>::_val(directory_instance)).
But the two solutions I've come to have 2 different problems: 
Solution 1: 
set<decltype(directory::directory_id)> directories;
odb::query<file>::directory->directory_id.in_range(directories.cbegin(),directories.cend());
Fails with a sql exception, "wrong syntax near directory_id"
Solution 2:
std::set<directory> directories
odb::query<file>::directory.in_range(directories.cbegin(),directories.cend());
Can't compile because there is, I assume, no overload of in_range() for directory. (Error is 'odb::mssql::val_bind<T>::val_bind' : none of the 2 overloads could convert all the argument types)

Can I solve this with the odb query language?

Thanks in advance and

best regards/Mit freundlichen Gr??en

Markus Klemm

From finjulhich at gmail.com  Thu Sep  3 10:18:42 2015
From: finjulhich at gmail.com (MM)
Date: Thu Sep  3 10:18:48 2015
Subject: [odb-users] Segmentation fault while executing odb compiler
	avoided adding a typedef sentence
Message-ID: <CA+Jb2D1hbZ64S5n5GAHUy2kOq6rbLtG2TJrnb1mCJ6oGXhBQdw@mail.gmail.com>

Just checking my previous post about the reproducer arrived to the mailing
list.
I am having gmail issues, it's rejecting some sends becaus it's thinking
i'm spaming:-)

Thanks
From boris at codesynthesis.com  Thu Sep  3 10:50:37 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Thu Sep  3 10:50:46 2015
Subject: [odb-users] Query object A by range of objects B, with 1:n
	relation
In-Reply-To: <91197426-FA08-473F-89F5-BA264AFC1379@markusklemm.net>
References: <91197426-FA08-473F-89F5-BA264AFC1379@markusklemm.net>
Message-ID: <boris.20150903164710@codesynthesis.com>

Hi Markus,

Markus Klemm <markus@markusklemm.net> writes:

> set<decltype(directory::directory_id)> directories;
> odb::query<file>::directory->directory_id.in_range(directories.cbegin(),directories.cend());
> Fails with a sql exception, "wrong syntax near directory_id"

I believe you are getting this error because the range is empty. This,
BTW, has been fixed for the next release of ODB.

Also, chances are, your next question will be how to find files
that belong to directories that in turn match a certain query
condition. For that you would need to use ODB views (Chapter 10).

Boris

From markus at markusklemm.net  Thu Sep  3 11:09:49 2015
From: markus at markusklemm.net (Markus Klemm)
Date: Thu Sep  3 11:10:04 2015
Subject: [odb-users] Query object A by range of objects B,
	with 1:n relation
In-Reply-To: <boris.20150903164710@codesynthesis.com>
References: <91197426-FA08-473F-89F5-BA264AFC1379@markusklemm.net>
	<boris.20150903164710@codesynthesis.com>
Message-ID: <DACD9253-FCE8-45A6-8205-39506A74A397@markusklemm.net>

I figured out the problem in the instant you replied: Solution 1 was right, but I accidentialy built the final query with the order query first.

But thank you very much for the odb view tip, I stick very close to the documentation, and read it thru first, but I wasn't aware/forggot that views are the way to go for this transitive purpose: Thank you and I'm sorry.

Good day and Mit freundlichen Gr??en

Markus Klemm



> Am 03.09.2015 um 16:50 schrieb Boris Kolpackov <boris@codesynthesis.com>:
> 
> Hi Markus,
> 
> Markus Klemm <markus@markusklemm.net> writes:
> 
>> set<decltype(directory::directory_id)> directories;
>> odb::query<file>::directory->directory_id.in_range(directories.cbegin(),directories.cend());
>> Fails with a sql exception, "wrong syntax near directory_id"
> 
> I believe you are getting this error because the range is empty. This,
> BTW, has been fixed for the next release of ODB.
> 
> Also, chances are, your next question will be how to find files
> that belong to directories that in turn match a certain query
> condition. For that you would need to use ODB views (Chapter 10).
> 
> Boris

From boris at codesynthesis.com  Fri Sep  4 09:31:47 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Fri Sep  4 09:31:56 2015
Subject: [odb-users] Segmentation fault while executing odb compiler
	avoided adding a typedef sentence
In-Reply-To: <CA+Jb2D2KT6URXB35Q=pAW+hMP4RFtaEZFmMWzGJTjfdYz=jRhQ@mail.gmail.com>
References: <CA+Jb2D2KT6URXB35Q=pAW+hMP4RFtaEZFmMWzGJTjfdYz=jRhQ@mail.gmail.com>
Message-ID: <boris.20150904152242@codesynthesis.com>

MM <finjulhich@gmail.com> writes:

> Please find attached 5 c++ files.

Thanks for the test. The problem is again with the order in which
ODB "sees" C++ classes. I've added some more diagnostics (available
in master) with which your test case produces:

derived.hpp:4:8: error: base class derived_from_base must be defined before derived class Derived
derived_odb.hpp:5:12: info: class derived_from_base is define here

I've re-arranged the order of things in your derived_odb.hpp to be:

#include "base_odb.hpp"

#include "cloneable.hpp"
struct Derived;
using derived_from_base = cloneable<Base,Derived>;
#pragma db object(derived_from_base) abstract definition

#include "derived.hpp"
#pragma db object(Derived) table("TD") definition

That and adding '#pragma once' to cloneable.hpp fixes the error and
even produces compilable code.

If you meditate on it, the order in which things are now included/defined
actually makes quite a bit of sense.

Boris

From boris at codesynthesis.com  Fri Sep  4 09:37:55 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Fri Sep  4 09:38:04 2015
Subject: [odb-users] Segmentation fault while executing odb compiler
	avoided adding a typedef sentence
In-Reply-To: <CA+Jb2D1hbZ64S5n5GAHUy2kOq6rbLtG2TJrnb1mCJ6oGXhBQdw@mail.gmail.com>
References: <CA+Jb2D1hbZ64S5n5GAHUy2kOq6rbLtG2TJrnb1mCJ6oGXhBQdw@mail.gmail.com>
Message-ID: <boris.20150904153652@codesynthesis.com>

MM <finjulhich@gmail.com> writes:

> Just checking my previous post about the reproducer arrived to the mailing
> list.

You can always check the archives:

http://www.codesynthesis.com/pipermail/odb-users/

Boris

From boris at codesynthesis.com  Fri Sep  4 10:41:58 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Fri Sep  4 10:42:06 2015
Subject: [odb-users] ORDER BY clause in ODB query language
In-Reply-To: <2FE30C1A-F0F9-4594-99FD-6F454B0D2A5D@markusklemm.net>
References: <B558C998-508E-4F20-BA05-DF6C98A3D1BB@markusklemm.net>
	<boris.20150827140741@codesynthesis.com>
	<2FE30C1A-F0F9-4594-99FD-6F454B0D2A5D@markusklemm.net>
Message-ID: <boris.20150904163653@codesynthesis.com>

Hi Markus,

Markus Klemm <markus@markusklemm.net> writes:

> Raw Sourcecode below [...]

Wouldn't something like this work:

using dir_query = odb::query<directory>;

dir_query q ("ORDER BY");

switch (column)
{
case added:
  q += dir_query::added;
...
}

switch (order)
{
case desc:
  q += "DESC";
...
}

db.query (q);

Boris

From finjulhich at gmail.com  Fri Sep  4 16:40:29 2015
From: finjulhich at gmail.com (MM)
Date: Fri Sep  4 16:40:36 2015
Subject: [odb-users] Segmentation fault while executing odb compiler
	avoided adding a typedef sentence
Message-ID: <CA+Jb2D2cshG8y87zyD9NyKK7DAymgogsFBwEKE5S8uW6qcuJbQ@mail.gmail.com>

On 4 September 2015 at 14:31, Boris Kolpackov <boris@codesynthesis.com>
 wrote:

> MM <finjulhich@gmail.com> writes:
>
> > Please find attached 5 c++ files.
>
> Thanks for the test. The problem is again with the order in which
> ODB "sees" C++ classes. I've added some more diagnostics (available
> in master) with which your test case produces:
>
> derived.hpp:4:8: error: base class derived_from_base must be defined
> before derived class Derived
> derived_odb.hpp:5:12: info: class derived_from_base is define here
>
> I've re-arranged the order of things in your derived_odb.hpp to be:
>
> #include "base_odb.hpp"
>
> #include "cloneable.hpp"
> struct Derived;
> using derived_from_base = cloneable<Base,Derived>;
> #pragma db object(derived_from_base) abstract definition
>
> #include "derived.hpp"
> #pragma db object(Derived) table("TD") definition
>
> That and adding '#pragma once' to cloneable.hpp fixes the error and
> even produces compilable code.
>
> If you meditate on it, the order in which things are now included/defined
> actually makes quite a bit of sense.
>
> Boris
>


Indeed, reordering fixes the odb compiler crash.  My executable builds.
Post successful link, I generate the sql once with odb.

I have the following snippet that tries to adapt
boost::gregorian::date_period as I don't see it in the profile..

namespace boost
{
namespace gregorian
{

#pragma db value(date_period) transient definition
#pragma db member(date_period::begin) virtual(date) get(begin())\
  set(this = ::boost::gregorian::date_period((?), this.end()))
#pragma db member(date_period::end) virtual(date) get(end())\
  set(this = ::boost::gregorian::date_period(this.begin(), (?)))

}
}

odb barks for the same reason:-)

Is this why only date is in the boost/date_time subprofile :-)

Any workaround you think of? My onw wrapper class around date_period?

Thanks
MM
From soroush.rabiei at gmail.com  Sun Sep  6 06:53:31 2015
From: soroush.rabiei at gmail.com (Soroush Rabiei)
Date: Sun Sep  6 06:53:58 2015
Subject: [odb-users] nullable unsigned variables
Message-ID: <CAOMsk2QH_7QeTuw8Mu-pSmH06i7SUi-0oHoc8V0aeZ+svhAMBg@mail.gmail.com>

Hi

I have a problem using nullable types aka `odb::nullable'.  It seems it
can't handle `unsigned' variables. Here is a small test and it's output
error:

#include <odb/core.hxx>
> #include <odb/nullable.hxx>
> class foo
> {
>     friend class odb::access;
>     private:
>     odb::nullable<unsigned short int> x;
> };
> #pragma db object(foo) no_id table("foo_table")
> #pragma db member(foo::x) column("X")


I'm getting these errors:

g++ --std=c++11 -c -Wall -Wextra -Wno-unknown-pragmas --pedantic
> test-odb.cxx
> test-odb.cxx: In static member function ?static void
> odb::access::object_traits_impl<foo,
> (odb::database_id)3u>::init(odb::access::object_traits_impl<foo,
> (odb::database_id)3u>::image_type&, const object_type&,
> odb::oracle::statement_kind)?:
> test-odb.cxx:70:30: error: no matching function for call to
> ?odb::oracle::value_traits<odb::nullable<short unsigned int>,
> (odb::oracle::database_type_id)0u>::set_image(int&, bool&, const
> odb::nullable<short unsigned int>&)?
> test-odb.cxx:70:30: note: candidates are:
> In file included from test-odb.cxx:13:0:
> /usr/include/odb/oracle/traits.hxx:351:7: note: static void
> odb::oracle::wrapped_value_traits<W, ID,
> true>::set_image(odb::oracle::wrapped_value_traits<W, ID,
> true>::image_type&, bool&, const W&) [with W = odb::nullable<short unsigned
> int>; odb::oracle::database_type_id ID = (odb::oracle::database_type_id)0u;
> odb::oracle::wrapped_value_traits<W, ID, true>::image_type = unsigned int]
> /usr/include/odb/oracle/traits.hxx:351:7: note:   no known conversion for
> argument 1 from ?int? to
> ?odb::oracle::wrapped_value_traits<odb::nullable<short unsigned int>,
> (odb::oracle::database_type_id)0u, true>::image_type& {aka unsigned int&}?
> /usr/include/odb/oracle/traits.hxx:373:7: note: static void
> odb::oracle::wrapped_value_traits<W, ID, true>::set_image(char*,
> std::size_t, std::size_t&, bool&, const W&) [with W = odb::nullable<short
> unsigned int>; odb::oracle::database_type_id ID =
> (odb::oracle::database_type_id)0u; std::size_t = long unsigned int]
> /usr/include/odb/oracle/traits.hxx:373:7: note:   candidate expects 5
> arguments, 3 provided
> /usr/include/odb/oracle/traits.hxx:388:7: note: static void
> odb::oracle::wrapped_value_traits<W, ID, true>::set_image(char*,
> std::size_t&, bool&, const W&) [with W = odb::nullable<short unsigned int>;
> odb::oracle::database_type_id ID = (odb::oracle::database_type_id)0u;
> std::size_t = long unsigned int]
> /usr/include/odb/oracle/traits.hxx:388:7: note:   candidate expects 4
> arguments, 3 provided
> /usr/include/odb/oracle/traits.hxx:408:7: note: static void
> odb::oracle::wrapped_value_traits<W, ID, true>::set_image(bool (*&)(const
> void*, ub4*, const void**, ub4*, odb::oracle::chunk_position*, void*, ub4),
> const void*&, bool&, const W&) [with W = odb::nullable<short unsigned int>;
> odb::oracle::database_type_id ID = (odb::oracle::database_type_id)0u;
> odb::oracle::param_callback_type = bool (*)(const void*, unsigned int*,
> const void**, unsigned int*, odb::oracle::chunk_position*, void*, unsigned
> int); ub4 = unsigned int]
> /usr/include/odb/oracle/traits.hxx:408:7: note:   candidate expects 4
> arguments, 3 provided
> make: *** [test-odb.o] Error 1
From yoav.weiss at intel.com  Sun Sep  6 11:10:29 2015
From: yoav.weiss at intel.com (Weiss, Yoav)
Date: Sun Sep  6 11:10:41 2015
Subject: [odb-users] A view on top of several tables from different
	sqlite files (DB)
In-Reply-To: <boris.20150902172345@codesynthesis.com>
References: <E650D64AEB646045AE83BF4F2E97B8482E5306FC@HASMSX110.ger.corp.intel.com>
	<boris.20150902172345@codesynthesis.com>
Message-ID: <E650D64AEB646045AE83BF4F2E97B8482E531EB5@HASMSX110.ger.corp.intel.com>

Sounds promising. I still have an issue though.

When I call m_p_db->execute ("ATTACH DATABASE cell.sqlite AS cells") I get:
operation can only be performed in transaction

So I did: m_p_db->connection()->execute ("ATTACH DATABASE cell.sqlite AS cells")
Which worked fine, but the first query (using m_p_db->query_one()) on a table inside cells DB, I got:
1: no such table: hier_version

I think this is caused by me using the connection pointer, but I'm not sure.

Table names are unique across databases

Yoav


-----Original Message-----
From: Boris Kolpackov [mailto:boris@codesynthesis.com] 
Sent: Wednesday, September 02, 2015 18:32
To: Weiss, Yoav
Cc: odb-users@codesynthesis.com
Subject: Re: [odb-users] A view on top of several tables from different sqlite files (DB)

Hi Yoav,

Weiss, Yoav <yoav.weiss@intel.com> writes:

> #pragma db view object (Cell = c) \
>                 object (Instance = i : c::m_id == i::m_cell_id) \
>                 query((?) + " GROUP BY " + c::m_name) struct 
> CellOccCount {
>     #pragma db column("c.cell_name")
>     std::string m_name;
> 
>     #pragma db column ("count(*)")
>     long m_count;
> };
> 
> And the way I query the view is:
> odb::result<CellOccCount> 
> results(m_p_db->query<CellOccCount>(query_cond));
> 
> Now suppose these 2 tables (Cell and Instance) are located in 2 
> different sqlite databases. My questions are:
> 
> 1. How should I modify the view (if at all) to reflect this?

The view doesn't need to be modified, but see #2 below.


> 2. On which odb::database should I initiate the query (assuming I have
>    2 instances - one for each database)?

No, that's not going to work. You cannot have a query span multiple SQLite connections. Instead, what you will have to do is attach the second database file to the existing connection with the SQLite ATTACH DATABASE statement:

https://www.sqlite.org/lang_attach.html

In ODB you can use execute(), for example:

m_p_db->execute ("ATTACH DATABASE cell.sqlite AS cells");

If you don't have identically names tables in the two databases, then you should be all set. Otherwise, you would have to qualify the table names, for example:

#pragma db object table("cells.Cell")
class Cell
{
  ...
};

Boris
---------------------------------------------------------------------
Intel Israel (74) Limited

This e-mail and any attachments may contain confidential material for
the sole use of the intended recipient(s). Any review or distribution
by others is strictly prohibited. If you are not the intended
recipient, please contact the sender and delete all copies.


From finjulhich at gmail.com  Sun Sep  6 17:19:34 2015
From: finjulhich at gmail.com (MM)
Date: Sun Sep  6 17:19:42 2015
Subject: [odb-users] 2-table query
Message-ID: <CADsB1iDM5J7FcTLwuKD5uij_rNi=i249bjkpnTHUGZmAxvWivQ@mail.gmail.com>

Hello,

I have the following 2 tables and their relevant columns:

M
==
id: PRIMARY KEY
month1: TEXT (encoded month,year combination)
month2: TEXT

F
==
id: PRIMARY KEY
mid: NOT NULL REFERENCES M("id")
month: TEXT

There can 0..N rows in F for each row in M.

My question is two-fold:

   1. How do I select all the F rows such that their month is 1 of the
   2months in the relevant M row (the one where F.mid=M.id)?
   2. How do I express this with an odb:query? Both M and F tables are the
   mapping of c++ classes.


db is sqlite. F has about 100 000 rows. The result of the query is about
200.

Thanks,
MM
From markus at markusklemm.net  Mon Sep  7 08:19:17 2015
From: markus at markusklemm.net (Markus Klemm)
Date: Mon Sep  7 08:19:32 2015
Subject: [odb-users] Inverse pragma using composite value types
Message-ID: <632BCDC6-2C6F-4D49-9968-21416609E465@markusklemm.net>

Hi again everyone,

I have the problem that the inverse pragma don't work with composite value types. I fails to parse the . operator correctly.
The problematic scenario is that object A's ID is part of object B's composite ID.
I'm of course not sure if ODB ever had in mind to offer that solution or if it is expected that I use a odb view.

Example (or link here for syntax highlighting https://gist.github.com/Superlokkus/44d69167ca1a82e150e9 )
Given tables:

create table measurement_set(
	measurement_id uniqueidentifier	primary key
)
create table measurement_parameter(
	measurement_set_id uniqueidentifier not null foreign key references measurement_set(id_col),
	parameter_name nvarchar(32) not null,
	--other columns
	constraint pk primary key(measurement_set_id,parameter_name)
)
I suppose the solution should be:

struct measurement_set;

#pragma db value
struct measurement_parameter_key{
#pragma db column("measurement_set_id") type("uniqueidentifier") not_null
	boost::uuids::uuid measurement_set_id;
#pragma db column("parameter_name") type("nvarchar(32)") not_null
	std::wstring parameter_name;
};

#pragma db object table("measurement_parameter")
struct measurement_parameter{
#pragma db id
	measurement_parameter_key measurement_set_id;
/* Other members*/
};

#pragma db object table("measurement_set")
struct measurement_set {
#pragma db column("measurement_id") type("uniqueidentifier") id not_null
	boost::uuids::uuid measurement_id;
#pragma db inverse(measurement_set_id.measurement_set_id)
								  //^  odb parser: error: ')' expected at the end of db pragma inverse
	std::set<measurement_parameter> parameters;
/* Other members*/
};

Regards and Boris thank you especially in advance for your time and patience

Markus Klemm

From adanesh at noornet.net  Sat Sep  5 06:26:06 2015
From: adanesh at noornet.net (=?utf-8?B?2LnZhNuMINiv2KfZhti0?=)
Date: Mon Sep  7 10:14:31 2015
Subject: [odb-users] Problem with QDate and Sqlite when it's mapped with
	TEXT type
Message-ID: <9c80cadbfcf747068d7a391042134ea4@exch.crcis.local>

Hi,

First, thanks for the work on this great tool.

I have a problem with QDate and sqlite database when QDate is mapped with TEXT type (default behavior). When it is mapped with INTEGER there is no problem, but when it is mapped with TEXT, it is persisted correctly, but retrieved incorrectly (it looks like an uninitialized variable!)



Test code:

Using odb-2.4.0 and Visual Studio 2013 and Qt 5.4.0



1-      odb.exe --std c++11 --profile qt --database sqlite --generate-query --generate-schema "%(Identity)" -I "$(MSBuildProjectDirectory)" -I "$(QTDIR)\\include" -I "$(QTDIR)\\include\\QtCore"

2-      test_qdate.h:


#pragma once

#include <QString>
#include <QDate>
#include "odb/core.hxx"

#pragma db object pointer(std::shared_ptr)
class ProgramModel
{
public:
#pragma db id auto
   int m_id;
   QString m_title;
#pragma db type("TEXT")
   QDate m_insertDate;
};



3-      test_qdate.cpp:


#include <memory>
#include <assert.h>
#include <odb/transaction.hxx>
#include <odb/database.hxx>
#include <odb/sqlite/database.hxx>
#include "odb/schema-catalog.hxx"
#include "test_qdate.h"
#include "test_qdate-odb.hxx"

auto DATABASE_NAME = L"test.sqlite";

void createDB()
{
   auto db = std::make_shared<odb::sqlite::database>(DATABASE_NAME, SQLITE_OPEN_READWRITE | SQLITE_OPEN_CREATE);
   odb::transaction t(db->begin());
   try
   {
      odb::schema_catalog::create_schema(*db, "", false);
   }
   catch (odb::exception&)
   {
   }
   t.commit();
}

int persist(std::shared_ptr<ProgramModel>& program)
{
   auto db = std::make_shared<odb::sqlite::database>(DATABASE_NAME, SQLITE_OPEN_READWRITE);
   odb::transaction t(db->begin());
   auto id = db->persist(program);
   t.commit();
   return id;
}

std::shared_ptr<ProgramModel> load(int id)
{
   auto db = std::make_shared<odb::sqlite::database>(DATABASE_NAME, SQLITE_OPEN_READWRITE);
   odb::transaction t(db->begin());
   auto result = db->load<ProgramModel>(id);
   t.commit();
   return result;
}

int main(int argc, char* argv[])
{
   createDB();
   auto item1 = std::make_shared<ProgramModel>();
   item1->m_title = "test";
   item1->m_insertDate = QDate::currentDate();
   //---
   auto item2 = load(persist(item1));
   assert(item1->m_insertDate == item2->m_insertDate);
   return 0;
}

Best regards,
Ali
From markus at markusklemm.net  Mon Sep  7 10:59:42 2015
From: markus at markusklemm.net (Markus Klemm)
Date: Mon Sep  7 10:59:55 2015
Subject: [odb-users] ORDER BY clause in ODB query language
Message-ID: <3868BE88-67A0-4118-989D-090B23CE52C6@markusklemm.net>

Hi Boris,

I didn't use the += operator, because it causes compiler errors for at least the wstring members. Surprisingly the bit type works. 
I updated the according gist ( https://gist.github.com/Superlokkus/a994d42534beca9ed5c6 ), added the erros to the relevant lines and also added the object definition.

But as always, already a big thank you, for your efforts and this still great library.

Regards

Markus Klemm

> Am 04.09.2015 um 16:41 schrieb Boris Kolpackov <boris@codesynthesis.com>:
> 
> Hi Markus,
> 
> Markus Klemm <markus@markusklemm.net> writes:
> 
>> Raw Sourcecode below [...]
> 
> Wouldn't something like this work:
> 
> using dir_query = odb::query<directory>;
> 
> dir_query q ("ORDER BY");
> 
> switch (column)
> {
> case added:
>  q += dir_query::added;
> ...
> }
> 
> switch (order)
> {
> case desc:
>  q += "DESC";
> ...
> }
> 
> db.query (q);
> 
> Boris

From boris at codesynthesis.com  Mon Sep  7 12:00:54 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Sep  7 12:01:01 2015
Subject: [odb-users] A view on top of several tables from different
	sqlite files (DB)
In-Reply-To: <E650D64AEB646045AE83BF4F2E97B8482E531EB5@HASMSX110.ger.corp.intel.com>
References: <E650D64AEB646045AE83BF4F2E97B8482E5306FC@HASMSX110.ger.corp.intel.com>
	<boris.20150902172345@codesynthesis.com>
	<E650D64AEB646045AE83BF4F2E97B8482E531EB5@HASMSX110.ger.corp.intel.com>
Message-ID: <boris.20150907175558@codesynthesis.com>

Hi Yoav,

Weiss, Yoav <yoav.weiss@intel.com> writes:

> So I did: m_p_db->connection()->execute ("ATTACH DATABASE cell.sqlite
> AS cells") Which worked fine, but the first query (using 
> m_p_db->query_one()) on a table inside cells DB, I got: 1: no such
> table: hier_version
> 
> I think this is caused by me using the connection pointer, but I'm
> not sure.

It would only be a problem if your application needed several
connections at the same time. If you executed the two statements
one after another, the query_one() would reuse the connection
that was created for the first.

Try to qualify the table name with the database name as I suggested
in my previous reply. I am suspecting this ability to use unqualified
table names with multiple databases is a recent addition to SQLite.

Boris

From finjulhich at gmail.com  Mon Sep  7 12:23:10 2015
From: finjulhich at gmail.com (MM)
Date: Mon Sep  7 12:23:17 2015
Subject: [odb-users] Re: 2-table query
In-Reply-To: <CADsB1iDM5J7FcTLwuKD5uij_rNi=i249bjkpnTHUGZmAxvWivQ@mail.gmail.com>
References: <CADsB1iDM5J7FcTLwuKD5uij_rNi=i249bjkpnTHUGZmAxvWivQ@mail.gmail.com>
Message-ID: <CADsB1iAicGKh5OhitdSSpDe1BS7Djk11HbXoy2D_Y9fLT+r4iQ@mail.gmail.com>

On 6 September 2015 at 22:19, MM <finjulhich@gmail.com> wrote:

> Hello,
>
> I have the following 2 tables and their relevant columns:
>
> M
> ==
> id: PRIMARY KEY
> month1: TEXT (encoded month,year combination)
> month2: TEXT
>
> F
> ==
> id: PRIMARY KEY
> mid: NOT NULL REFERENCES M("id")
> month: TEXT
>
> There can 0..N rows in F for each row in M.
>
> My question is two-fold:
>
>    1. How do I select all the F rows such that their month is 1 of the
>    2months in the relevant M row (the one where F.mid=M.id)?
>
> from stackoverflow, this is simply:

SELECT * FROM
F LEFT OUTER JOIN M ON F.mid = M.ID
WHERE (F.month = M.month1 OR F.month = M.month2)


>
>    1. How do I express this with an odb:query? Both M and F tables are
>    the mapping of c++ classes.
>
> db is sqlite. F has about 100 000 rows. The result of the query is about
> 200.
>
> Thanks,
> MM
>
Is this expressable as a high-level odb query?

typedef odb::query<F> Query;

db.query( Query::.month.in( .......... ,   ......... ) ) ?

MM
From boris at codesynthesis.com  Mon Sep  7 12:33:44 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Sep  7 12:33:52 2015
Subject: [odb-users] Problem with QDate and Sqlite when it's mapped with
	TEXT type
In-Reply-To: <9c80cadbfcf747068d7a391042134ea4@exch.crcis.local>
References: <9c80cadbfcf747068d7a391042134ea4@exch.crcis.local>
Message-ID: <boris.20150907182731@codesynthesis.com>

Hi Ali,

> I have a problem with QDate and sqlite database when QDate is mapped
> with TEXT type (default behavior). When it is mapped with INTEGER
> there is no problem, but when it is mapped with TEXT, it is persisted
> correctly, but retrieved incorrectly (it looks like an uninitialized
> variable!)

We have the TEXT mapping exercised in the ODB test suite and everything
works correctly, including on Windows. So we will have to do a little
debugging to figure out what's going on. Specifically, can you set the
breakpoint in file:

libodb-qt/odb/qt/date-time/sqlite/qdate-traits.hxx

Line 37 (the 'if (is_null)' statement in set_value()). Then run
your test and see what happens there. Specifically, you want to
look at i.data() and 'n' and verify that they describe the correct
string representation of the date.

Boris

From boris at codesynthesis.com  Mon Sep  7 12:36:01 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Sep  7 12:36:07 2015
Subject: [odb-users] Re: 2-table query
In-Reply-To: <CADsB1iAicGKh5OhitdSSpDe1BS7Djk11HbXoy2D_Y9fLT+r4iQ@mail.gmail.com>
References: <CADsB1iDM5J7FcTLwuKD5uij_rNi=i249bjkpnTHUGZmAxvWivQ@mail.gmail.com>
	<CADsB1iAicGKh5OhitdSSpDe1BS7Djk11HbXoy2D_Y9fLT+r4iQ@mail.gmail.com>
Message-ID: <boris.20150907183432@codesynthesis.com>

Hi,

MM <finjulhich@gmail.com> writes:

> SELECT * FROM
> F LEFT OUTER JOIN M ON F.mid = M.ID
> WHERE (F.month = M.month1 OR F.month = M.month2)

Good, this is much better than saying "I need to achieve X, do it
for me".


> Is this expressable as a high-level odb query?

Yes, pretty easily using ODB views.

Boris

From boris at codesynthesis.com  Mon Sep  7 12:41:54 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Sep  7 12:42:00 2015
Subject: [odb-users] Inverse pragma using composite value types
In-Reply-To: <632BCDC6-2C6F-4D49-9968-21416609E465@markusklemm.net>
References: <632BCDC6-2C6F-4D49-9968-21416609E465@markusklemm.net>
Message-ID: <boris.20150907184023@codesynthesis.com>

Hi Markus,

Markus Klemm <markus@markusklemm.net> writes:

> I have the problem that the inverse pragma don't work with composite
> value types. I fails to parse the . operator correctly. The problematic
> scenario is that object A's ID is part of object B's composite ID.

No, ODB doesn't support inverse relationships on "half an object id".

Boris

From boris at codesynthesis.com  Mon Sep  7 13:13:45 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Sep  7 13:13:52 2015
Subject: [odb-users] nullable unsigned variables
In-Reply-To: <CAOMsk2QH_7QeTuw8Mu-pSmH06i7SUi-0oHoc8V0aeZ+svhAMBg@mail.gmail.com>
References: <CAOMsk2QH_7QeTuw8Mu-pSmH06i7SUi-0oHoc8V0aeZ+svhAMBg@mail.gmail.com>
Message-ID: <boris.20150907191156@codesynthesis.com>

Hi Soroush,

Soroush Rabiei <soroush.rabiei@gmail.com> writes:

> I have a problem using nullable types aka `odb::nullable'.  It seems it
> can't handle `unsigned' variables.

Yes, that's an Oracle-specific bug in the ODB compiler which I have
now fixed:

http://scm.codesynthesis.com/?p=odb/odb.git;a=commit;h=f70642bcdc61d679981a55ce9bcd88c08c8dadc1

Thanks for the report and the test case, much appreciated!

Boris

From adanesh at noornet.net  Tue Sep  8 04:35:07 2015
From: adanesh at noornet.net (=?utf-8?B?2LnZhNuMINiv2KfZhti0?=)
Date: Tue Sep  8 04:35:18 2015
Subject: [odb-users] Problem with QDate and Sqlite when it's mapped with
	TEXT type
In-Reply-To: <boris.20150907182731@codesynthesis.com>
References: <9c80cadbfcf747068d7a391042134ea4@exch.crcis.local>
	<boris.20150907182731@codesynthesis.com>
Message-ID: <a322c6dec57b495b95e9a352d1d8a62f@exch.crcis.local>

Hi Boris,
Thanks for your reply. The problem is related to my operating system locale, because Qt uses operating system locale. My OS locale is Persian.

odb\qt\date-time\sqlite\qdate-traits.hxx, Line 64:

std::string s (v.toString ("yyyy-MM-dd").toLatin1 ().constData ());

must be changed to:

std::string s(QLocale(QLocale::English).toString(v, "yyyy-MM-dd").toLatin1().constData());

There are some other places that must be corrected.
Best regards,

Ali

-----Original Message-----
From: Boris Kolpackov [mailto:boris@codesynthesis.com]
Sent: Monday, September 7, 2015 9:04 PM
To: ??? ????
Cc: odb-users@codesynthesis.com
Subject: Re: [odb-users] Problem with QDate and Sqlite when it's mapped with TEXT type



Hi Ali,



> I have a problem with QDate and sqlite database when QDate is mapped

> with TEXT type (default behavior). When it is mapped with INTEGER

> there is no problem, but when it is mapped with TEXT, it is persisted

> correctly, but retrieved incorrectly (it looks like an uninitialized

> variable!)



We have the TEXT mapping exercised in the ODB test suite and everything works correctly, including on Windows. So we will have to do a little debugging to figure out what's going on. Specifically, can you set the breakpoint in file:



libodb-qt/odb/qt/date-time/sqlite/qdate-traits.hxx



Line 37 (the 'if (is_null)' statement in set_value()). Then run your test and see what happens there. Specifically, you want to look at i.data() and 'n' and verify that they describe the correct string representation of the date.



Boris
From yoav.weiss at intel.com  Tue Sep  8 08:57:39 2015
From: yoav.weiss at intel.com (Weiss, Yoav)
Date: Tue Sep  8 08:57:50 2015
Subject: [odb-users] A view on top of several tables from different
	sqlite files (DB)
In-Reply-To: <boris.20150907175558@codesynthesis.com>
References: <E650D64AEB646045AE83BF4F2E97B8482E5306FC@HASMSX110.ger.corp.intel.com>
	<boris.20150902172345@codesynthesis.com>
	<E650D64AEB646045AE83BF4F2E97B8482E531EB5@HASMSX110.ger.corp.intel.com>
	<boris.20150907175558@codesynthesis.com>
Message-ID: <E650D64AEB646045AE83BF4F2E97B8482E532C99@HASMSX110.ger.corp.intel.com>

Hi Boris,

My issue was resolved. I simply saved the connection ptr (by using the database::connection() API) and used it for all subsequent queries and transactions. I didn't need to qualify my tables since they have unique names across the 2 database schemas. I simply used --schema-name switch in the odb code generation to create specific tables in specific databases.

Thanks for your help.

Yoav


-----Original Message-----
From: Boris Kolpackov [mailto:boris@codesynthesis.com] 
Sent: Monday, September 07, 2015 19:01
To: Weiss, Yoav
Cc: odb-users@codesynthesis.com
Subject: Re: [odb-users] A view on top of several tables from different sqlite files (DB)

Hi Yoav,

Weiss, Yoav <yoav.weiss@intel.com> writes:

> So I did: m_p_db->connection()->execute ("ATTACH DATABASE cell.sqlite 
> AS cells") Which worked fine, but the first query (using
> m_p_db->query_one()) on a table inside cells DB, I got: 1: no such
> table: hier_version
> 
> I think this is caused by me using the connection pointer, but I'm not 
> sure.

It would only be a problem if your application needed several connections at the same time. If you executed the two statements one after another, the query_one() would reuse the connection that was created for the first.

Try to qualify the table name with the database name as I suggested in my previous reply. I am suspecting this ability to use unqualified table names with multiple databases is a recent addition to SQLite.

Boris
---------------------------------------------------------------------
Intel Israel (74) Limited

This e-mail and any attachments may contain confidential material for
the sole use of the intended recipient(s). Any review or distribution
by others is strictly prohibited. If you are not the intended
recipient, please contact the sender and delete all copies.


From markus at markusklemm.net  Tue Sep  8 10:50:37 2015
From: markus at markusklemm.net (Markus Klemm)
Date: Tue Sep  8 10:50:52 2015
Subject: [odb-users] Inverse pragma using composite value types
In-Reply-To: <boris.20150907184023@codesynthesis.com>
References: <632BCDC6-2C6F-4D49-9968-21416609E465@markusklemm.net>
	<boris.20150907184023@codesynthesis.com>
Message-ID: <0723AC01-43E4-4A13-9519-F3EE66B04A97@markusklemm.net>

Good day Boris,

so I guess support for views is not intented either?! Because 
#pragma db view object(measurement_parameter) object(measurement_set:measurement_parameter::measurement_parameter_key. measurement_set_id)
which is what I'm guessing would be the right prgma for a object view, is not working due a missing query conversion.

I wouldn't call it "half an object id" but rather "natural key" (instead of a surrogate key).

Regards

Markus Klemm

> Am 07.09.2015 um 18:41 schrieb Boris Kolpackov <boris@codesynthesis.com>:
> 
> Hi Markus,
> 
> Markus Klemm <markus@markusklemm.net> writes:
> 
>> I have the problem that the inverse pragma don't work with composite
>> value types. I fails to parse the . operator correctly. The problematic
>> scenario is that object A's ID is part of object B's composite ID.
> 
> No, ODB doesn't support inverse relationships on "half an object id".
> 
> Boris

From boris at codesynthesis.com  Tue Sep  8 10:58:30 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Tue Sep  8 10:58:35 2015
Subject: [odb-users] Inverse pragma using composite value types
In-Reply-To: <0723AC01-43E4-4A13-9519-F3EE66B04A97@markusklemm.net>
References: <632BCDC6-2C6F-4D49-9968-21416609E465@markusklemm.net>
	<boris.20150907184023@codesynthesis.com>
	<0723AC01-43E4-4A13-9519-F3EE66B04A97@markusklemm.net>
Message-ID: <boris.20150908165546@codesynthesis.com>

Hi Markus,

Markus Klemm <markus@markusklemm.net> writes:

> so I guess support for views is not intented either?!

No you should be able to slice & dice it pretty much any way you
want with an ODB view.

> Because #pragma db view object(measurement_parameter)
> object(measurement_set:measurement_parameter::measurement_parameter_key.measurement_set_id)

Since this is not a relationship in ODB terms, you will have to spell
out the actual JOIN condition. See the manual, it has examples of this.

Boris

From boris at codesynthesis.com  Tue Sep  8 11:05:03 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Tue Sep  8 11:05:08 2015
Subject: [odb-users] A view on top of several tables from different
	sqlite files (DB)
In-Reply-To: <E650D64AEB646045AE83BF4F2E97B8482E532C99@HASMSX110.ger.corp.intel.com>
References: <E650D64AEB646045AE83BF4F2E97B8482E5306FC@HASMSX110.ger.corp.intel.com>
	<boris.20150902172345@codesynthesis.com>
	<E650D64AEB646045AE83BF4F2E97B8482E531EB5@HASMSX110.ger.corp.intel.com>
	<boris.20150907175558@codesynthesis.com>
	<E650D64AEB646045AE83BF4F2E97B8482E532C99@HASMSX110.ger.corp.intel.com>
Message-ID: <boris.20150908165900@codesynthesis.com>

Hi Yoav,

Weiss, Yoav <yoav.weiss@intel.com> writes:

> My issue was resolved. I simply saved the connection ptr (by using
> the database::connection() API) and used it for all subsequent
> queries and transactions. 

Glad to hear it is working now.

For completeness, if you don't want to have to deal with connection_ptr,
then the "proper" way to implement this would be to provide a derived
connection_pool implementation that attaches all the necessary databases
for every connection that is created. If you are interested, see Section
18.3, "SQLite Connection and Connection Factory". The function you would
want to override is connection_pool_factory::create() in which you would
first call the base version to actually create the connection and then
attach all the databases to it before returning.

Boris

From boris at codesynthesis.com  Tue Sep  8 11:29:52 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Tue Sep  8 11:29:58 2015
Subject: [odb-users] ORDER BY clause in ODB query language
In-Reply-To: <3868BE88-67A0-4118-989D-090B23CE52C6@markusklemm.net>
References: <3868BE88-67A0-4118-989D-090B23CE52C6@markusklemm.net>
Message-ID: <boris.20150908172837@codesynthesis.com>

Hi Markus,

Markus Klemm <markus@markusklemm.net> writes:

> I didn't use the += operator, because it causes compiler errors

I've added the missing operator. Can you apply this patch and let
me know if there are any issues remaining:

http://scm.codesynthesis.com/?p=odb/libodb-mssql.git;a=commit;h=2d228c59fc6bd96944f91912c2b174cc63f56aab

Thanks,
Boris

From markus at markusklemm.net  Tue Sep  8 12:47:24 2015
From: markus at markusklemm.net (Markus Klemm)
Date: Tue Sep  8 12:47:49 2015
Subject: [odb-users] ORDER BY clause in ODB query language
In-Reply-To: <boris.20150908172837@codesynthesis.com>
References: <3868BE88-67A0-4118-989D-090B23CE52C6@markusklemm.net>
	<boris.20150908172837@codesynthesis.com>
Message-ID: <751E0A8C-D4A8-4AFC-897C-A51ED2F31FA8@markusklemm.net>

I'd love to but:

in contrast to the 2.4.0 package, opening the project with visual studio 2013 express does not work anymore (output below), but because my shift just ends in a couple of minutes I applied the patch on the used package, e.g. used header, manually instead.

But I'm not sure if it's due that I just changed the header or if there is a problem, it didn't work. But I wanted to give a 'quick' response:

[Microsoft][SQL Server Native Client 11.0][SQL Server]Falsche Syntax in der N?he von ')'.
8180 (42000): [Microsoft][SQL Server Native Client 11.0][SQL Server]Anweisung(en
) konnte(n) nicht vorbereitet werden."

Translation: Wrong syntax around ')'... query could not be perpared. (I'm sorry for the german error message, microsoft ;-( )

Visual Studio 2013 express error:

C:\Users\klm\Downloads\libodb-mssql\odb\mssql\libodb-mssql-vc12.vcxproj : error  : Unable to read the project file "libodb-mssql-vc12.vcxproj".
C:\Users\klm\Downloads\libodb-mssql\odb\mssql\libodb-mssql-vc12.vcxproj(173,3): The element <#text> beneath element <ItemGroup> is unrecognized.

Regards/ Mit freundlichen Gr??en

Markus Klemm

Gesendet via Mobiltelefon

> Am 08.09.2015 um 17:29 schrieb Boris Kolpackov <boris@codesynthesis.com>:
> 
> Hi Markus,
> 
> Markus Klemm <markus@markusklemm.net> writes:
> 
>> I didn't use the += operator, because it causes compiler errors
> 
> I've added the missing operator. Can you apply this patch and let
> me know if there are any issues remaining:
> 
> http://scm.codesynthesis.com/?p=odb/libodb-mssql.git;a=commit;h=2d228c59fc6bd96944f91912c2b174cc63f56aab
> 
> Thanks,
> Boris

From boris at codesynthesis.com  Wed Sep  9 11:12:59 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Wed Sep  9 11:13:04 2015
Subject: [odb-users] ORDER BY clause in ODB query language
In-Reply-To: <751E0A8C-D4A8-4AFC-897C-A51ED2F31FA8@markusklemm.net>
References: <3868BE88-67A0-4118-989D-090B23CE52C6@markusklemm.net>
	<boris.20150908172837@codesynthesis.com>
	<751E0A8C-D4A8-4AFC-897C-A51ED2F31FA8@markusklemm.net>
Message-ID: <boris.20150909170841@codesynthesis.com>

Hi Markus,

Markus Klemm <markus@markusklemm.net> writes:

> in contrast to the 2.4.0 package, opening the project with visual
> studio 2013 express does not work anymore...

By applying the patch I meant applying the changes from this specific
commit to what you are currently using. Getting the whole library
from master is not a good idea for multiple reasons.


> Translation: Wrong syntax around ')'... query could not be perpared.

Ok, so your application now compiles fine but you get this error at
runtime. Can you enable statement tracing for the affected transaction
so that we can see the query that causes this error:

t.tracer (odb::stderr_tracer);

Boris

From pietrino.atzeni at di-side.com  Thu Sep 10 03:17:10 2015
From: pietrino.atzeni at di-side.com (Pietrino Atzeni)
Date: Thu Sep 10 03:17:48 2015
Subject: [odb-users] Cannot compile on Yosemite 10.10.5
Message-ID: <CA+fbFz56qqRum-rfyT42OV_+5pH-0265E8wq2SkEwrEMMVvBjA@mail.gmail.com>

Hi,
I just installed a fresh copy of Yosemite, and tried to compile odb,
without success.

I tried:

- compiling with clang (Apple LLVM version 6.1.0 (clang-602.0.53)), but
during ./configure it says "g++ does not support plugins; reconfigure GCC
with --enable-plugin"
- compiling with gcc-4.9, installed with brew, but I get some errors when
using odb in my project (I can provide further details if needed);
- compiling with gcc-5.2, as above.

Is there anything more I can try?

Thanks in advance,
Pietro
From markus at markusklemm.net  Thu Sep 10 04:13:11 2015
From: markus at markusklemm.net (Markus Klemm)
Date: Thu Sep 10 04:13:26 2015
Subject: [odb-users] ORDER BY clause in ODB query language
In-Reply-To: <boris.20150909170841@codesynthesis.com>
References: <3868BE88-67A0-4118-989D-090B23CE52C6@markusklemm.net>
	<boris.20150908172837@codesynthesis.com>
	<751E0A8C-D4A8-4AFC-897C-A51ED2F31FA8@markusklemm.net>
	<boris.20150909170841@codesynthesis.com>
Message-ID: <08D5CFDB-23A9-410E-9029-5A342EF66345@markusklemm.net>

Alright I checked out tag 2.4.0 (which differs from the downloadable 2.4.0 package btw) and cherrypicked your commit.

It broke where I want to filter the files by the directories (1:n bidirect.) and still want a particular ordering of the files.

const std::vector<decltype(directory::directory_id)> &directories
odb::query<file> filter_by_directory_query = odb::query<file>::directory->directory_id.in_range(directories.cbegin(), directories.cend());
odb::query<file> order_by_query("ORDER BY" + odb::query<file>::read_time + "ASC");
odb::query<file> final_query = filter_by_directory_query + order_by_query;

I'm sorry in advance because I guess I did something wrong/unelegant again.

Resulting SQL Query:
SELECT [files].[file_id], [files].[file_name], [files].[directory_id], [files].[
fully_uploaded], [files].[partially_uploaded], [files].[read_time], [files].[upl
oader_log] FROM [files] LEFT JOIN [directories] AS [directory_id] ON [directory_
id].[directory_id]=[files].[directory_id] WHERE [directory_id].[directory_id] IN
 () ORDER BY [files].[read_time] ASC

Other querys looked fine:
 SELECT [directories].[directory_id], [directories].[full_path], [directories].[f
ile_type], [directories].[active], [directories].[added], [directories].[added_b
y] FROM [directories] ORDER BY [directories].[full_path] ASC
SELECT [file_types].[type_id], [file_types].[type_description], [file_types].[md
db_importer_flag], [file_types].[search_string], [file_types].[odbc_connection_t
emplate] FROM [file_types] WHERE [file_types].[type_id]=?
SELECT [directories].[directory_id], [directories].[full_path], [directories].[f
ile_type], [directories].[active], [directories].[added], [directories].[added_b
y] FROM [directories] ORDER BY [directories].[added] ASC

SELECT [files].[file_id], [files].[file_name], [files].[directory_id], [files].[
fully_uploaded], [files].[partially_uploaded], [files].[read_time], [files].[upl
oader_log] FROM [files] WHERE [files].[file_id]=?

Regards

Markus Klemm

From soroush.rabiei at gmail.com  Thu Sep 10 04:59:36 2015
From: soroush.rabiei at gmail.com (Soroush Rabiei)
Date: Thu Sep 10 05:00:03 2015
Subject: [odb-users] nullable unsigned variables
In-Reply-To: <boris.20150907191156@codesynthesis.com>
References: <CAOMsk2QH_7QeTuw8Mu-pSmH06i7SUi-0oHoc8V0aeZ+svhAMBg@mail.gmail.com>
	<boris.20150907191156@codesynthesis.com>
Message-ID: <CAOMsk2QoR=_GLqFWe9bPPWfmeseXcSveyQPXYiJDnNqmVgbRuQ@mail.gmail.com>

On Mon, Sep 7, 2015 at 5:13 PM, Boris Kolpackov <boris@codesynthesis.com>
wrote:

> Hi Soroush,
>
> Soroush Rabiei <soroush.rabiei@gmail.com> writes:
>
> > I have a problem using nullable types aka `odb::nullable'.  It seems it
> > can't handle `unsigned' variables.
>
> Yes, that's an Oracle-specific bug in the ODB compiler which I have
> now fixed:
>
>
> http://scm.codesynthesis.com/?p=odb/odb.git;a=commit;h=f70642bcdc61d679981a55ce9bcd88c08c8dadc1
>
> Thanks for the report and the test case, much appreciated!
>
> Boris
>

Thanks for the patch!
From markus at markusklemm.net  Thu Sep 10 05:20:09 2015
From: markus at markusklemm.net (Markus Klemm)
Date: Thu Sep 10 05:20:25 2015
Subject: [odb-users] ORDER BY clause in ODB query language
In-Reply-To: <08D5CFDB-23A9-410E-9029-5A342EF66345@markusklemm.net>
References: <3868BE88-67A0-4118-989D-090B23CE52C6@markusklemm.net>
	<boris.20150908172837@codesynthesis.com>
	<751E0A8C-D4A8-4AFC-897C-A51ED2F31FA8@markusklemm.net>
	<boris.20150909170841@codesynthesis.com>
	<08D5CFDB-23A9-410E-9029-5A342EF66345@markusklemm.net>
Message-ID: <4A23B04E-A40A-4D5E-B88C-F7B976555D72@markusklemm.net>

I'm sorry this bug is not causes by your fix, it judt slipped my tests before. 

Mit freundlichen Gr??en

Markus Klemm

Gesendet via Mobiltelefon

> Am 10.09.2015 um 10:13 schrieb Markus Klemm <markus@markusklemm.net>:
> 
> Alright I checked out tag 2.4.0 (which differs from the downloadable 2.4.0 package btw) and cherrypicked your commit.
> 
> It broke where I want to filter the files by the directories (1:n bidirect.) and still want a particular ordering of the files.
> 
> const std::vector<decltype(directory::directory_id)> &directories
> odb::query<file> filter_by_directory_query = odb::query<file>::directory->directory_id.in_range(directories.cbegin(), directories.cend());
> odb::query<file> order_by_query("ORDER BY" + odb::query<file>::read_time + "ASC");
> odb::query<file> final_query = filter_by_directory_query + order_by_query;
> 
> I'm sorry in advance because I guess I did something wrong/unelegant again.
> 
> Resulting SQL Query:
> SELECT [files].[file_id], [files].[file_name], [files].[directory_id], [files].[
> fully_uploaded], [files].[partially_uploaded], [files].[read_time], [files].[upl
> oader_log] FROM [files] LEFT JOIN [directories] AS [directory_id] ON [directory_
> id].[directory_id]=[files].[directory_id] WHERE [directory_id].[directory_id] IN
> () ORDER BY [files].[read_time] ASC
> 
> Other querys looked fine:
> SELECT [directories].[directory_id], [directories].[full_path], [directories].[f
> ile_type], [directories].[active], [directories].[added], [directories].[added_b
> y] FROM [directories] ORDER BY [directories].[full_path] ASC
> SELECT [file_types].[type_id], [file_types].[type_description], [file_types].[md
> db_importer_flag], [file_types].[search_string], [file_types].[odbc_connection_t
> emplate] FROM [file_types] WHERE [file_types].[type_id]=?
> SELECT [directories].[directory_id], [directories].[full_path], [directories].[f
> ile_type], [directories].[active], [directories].[added], [directories].[added_b
> y] FROM [directories] ORDER BY [directories].[added] ASC
> 
> SELECT [files].[file_id], [files].[file_name], [files].[directory_id], [files].[
> fully_uploaded], [files].[partially_uploaded], [files].[read_time], [files].[upl
> oader_log] FROM [files] WHERE [files].[file_id]=?
> 
> Regards
> 
> Markus Klemm
> 

From boris at codesynthesis.com  Thu Sep 10 05:28:21 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Thu Sep 10 05:28:25 2015
Subject: [odb-users] Cannot compile on Yosemite 10.10.5
In-Reply-To: <CA+fbFz56qqRum-rfyT42OV_+5pH-0265E8wq2SkEwrEMMVvBjA@mail.gmail.com>
References: <CA+fbFz56qqRum-rfyT42OV_+5pH-0265E8wq2SkEwrEMMVvBjA@mail.gmail.com>
Message-ID: <boris.20150910112438@codesynthesis.com>

Hi Pietrino,

Pietrino Atzeni <pietrino.atzeni@di-side.com> writes:

> - compiling with clang (Apple LLVM version 6.1.0 (clang-602.0.53)), but
> during ./configure it says "g++ does not support plugins; reconfigure GCC
> with --enable-plugin"
> - compiling with gcc-4.9, installed with brew, but I get some errors when
> using odb in my project (I can provide further details if needed);
> - compiling with gcc-5.2, as above.

Check this page, it has more information on using ODB on Mac OS:

http://codesynthesis.com/products/odb/doc/install-macosx.xhtml

If you are still having issues, please do provide more information.

Boris

From boris at codesynthesis.com  Thu Sep 10 10:31:34 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Thu Sep 10 10:31:40 2015
Subject: [odb-users] Problem with QDate and Sqlite when it's mapped with
	TEXT type
In-Reply-To: <a322c6dec57b495b95e9a352d1d8a62f@exch.crcis.local>
References: <9c80cadbfcf747068d7a391042134ea4@exch.crcis.local>
	<boris.20150907182731@codesynthesis.com>
	<a322c6dec57b495b95e9a352d1d8a62f@exch.crcis.local>
Message-ID: <boris.20150910162523@codesynthesis.com>

Hi Ali,

> The problem is related to my operating system locale, because Qt
> uses operating system locale. My OS locale is Persian.
> 
> odb\qt\date-time\sqlite\qdate-traits.hxx, Line 64:
> 
> std::string s (v.toString ("yyyy-MM-dd").toLatin1 ().constData ());
> 
> must be changed to:
> 
> std::string s(QLocale(QLocale::English).toString(v, "yyyy-MM-dd").toLatin1().constData());

Thanks for tracking it down. The problem with the proposed fix is
that locales are quite heavy-weight. I am not 100% sure about Qt
but standard ones are. So I am not sure it would make sense to
impose this overhead on cases where it is not necessary, which
is probably 99% of all of them.

We have the same issue with using stringstream to parse integers,
etc. When we switch to C++11, the plan is to use to_string()/sto*()
instead which always use "C" locale. I wonder if there is a similar,
locale-independent Date/Time parsing support in Qt?

Boris

From finjulhich at gmail.com  Fri Sep 11 14:57:47 2015
From: finjulhich at gmail.com (MM)
Date: Fri Sep 11 14:57:55 2015
Subject: [odb-users] Re: 2-table query
In-Reply-To: <boris.20150907183432@codesynthesis.com>
References: <CADsB1iDM5J7FcTLwuKD5uij_rNi=i249bjkpnTHUGZmAxvWivQ@mail.gmail.com>
	<CADsB1iAicGKh5OhitdSSpDe1BS7Djk11HbXoy2D_Y9fLT+r4iQ@mail.gmail.com>
	<boris.20150907183432@codesynthesis.com>
Message-ID: <CADsB1iCJVCGQ2VnJadsdv=dDVZcxrccar-_24Zuqd8Pqii0erA@mail.gmail.com>

On 7 September 2015 at 17:36, Boris Kolpackov <boris@codesynthesis.com>
wrote:

> Hi,
>
> MM <finjulhich@gmail.com> writes:
>
> > SELECT * FROM
> > F LEFT OUTER JOIN M ON F.mid = M.ID
> > WHERE (F.month = M.month1 OR F.month = M.month2)
>
> Good, this is much better than saying "I need to achieve X, do it
> for me".
>
>
> > Is this expressable as a high-level odb query?
>
> Yes, pretty easily using ODB views.
>
> Boris
>

I am struggling after reading chapter 10. Let me put the schema here again:

M
==
id: PRIMARY KEY
month1: TEXT (encoded month,year combination)
month2: TEXT

F
==
id: PRIMARY KEY
mid: NOT NULL REFERENCES M("id")
month: TEXT

There can 0..N rows in F for each row in M.

The intended SQL query is:
> SELECT * FROM
> F LEFT OUTER JOIN M ON F.mid = M.ID <http://m.id/>
> WHERE (F.month = M.month1 OR F.month = M.month2)


class F : public SomeBase {   /// SomeBase provides get_m and set_m
public:
  const M* get_m() const override;   /// this is how I access the related M
instance
  static const std::string type;   /// identifies which of the derived
classes from SomeBase this is
};

#pragma db object(F) table("F") definition
#pragma db member(F::mid) virtual(std::uint32_t)\
  get(this.get_m()? this.get_m()->id: std::uint32_t())\
  set(this.set_m(find_m(F::type,(?))))


1)
I explicitly load all the M instances and populate them myself in a
boost::ptr_map
db.query<M>();
//// use the iterator .load() to get a raw pointer to M and take ownership
of it.

2) I then load the F instances. Either I load them all, the same way as I
load the Ms (store them in their own ptr_map), or I want to load only those
Fs where the month in either month1 or month2 of the relevant M, and store
the loaded instances (by getting raw pointers and storing them in my
ptr_map)

I added the following view:

//// Specifying the joint condition to be    F::get_m   to tell odb that's
how you get the M instance from the F instance
#pragma db view object(F) object(M: F::get_m)\
  query( F::month.in (M::month1, M::month2) )
struct F_current {
  F* f;
};

odb generates the ?xx files, but it doesn't compile

All the M objects are already loaded and stored in the ptr_map. I do not
want the view loading to load them again.
The odb loader of the F instances can find the relevant instances because I
provided the get and set accessors in the pragma for F.


Is this the way to express this view?

Regards,

MM
From adanesh at noornet.net  Sat Sep 12 00:09:09 2015
From: adanesh at noornet.net (=?utf-8?B?2LnZhNuMINiv2KfZhti0?=)
Date: Sat Sep 12 00:09:19 2015
Subject: [odb-users] Problem with QDate and Sqlite when it's mapped with
	TEXT type
In-Reply-To: <boris.20150910162523@codesynthesis.com>
References: <9c80cadbfcf747068d7a391042134ea4@exch.crcis.local>
	<boris.20150907182731@codesynthesis.com>
	<a322c6dec57b495b95e9a352d1d8a62f@exch.crcis.local>
	<boris.20150910162523@codesynthesis.com>
Message-ID: <27bcd53e22fc41819a170195d668732e@exch.crcis.local>

Hi Boris,
Very good point! Using QLocate::c() static function, the problem of the proposed fix can be solved.

std::string s(QLocale::c().toString(v, "yyyy-MM-dd").toLatin1().constData());

Best regards,
Ali

-----Original Message-----
From: Boris Kolpackov [mailto:boris@codesynthesis.com] 
Sent: Thursday, September 10, 2015 7:02 PM
To: ??? ????
Cc: odb-users@codesynthesis.com
Subject: Re: [odb-users] Problem with QDate and Sqlite when it's mapped with TEXT type

Hi Ali,

> The problem is related to my operating system locale, because Qt uses 
> operating system locale. My OS locale is Persian.
> 
> odb\qt\date-time\sqlite\qdate-traits.hxx, Line 64:
> 
> std::string s (v.toString ("yyyy-MM-dd").toLatin1 ().constData ());
> 
> must be changed to:
> 
> std::string s(QLocale(QLocale::English).toString(v, 
> "yyyy-MM-dd").toLatin1().constData());

Thanks for tracking it down. The problem with the proposed fix is that locales are quite heavy-weight. I am not 100% sure about Qt but standard ones are. So I am not sure it would make sense to impose this overhead on cases where it is not necessary, which is probably 99% of all of them.

We have the same issue with using stringstream to parse integers, etc. When we switch to C++11, the plan is to use to_string()/sto*() instead which always use "C" locale. I wonder if there is a similar, locale-independent Date/Time parsing support in Qt?

Boris

From andrew at a-cunningham.com  Fri Sep 11 16:39:49 2015
From: andrew at a-cunningham.com (Andrew Cunningham)
Date: Sat Sep 12 06:41:14 2015
Subject: [odb-users] SQLIte "INSERT" slows dramatically...
Message-ID: <CAGTJuHgHmmq_CpkAoSrSQBmgZ=X4=VcGijFXArNJ2_7ExGzQWw@mail.gmail.com>

Hi,

In general I am very happy with the performance ODB 2.4.0 /SQLite.


However I am running into some serious performance issues when creating
?large? databases - I  am seeing  SQLIte performance completely tanking ,
but only at a certain database size.


Assume I have am inserting a thousand ?subsystem? objects to the database,
each of those ?subsystem? objects has a std::vector of ?elements? that can
be size 100-50,000

This is being done in a single transaction..



When creating this database, performance is perfectly fine until at some
point the time to persist a ?subsystem and its elements? ( and all
subsequent ones)  increases dramatically by a factor of 1000.



For example, early on in the database creation a subsystem with ~1000
elements might take 0.08s to persist (along with its elements)

However, at a certain point, persisting a similar object  starts to take
118s ( 1000x longer).



Breaking into SQLite seems to show that the code is often in this phase.
Looks like some kind of database paging/moving is going on?.



>             odb-sqlite-2.4-vc10.dll!winRead(sqlite3_file * id, void *
pBuf, int amt, __int64 offset)  Line 35654 + 0x2 bytes                C

                odb-sqlite-2.4-vc10.dll!readDbPage(PgHdr * pPg, unsigned
int iFrame)  Line 44461 + 0x19 bytes C

                odb-sqlite-2.4-vc10.dll!sqlite3PagerAcquire(Pager * pPager,
unsigned int pgno, PgHdr * * ppPage, int flags)  Line 46979 + 0xb bytes   C

                odb-sqlite-2.4-vc10.dll!getAndInitPage(BtShared * pBt,
unsigned int pgno, MemPage * * ppPage, int bReadonly)  Line 54790 + 0xd
bytes         C

                odb-sqlite-2.4-vc10.dll!moveToRoot(BtCursor * pCur)  Line
57596 + 0x22 bytes  C

                odb-sqlite-2.4-vc10.dll!sqlite3VdbeExec(Vdbe * p)  Line
9732 + 0x8 bytes             C

                odb-sqlite-2.4-vc10.dll!sqlite3Step(Vdbe * p)  Line 3812 C

                odb-sqlite-2.4-vc10.dll!sqlite3_step(sqlite3_stmt *
pStmt)  Line
3878 + 0x8 bytes              C


odb-sqlite-2.4-vc10.dll!odb::sqlite::insert_statement::execute()  Line 649
+ 0xd bytes    C++



Monitoring the process with ?Process Monitor?, shows that SQLite is making
a very large number of calls to the Windows API ReadFile at this point.



Any ideas welcome!


Andrew
From sean.clarke at sec-consulting.co.uk  Sun Sep 13 11:20:41 2015
From: sean.clarke at sec-consulting.co.uk (Sean Clarke)
Date: Sun Sep 13 11:20:50 2015
Subject: [odb-users] include issues and unknown class errors on many-to-one
	relationships
Message-ID: <CAHCKNJgPDNF2hrZ2ymJX3vmO1H7iHQF9KQ0xUXXFsKBOTrjxFQ@mail.gmail.com>

Hi,
    For some reason I keep getting include issues when I try and set up
some basic many-to-one relationships- I can get your examples to work
because they are in the same file, yet mine (when I split across multiple
headers) seem to get in a bit of a state....

------------------------------------------------
Job.hpp
------------------------------------------------
#pragma once

class Job {
// lots of stuff

#pragma db value_not_null inverse(m_job)
std::vector<std::weak_ptr<SomeJobItem>> m_job_items;
};
------------------------------------------------

and then in another header file I have:

------------------------------------------------
SomeJobItem.hpp
------------------------------------------------
#pragma once

class SomeJobItem {
// lots of stuff

#pragma db not_null
std::weak_ptr<Job> m_job;
};
------------------------------------------------

This gives errors that it can't find the class (even though I include the
headers in each of them)

 error: ?JSomJobItem? was not declared in this scope
 std::vector<std::weak_ptr<SomeJobItem>> m_job_items;

etc.

I can't forward declare the classes as ODB then complains that the type
maps to a non DB type...

Any ideas on what I am doing wrong on these simple examples?

Regards
Sean Clarke
From sean.clarke at sec-consulting.co.uk  Sun Sep 13 15:23:38 2015
From: sean.clarke at sec-consulting.co.uk (Sean Clarke)
Date: Sun Sep 13 15:23:46 2015
Subject: [odb-users] Moving odb::result<T>
Message-ID: <CAHCKNJhtyoUOxR-Vfitq3_4vuEDPCNA56W+yf3C4mxvUnrzNMA@mail.gmail.com>

Hi,
    I have been doing some tidying etc. and moved some queries into a
factory class, however I am getting segsv issues when I try and move the
results of some queries...

Program received signal SIGSEGV, Segmentation fault.
0x00000000004eca92 in odb::result<Job>::size (this=0x7fffffffd6f0)
    at orm_test/libodb-2.4.0/odb/result.hxx:209
209      return impl_ ? impl_->size () : 0;

I just perform a simple query and return the resultant odb::result using a
std::move....

If I don't call the std::move and just return the result it does not give a
segv, but doesn't work  - but I sort of expected this from the info you
assisted me with previously regarding multiple iterations over the
container/forward_iterator<>.

I can "fudge" it by copying data to a std::vector<T>, but would rather not
if at all possible.


Regards
Sean Clarke
From finjulhich at gmail.com  Sun Sep 13 15:53:54 2015
From: finjulhich at gmail.com (MM)
Date: Sun Sep 13 15:54:03 2015
Subject: [odb-users] include issues and unknown class errors on
	many-to-one relationships
In-Reply-To: <CAHCKNJgPDNF2hrZ2ymJX3vmO1H7iHQF9KQ0xUXXFsKBOTrjxFQ@mail.gmail.com>
References: <CAHCKNJgPDNF2hrZ2ymJX3vmO1H7iHQF9KQ0xUXXFsKBOTrjxFQ@mail.gmail.com>
Message-ID: <CADsB1iAv5f15VHpv1kUftBn2C7JF6e5sr2nfEyatXs6Ek69Qhw@mail.gmail.com>

On 13 September 2015 at 16:20, Sean Clarke <sean.clarke@sec-consulting.co.uk
> wrote:

> Hi,
>     For some reason I keep getting include issues when I try and set up
> some basic many-to-one relationships- I can get your examples to work
> because they are in the same file, yet mine (when I split across multiple
> headers) seem to get in a bit of a state....
>
> ------------------------------------------------
> Job.hpp
> ------------------------------------------------
> #pragma once
>
> class Job {
> // lots of stuff
>
> #pragma db value_not_null inverse(m_job)
> std::vector<std::weak_ptr<SomeJobItem>> m_job_items;
> };
> ------------------------------------------------
>
> and then in another header file I have:
>
> ------------------------------------------------
> SomeJobItem.hpp
> ------------------------------------------------
> #pragma once
>
> class SomeJobItem {
> // lots of stuff
>
> #pragma db not_null
> std::weak_ptr<Job> m_job;
> };
> ------------------------------------------------
>
> This gives errors that it can't find the class (even though I include the
> headers in each of them)
>
>  error: ?JSomJobItem? was not declared in this scope
>  std::vector<std::weak_ptr<SomeJobItem>> m_job_items;
>
> etc.
>
> I can't forward declare the classes as ODB then complains that the type
> maps to a non DB type...
>
> Any ideas on what I am doing wrong on these simple examples?
>
> Regards
> Sean Clarke


I would try to extract the minimal independent example that you can email
here.
It often happened to me that with the extracted example, an obvious error
appeared that I couldn't see in the real code.
Otherwise, a minimal example can help Boris or the list pinpoint a bug in
the compiler, or just a wrong usage of the pragmas.

Rds,
From markus at markusklemm.net  Mon Sep 14 05:26:47 2015
From: markus at markusklemm.net (Markus Klemm)
Date: Mon Sep 14 05:27:01 2015
Subject: [odb-users] Assumed bug in odb::query in_range when range is empty
Message-ID: <3314C3F1-850D-4ABF-A619-A5DCEB3D012E@markusklemm.net>

Hi everyone,

I think I found a bug in odb::query<>::field:in_range(begin,end) : 
If the given range is empty i.e. begin == end, the resulting sql query ("WHERE [type].[field] IN ()") is malformend or at least MS SQL server does not like it.

It's basiclly the bug I encountered in the last messages of the mail thread ("ORDER BY clause in ODB query language"/Thu Sep 10 05:20:09 EDT 2015), but I minized it to the following code:

Full example/Gist with syntax-highlighting: https://gist.github.com/Superlokkus/e23a5be5bcebf738b1a0

#pragma db object table("files") bulk(5000) session pointer(std::shared_ptr<file>)
		class file
		{
		public:
#pragma db column("file_id") type("uniqueidentifier") id not_null
			boost::uuids::uuid file_id;

#pragma db column("file_name") type("nvarchar(256)") not_null
			std::wstring file_name;

		};

std::vector<decltype(file::file_id)> files;
odb::query<file> query = odb::query<file>::file_id.in_range(files.cbegin(), files.cend());
odb::session s;
odb::transaction t(db->begin());
t.tracer(odb::stderr_tracer);
odb::result<file> result(
	db->query <file>(query));
for (const auto &f : result){
	std::wcout << f.file_name << std::endl;
}

Output
"SELECT [files].[file_id], [files].[file_name] FROM [files] WHERE [files].[file_i
d] IN ()
ODB database error: 102 (42000): [Microsoft][SQL Server Native Client 11.0][SQL
Server]Falsche Syntax in der N?he von ')'.
8180 (42000): [Microsoft][SQL Server Native Client 11.0][SQL Server]Anweisung(en
) konnte(n) nicht vorbereitet werden."

Falsche Syntax in der N?he von ')'. == Wrong syntax near ')'
Anweisung(en) konnte(n) nicht vorbereitet werden. == Could not prepare query

I wish you a nice (or at least bearable) monday

Markus Klemm

From finjulhich at gmail.com  Mon Sep 14 05:56:36 2015
From: finjulhich at gmail.com (MM)
Date: Mon Sep 14 05:56:44 2015
Subject: [odb-users] sqlite mismatch in result column count
Message-ID: <CADsB1iCAMyME1wb3NxHkCFMVWN1+ZzkMcR9vFjicaX9cxZPY2g@mail.gmail.com>

Hello,

This assertion, from sqlite, gets triggered:

statement.cxx:331: bool odb::sqlite::statement::bind_result(const
odb::sqlite::bind*, size_t, bool): Assertion `col == col_count' failed.

Looking at the source code, there are 3 variables that relate to column
count:

 bool statement::
    bind_result (const bind* p, size_t count, bool truncated)
    {
    }

   1. The 'count' argument of bind_result: looking at the core file, its
   value is 4
   2. col_count: I couldn't find its value from the core file. Is it the
   expected number of columns that is deduced from the SQL query before it is
   executed?
   If so, that number is 4 also.
   3. col, I suppose this is the actual number of columns that is returned
   after executing the SQL query. I traced the odb query, executed the SQL
   query I see traced out, it has 4 columns, and the result also includes 4
   columns.
   4. The query is the result of a call:    db.query<F>();
   F is a struct, where the number of non transient fields overall is 4 as
   well.
   F(2 virtual members) derives from B1(odb abstract, 1 member) which
   derives from B0(odb abstract, 1 member).

I will try to printout the values of those variables in my locally modified
libodb-sqlite to see what's going on.

Rds,
From finjulhich at gmail.com  Mon Sep 14 08:43:26 2015
From: finjulhich at gmail.com (MM)
Date: Mon Sep 14 08:43:33 2015
Subject: [odb-users] Re: sqlite mismatch in result column count
In-Reply-To: <CADsB1iCAMyME1wb3NxHkCFMVWN1+ZzkMcR9vFjicaX9cxZPY2g@mail.gmail.com>
References: <CADsB1iCAMyME1wb3NxHkCFMVWN1+ZzkMcR9vFjicaX9cxZPY2g@mail.gmail.com>
Message-ID: <CADsB1iCxf+UuALzd1SOGHCU4H=NnojgMY8NCnnAx1TPyefX5tQ@mail.gmail.com>

On 14 September 2015 at 10:56, MM <finjulhich@gmail.com> wrote:

> Hello,
>
> This assertion, from sqlite, gets triggered:
>
> statement.cxx:331: bool odb::sqlite::statement::bind_result(const
> odb::sqlite::bind*, size_t, bool): Assertion `col == col_count' failed.
>
> Looking at the source code, there are 3 variables that relate to column
> count:
>
>  bool statement::
>     bind_result (const bind* p, size_t count, bool truncated)
>     {
>     }
>
>    1. The 'count' argument of bind_result: looking at the core file, its
>    value is 4
>    2. col_count: I couldn't find its value from the core file. Is it the
>    expected number of columns that is deduced from the SQL query before it is
>    executed?
>    If so, that number is 4 also.
>    3. col, I suppose this is the actual number of columns that is
>    returned after executing the SQL query. I traced the odb query, executed
>    the SQL query I see traced out, it has 4 columns, and the result also
>    includes 4 columns.
>    4. The query is the result of a call:    db.query<F>();
>    F is a struct, where the number of non transient fields overall is 4
>    as well.
>    F(2 virtual members) derives from B1(odb abstract, 1 member) which
>    derives from B0(odb abstract, 1 member).
>
> I will try to printout the values of those variables in my locally
> modified libodb-sqlite to see what's going on.
>
> Rds,
>
> After some debugging, I find that the value of 'col' is 3, not 4.
The loop goes through i=0,1,2,3 as expected.

if (b.buffer == 0) { // Skip NULL entries.
}

This happens once, and that is why col is 3.

I printed b.type, and 3 times, it is 0, ie bind::integer

The 4rd column (the missing one) is of type std::string in the c++ class
(B1), and of type TEXT in the db.

B1 is pragma'd like this
#pragma db object(B1) abstract definition

I assume this means to indeed declare all of its members as non transient.

I'm gonna explicitly declare the member here with #pragma member and see if
there's a difference

Rds,
From boris at codesynthesis.com  Mon Sep 14 09:16:42 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Sep 14 09:16:52 2015
Subject: [odb-users] include issues and unknown class errors on
	many-to-one relationships
In-Reply-To: <CAHCKNJgPDNF2hrZ2ymJX3vmO1H7iHQF9KQ0xUXXFsKBOTrjxFQ@mail.gmail.com>
References: <CAHCKNJgPDNF2hrZ2ymJX3vmO1H7iHQF9KQ0xUXXFsKBOTrjxFQ@mail.gmail.com>
Message-ID: <boris.20150914151527@codesynthesis.com>

Hi Sean,

Sean Clarke <sean.clarke@sec-consulting.co.uk> writes:

>     For some reason I keep getting include issues when I try and set up
> some basic many-to-one relationships- I can get your examples to work
> because they are in the same file, yet mine (when I split across multiple
> headers) seem to get in a bit of a state....

Have you read Section 6.3, "Circular Relationships"?

Boris

From fredrik.strand at qliro.com  Mon Sep 14 08:05:55 2015
From: fredrik.strand at qliro.com (Fredrik Strand)
Date: Mon Sep 14 09:30:12 2015
Subject: [odb-users] Set Query timeout
Message-ID: <AM3PR02MB03358CA068950C0C90B9782895D0@AM3PR02MB033.eurprd02.prod.outlook.com>

Hi,

It's possible to set query timeout using raw odbc calls. However I don't find a way to set query timeout using ODB. Nor do I see any support for this in the source code.

Is there a way to set query timeout using ODB?

Fredrik Strand
Lead Developer

Qliro AB
Sveav?gen 151,
Stockholm 113 46
Mobile: +46733 30 83 82
E-mail: fredrik.strand@qliro.com

[footer]

Confidentiality
Information contained in this e-mail is intended for the use of the
addressee only, and is confidential. Any dissemination, distribution,
copying or use of this communication without prior permission of
the addressee is strictly prohibited. If you are not the intended
addressee you must delete this e-mail and its attachments.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: image001.jpg
Type: image/jpeg
Size: 6743 bytes
Desc: image001.jpg
Url : http://codesynthesis.com/pipermail/odb-users/attachments/20150914/a5261ebb/image001.jpg
From boris at codesynthesis.com  Mon Sep 14 09:35:47 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Sep 14 09:35:56 2015
Subject: [odb-users] Re: sqlite mismatch in result column count
In-Reply-To: <CADsB1iCxf+UuALzd1SOGHCU4H=NnojgMY8NCnnAx1TPyefX5tQ@mail.gmail.com>
References: <CADsB1iCAMyME1wb3NxHkCFMVWN1+ZzkMcR9vFjicaX9cxZPY2g@mail.gmail.com>
	<CADsB1iCxf+UuALzd1SOGHCU4H=NnojgMY8NCnnAx1TPyefX5tQ@mail.gmail.com>
Message-ID: <boris.20150914153252@codesynthesis.com>

Hi,

MM <finjulhich@gmail.com> writes:

> The 4rd column (the missing one) is of type std::string in the c++ class
> (B1), and of type TEXT in the db.

You seem to be trying to map C++ classes to an existing schema. In this
case it is very helpful to still ask ODB to generate the schema and then
make sure that what gets generated is equivalent to the existing schema.

For SQLite, use '--generate-schema --schema-format sql' to get the
schema as a standalone .sql file.

Boris

From finjulhich at gmail.com  Mon Sep 14 09:40:56 2015
From: finjulhich at gmail.com (MM)
Date: Mon Sep 14 09:41:04 2015
Subject: [odb-users] Re: sqlite mismatch in result column count
In-Reply-To: <boris.20150914153252@codesynthesis.com>
References: <CADsB1iCAMyME1wb3NxHkCFMVWN1+ZzkMcR9vFjicaX9cxZPY2g@mail.gmail.com>
	<CADsB1iCxf+UuALzd1SOGHCU4H=NnojgMY8NCnnAx1TPyefX5tQ@mail.gmail.com>
	<boris.20150914153252@codesynthesis.com>
Message-ID: <CADsB1iCcbExR8s+bcdFWf-ZJswaXjsWM_0Xxd7+XxXeWhoUhrg@mail.gmail.com>

On 14 September 2015 at 14:35, Boris Kolpackov <boris@codesynthesis.com>
wrote:

> Hi,
>
> MM <finjulhich@gmail.com> writes:
>
> > The 4rd column (the missing one) is of type std::string in the c++ class
> > (B1), and of type TEXT in the db.
>
> You seem to be trying to map C++ classes to an existing schema. In this
> case it is very helpful to still ask ODB to generate the schema and then
> make sure that what gets generated is equivalent to the existing schema.
>
> For SQLite, use '--generate-schema --schema-format sql' to get the
> schema as a standalone .sql file.
>
> Boris
>

Yes, I do that too.
THe CREATE TABLE Statement does have 4 columns (3 INTEGER and 1 TEXT).

Normally,

#pragma odb (B1) abstract definition

is equivalent to

#pragma db object(B1) abstract definition
#pragma db member(B1::field)

with B1's base is abstract also, and F (derived from B) is not abstract

?
From boris at codesynthesis.com  Mon Sep 14 09:50:44 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Sep 14 09:50:54 2015
Subject: [odb-users] Re: sqlite mismatch in result column count
In-Reply-To: <CADsB1iCxf+UuALzd1SOGHCU4H=NnojgMY8NCnnAx1TPyefX5tQ@mail.gmail.com>
References: <CADsB1iCAMyME1wb3NxHkCFMVWN1+ZzkMcR9vFjicaX9cxZPY2g@mail.gmail.com>
	<CADsB1iCxf+UuALzd1SOGHCU4H=NnojgMY8NCnnAx1TPyefX5tQ@mail.gmail.com>
Message-ID: <boris.20150914154650@codesynthesis.com>

MM <finjulhich@gmail.com> writes:

> if (b.buffer == 0) { // Skip NULL entries.
> }
> 
> This happens once, and that is why col is 3.

This should only happen if you are using schema evolution support
and this column was soft-deleted. And in this case the corresponding
column from the statement's SELECT-list should have been removed as
well (see statement processing in the init() function in the same
file).

If you are not using schema evolution support, then this is either
a bug in ODB or a memory corruption in your application. Can you
run it under valgrind?

Boris

From finjulhich at gmail.com  Mon Sep 14 10:00:58 2015
From: finjulhich at gmail.com (MM)
Date: Mon Sep 14 10:01:05 2015
Subject: [odb-users] Re: sqlite mismatch in result column count
In-Reply-To: <CADsB1iCcbExR8s+bcdFWf-ZJswaXjsWM_0Xxd7+XxXeWhoUhrg@mail.gmail.com>
References: <CADsB1iCAMyME1wb3NxHkCFMVWN1+ZzkMcR9vFjicaX9cxZPY2g@mail.gmail.com>
	<CADsB1iCxf+UuALzd1SOGHCU4H=NnojgMY8NCnnAx1TPyefX5tQ@mail.gmail.com>
	<boris.20150914153252@codesynthesis.com>
	<CADsB1iCcbExR8s+bcdFWf-ZJswaXjsWM_0Xxd7+XxXeWhoUhrg@mail.gmail.com>
Message-ID: <CADsB1iCbx=8wnypRiUDdoPcDGt55KGP2AL8=B+MbHyUcibWx6A@mail.gmail.com>

On 14 September 2015 at 14:40, MM <finjulhich@gmail.com> wrote:

>
>
> On 14 September 2015 at 14:35, Boris Kolpackov <boris@codesynthesis.com>
> wrote:
>
>> Hi,
>>
>> MM <finjulhich@gmail.com> writes:
>>
>> > The 4rd column (the missing one) is of type std::string in the c++ class
>> > (B1), and of type TEXT in the db.
>>
>> You seem to be trying to map C++ classes to an existing schema. In this
>> case it is very helpful to still ask ODB to generate the schema and then
>> make sure that what gets generated is equivalent to the existing schema.
>>
>> For SQLite, use '--generate-schema --schema-format sql' to get the
>> schema as a standalone .sql file.
>>
>> Boris
>>
>
> Yes, I do that too.
> THe CREATE TABLE Statement does have 4 columns (3 INTEGER and 1 TEXT).
>
> Normally,
>
> #pragma odb (B1) abstract definition
>
> is equivalent to
>
> #pragma db object(B1) abstract definition
> #pragma db member(B1::field)
>
> with B1's base is abstract also, and F (derived from B) is not abstract
>
> ?
>
I just finished a long recompile because I had to add /usr/local/include to
test with the modified libodb...:-)

it now works.

I thought that as long as the entire B1 was not marked transient, all its
members would be considered, by non-abstract derived classes. This is:
    public std::string field

Adding
  #pragma db member(B1::field)
explicitly solved the problem

Rds,
From boris at codesynthesis.com  Mon Sep 14 10:11:19 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Sep 14 10:11:29 2015
Subject: [odb-users] Set Query timeout
In-Reply-To: <AM3PR02MB03358CA068950C0C90B9782895D0@AM3PR02MB033.eurprd02.prod.outlook.com>
References: <AM3PR02MB03358CA068950C0C90B9782895D0@AM3PR02MB033.eurprd02.prod.outlook.com>
Message-ID: <boris.20150914155111@codesynthesis.com>

Hi Fredrik,

Fredrik Strand <fredrik.strand@qliro.com> writes:

> It's possible to set query timeout using raw odbc calls. However I
> don't find a way to set query timeout using ODB. Nor do I see any
> support for this in the source code.

I assume you are interesting in the SQL_ATTR_QUERY_TIMEOUT statement
attribute. While ODB normally hides underlying database statements,
you can get hold of it for a prepared query (Section 4.5 in the
manual). Here is an outline based on the example from this section:

#include <odb/mssql/error.hxx>
#include <odb/mssql/statement.hxx>

using query = odb::query<person>;
using prep_query = odb::prepared_query<person>;
using result = odb::result<person>;

transaction t (db.begin ());

prep_query pq (db.prepare_query<person> ("age-query", query::age > 18));

// Set timeout.
//
{
  auto& s (static_cast<odb::mssql::statement&> (pq.statement ()));

  SQLHSTMT h (s.handle ());
  SQLULEN t (10); // In seconds.
  SQLRETURN r (SQLSetStmtAttr (h, 
                               SQL_ATTR_QUERY_TIMEOUT, 
                               (SQLPOINTER) t,
                               nullptr));

  // Translate errors to exceptions using ODB mechanisms.
  //
  if (!SQL_SUCCEEDED (r))
    odb::mssql::translate_error (r, s.connection (), s);
}

// Run the query.
//
result r (pq.execute ());
...

t.commit ();

Let me know if there are any problems with this approach.

Boris

From boris at codesynthesis.com  Mon Sep 14 10:12:42 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Sep 14 10:12:51 2015
Subject: [odb-users] Re: sqlite mismatch in result column count
In-Reply-To: <CADsB1iCbx=8wnypRiUDdoPcDGt55KGP2AL8=B+MbHyUcibWx6A@mail.gmail.com>
References: <CADsB1iCAMyME1wb3NxHkCFMVWN1+ZzkMcR9vFjicaX9cxZPY2g@mail.gmail.com>
	<CADsB1iCxf+UuALzd1SOGHCU4H=NnojgMY8NCnnAx1TPyefX5tQ@mail.gmail.com>
	<boris.20150914153252@codesynthesis.com>
	<CADsB1iCcbExR8s+bcdFWf-ZJswaXjsWM_0Xxd7+XxXeWhoUhrg@mail.gmail.com>
	<CADsB1iCbx=8wnypRiUDdoPcDGt55KGP2AL8=B+MbHyUcibWx6A@mail.gmail.com>
Message-ID: <boris.20150914161151@codesynthesis.com>

MM <finjulhich@gmail.com> writes:

> Adding
>   #pragma db member(B1::field)
> explicitly solved the problem

I don't think you know what exactly solved your problem... ;-)

Boris

From boris at codesynthesis.com  Mon Sep 14 10:45:31 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Sep 14 10:45:40 2015
Subject: [odb-users] Re: 2-table query
In-Reply-To: <CADsB1iCJVCGQ2VnJadsdv=dDVZcxrccar-_24Zuqd8Pqii0erA@mail.gmail.com>
References: <CADsB1iDM5J7FcTLwuKD5uij_rNi=i249bjkpnTHUGZmAxvWivQ@mail.gmail.com>
	<CADsB1iAicGKh5OhitdSSpDe1BS7Djk11HbXoy2D_Y9fLT+r4iQ@mail.gmail.com>
	<boris.20150907183432@codesynthesis.com>
	<CADsB1iCJVCGQ2VnJadsdv=dDVZcxrccar-_24Zuqd8Pqii0erA@mail.gmail.com>
Message-ID: <boris.20150914164211@codesynthesis.com>

MM <finjulhich@gmail.com> writes:

> odb generates the ?xx files, but it doesn't compile

What is the error?!


> #pragma db view object(F) object(M: F::get_m)\
>   query( F::month.in (M::month1, M::month2) )

ODB in() function only works on values, not members. I guess that's
the error but I shouldn't be guessing.

Boris

From markus at markusklemm.net  Mon Sep 14 12:09:23 2015
From: markus at markusklemm.net (Markus Klemm)
Date: Mon Sep 14 12:09:38 2015
Subject: [odb-users] statement-cache.txx compiler error when using object
	loading views
Message-ID: <F96FC22D-91F3-4135-9E82-5C6330123BDF@markusklemm.net>

My second an last mail for today;

Either I'm missing something essential about object loading views or I stumbled over a compile time bug of odb, in this case \libodb-mssql-2.4.0\odb\mssql\statement-cache.txx.
Given for a simple example (code below), I got this compiler error
'odb::mssql::object_statements<odb::access::object_traits<type>::object_type>' : class has no constructors	\libodb-mssql-2.4.0\odb\mssql\statement-cache.txx	36	
'static_cast' : cannot convert from 'odb::mssql::statements_base' to 'statements_type &' \libodb-mssql-2.4.0\odb\mssql\statement-cache.txx	33.

Code: https://gist.github.com/Superlokkus/ea4a28e197fea67cec4e
ODB-compile command I use: odb.exe -I <boost_path>\include -q -e -d mssql --profile boost --std c++11 <class.h>
I'm using MS MVC 2013

Regards

Markus Klemm

From boris at codesynthesis.com  Mon Sep 14 13:49:56 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Sep 14 13:50:06 2015
Subject: [odb-users] Assumed bug in odb::query in_range when range is empty
In-Reply-To: <3314C3F1-850D-4ABF-A619-A5DCEB3D012E@markusklemm.net>
References: <3314C3F1-850D-4ABF-A619-A5DCEB3D012E@markusklemm.net>
Message-ID: <boris.20150914191149@codesynthesis.com>

Hi Markus,

Markus Klemm <markus@markusklemm.net> writes:

> I think I found a bug in odb::query<>::field:in_range(begin,end) : 
> If the given range is empty i.e. begin == end, the resulting sql 
> query ("WHERE [type].[field] IN ()") is malformend [...]

Yes, as I mentioned in one of my previous emails, this is a known
bug that we have fixed for the next release:

http://scm.codesynthesis.com/?p=odb/libodb-mssql.git;a=commit;h=f25ff0d1163ea65abd709b6037098f6a2199a238

Boris

From boris at codesynthesis.com  Mon Sep 14 13:55:33 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Sep 14 13:55:44 2015
Subject: [odb-users] Moving odb::result<T>
In-Reply-To: <CAHCKNJhtyoUOxR-Vfitq3_4vuEDPCNA56W+yf3C4mxvUnrzNMA@mail.gmail.com>
References: <CAHCKNJhtyoUOxR-Vfitq3_4vuEDPCNA56W+yf3C4mxvUnrzNMA@mail.gmail.com>
Message-ID: <boris.20150914195205@codesynthesis.com>

Hi Sean,

Sean Clarke <sean.clarke@sec-consulting.co.uk> writes:

> I just perform a simple query and return the resultant odb::result using a
> std::move....

odb::result is copyable, you can just return it by value without move().
Second paragraph from Section 4.4, "Query Result":

"It is best to view an instance of odb::result as a handle to a stream,
 such as a socket stream. While we can make a copy of a result or assign
 one result to another, the two instances will refer to the same result
 stream. Advancing the current position in one instance will also advance
 it in another. The result instance is only usable within the transaction
 it was created in. Trying to manipulate the result after the transaction
 has terminated leads to undefined behavior."

My guess would be you return the result out of transaction scope.

Boris

From sean.clarke at sec-consulting.co.uk  Mon Sep 14 14:11:59 2015
From: sean.clarke at sec-consulting.co.uk (Sean Clarke)
Date: Mon Sep 14 14:12:06 2015
Subject: [odb-users] Moving odb::result<T>
In-Reply-To: <boris.20150914195205@codesynthesis.com>
References: <CAHCKNJhtyoUOxR-Vfitq3_4vuEDPCNA56W+yf3C4mxvUnrzNMA@mail.gmail.com>
	<boris.20150914195205@codesynthesis.com>
Message-ID: <CAHCKNJh50KQinabFhnuOnGfQOjV1RJC3CUKREZxoswgcdYDbDQ@mail.gmail.com>

> My guess would be you return the result out of transaction scope.

You are indeed correct!

Many thanks Boris, as helpful as ever.


Regards
Sean Clarke
---------------------------------------------
SEC Consulting Limited
Phone: +44 (0)23 8040 5599
Website: http://www.sec-consulting.co.uk
Email: sean.clarke@sec-consulting.co.uk

On 14 September 2015 at 18:55, Boris Kolpackov <boris@codesynthesis.com>
wrote:

> Hi Sean,
>
> Sean Clarke <sean.clarke@sec-consulting.co.uk> writes:
>
> > I just perform a simple query and return the resultant odb::result using
> a
> > std::move....
>
> odb::result is copyable, you can just return it by value without move().
> Second paragraph from Section 4.4, "Query Result":
>
> "It is best to view an instance of odb::result as a handle to a stream,
>  such as a socket stream. While we can make a copy of a result or assign
>  one result to another, the two instances will refer to the same result
>  stream. Advancing the current position in one instance will also advance
>  it in another. The result instance is only usable within the transaction
>  it was created in. Trying to manipulate the result after the transaction
>  has terminated leads to undefined behavior."
>
> My guess would be you return the result out of transaction scope.
>
> Boris
>
From boris at codesynthesis.com  Mon Sep 14 14:36:29 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Sep 14 14:36:38 2015
Subject: [odb-users] SQLIte "INSERT" slows dramatically...
In-Reply-To: <CAGTJuHgHmmq_CpkAoSrSQBmgZ=X4=VcGijFXArNJ2_7ExGzQWw@mail.gmail.com>
References: <CAGTJuHgHmmq_CpkAoSrSQBmgZ=X4=VcGijFXArNJ2_7ExGzQWw@mail.gmail.com>
Message-ID: <boris.20150914203254@codesynthesis.com>

Hi Andrew,

These two pages list various tuning options for using SQLite with
very large databases. In particular, I would try the WAL journal.

http://stackoverflow.com/questions/10209438/sqlite-end-transaction-is-taking-too-long

http://stackoverflow.com/a/6533930

Boris

From sean.clarke at sec-consulting.co.uk  Mon Sep 14 14:42:21 2015
From: sean.clarke at sec-consulting.co.uk (Sean Clarke)
Date: Mon Sep 14 14:42:28 2015
Subject: [odb-users] include issues and unknown class errors on
	many-to-one relationships
In-Reply-To: <boris.20150914151527@codesynthesis.com>
References: <CAHCKNJgPDNF2hrZ2ymJX3vmO1H7iHQF9KQ0xUXXFsKBOTrjxFQ@mail.gmail.com>
	<boris.20150914151527@codesynthesis.com>
Message-ID: <CAHCKNJhGGJ1m5yK2dYrwp3vha28-Lde=n_+8d_3mX_SftQ3uaA@mail.gmail.com>

Hi Boris,
   yes, I did - but I either get these errors or recursion include errors.
I'll keep at it and try some things, it is probably conflicting with my
pragma once guards - maybe I'll switch to the standard guards and put the
include (with #ifdef ODB_COMPILER) outside of that.

Regards
Sean Clarke
---------------------------------------------
SEC Consulting Limited
Phone: +44 (0)23 8040 5599
Website: http://www.sec-consulting.co.uk
Email: sean.clarke@sec-consulting.co.uk

On 14 September 2015 at 14:16, Boris Kolpackov <boris@codesynthesis.com>
wrote:

> Hi Sean,
>
> Sean Clarke <sean.clarke@sec-consulting.co.uk> writes:
>
> >     For some reason I keep getting include issues when I try and set up
> > some basic many-to-one relationships- I can get your examples to work
> > because they are in the same file, yet mine (when I split across multiple
> > headers) seem to get in a bit of a state....
>
> Have you read Section 6.3, "Circular Relationships"?
>
> Boris
>
From finjulhich at gmail.com  Mon Sep 14 16:27:52 2015
From: finjulhich at gmail.com (MM)
Date: Mon Sep 14 16:28:00 2015
Subject: [odb-users] Re: 2-table query
In-Reply-To: <boris.20150914164211@codesynthesis.com>
References: <CADsB1iDM5J7FcTLwuKD5uij_rNi=i249bjkpnTHUGZmAxvWivQ@mail.gmail.com>
	<CADsB1iAicGKh5OhitdSSpDe1BS7Djk11HbXoy2D_Y9fLT+r4iQ@mail.gmail.com>
	<boris.20150907183432@codesynthesis.com>
	<CADsB1iCJVCGQ2VnJadsdv=dDVZcxrccar-_24Zuqd8Pqii0erA@mail.gmail.com>
	<boris.20150914164211@codesynthesis.com>
Message-ID: <CADsB1iBD-MxARHcSwtoPH3FoF-UJr4mxszr5Ea1FOkn1jFUiVA@mail.gmail.com>

On 14 September 2015 at 15:45, Boris Kolpackov <boris@codesynthesis.com>
wrote:

> MM <finjulhich@gmail.com> writes:
>
> > odb generates the ?xx files, but it doesn't compile
>
> What is the error?!
>
>
> > #pragma db view object(F) object(M: F::get_m)\
> >   query( F::month.in (M::month1, M::month2) )
>
> ODB in() function only works on values, not members. I guess that's
> the error but I shouldn't be guessing.
>
> Boris
>

I see. I changed that. Thanks.

For the pragma syntax:

*object(**name* [*=* *alias*] [*join-type*] [*:* *join-condition*]*)*

In my case, I have 2 associated objects. F and M.

One can obtain the M object from the F object through this public member
function

struct F : public Base {
...
   const M* getM() const override;
};

I couldn't find out how to express the *join-condition.*
I tried (and used 2 aliases as well) this:

#pragma db view object(::NS::NS1::F = fctrct)   object(::NS::NS2::M = fmkt:
fctrct::getM)

 The error by the odb compiler is:

error: unable to resolve data member 'getM' specified with db pragma object

Basically, I don't understand the syntax of the join-condition. The example
in the manual show actual data members. I couldn't find an example with a
function member .

Rds,
From markus at markusklemm.net  Tue Sep 15 05:23:26 2015
From: markus at markusklemm.net (Markus Klemm)
Date: Tue Sep 15 05:23:41 2015
Subject: [odb-users] Order, Group by and Where in a view simultaneously
Message-ID: <55C52526-8827-42A7-AA6F-DA0AFBC23FD6@markusklemm.net>

In addition to Bruce Cresantas old question, I've got a question e.g. feature request: I know how to use a view for aggregate functions, and everything works, except when I also want to order that results and at the same moment have a where clause too. Because AFAIK I have to choose where to set the (?) placeholder in the views query pragma. Either I place it before the group-by clause for a where clause or behind for the order-by clause. Either way one gets a syntax expcetion from the server.

Because I think that could be a nontrivial problem for ODB because of the SQL syntax ordering; I guess this is rather a feature request.
It's not a big deal because I can use a compile-time static order-by clause in the view, but as many details, it's nice to have.

Or is there something I missed in the manual?

Example:

#pragma db view object(measurement) object(order) query((?) + "GROUP BY" + order::order_number)
		class measurement_view{
		public:
			int32_t order_number;

#pragma db column("count(" + measurement::measurement_id + ")")
			std::size_t count;
		};

	}
void get(std::unique_ptr<odb::database> db){
	odb::session s; 
	odb::transaction t(db->begin());
	t.tracer(odb::stderr_tracer);
	typedef odb::query<measurement_view> query_t;
	query_t query(query_t::measurement::tests_passed == false);
	query += query_t("ORDER BY" + odb::query<measurement_view>::order::order_number + "DESC");
	odb::result<measurement_view> result(db->query<measurement_view>(query));
}

Regards

Markus Klemm

From andrew at a-cunningham.com  Mon Sep 14 15:10:43 2015
From: andrew at a-cunningham.com (Andrew Cunningham)
Date: Tue Sep 15 05:31:07 2015
Subject: [odb-users] SQLIte "INSERT" slows dramatically...
In-Reply-To: <boris.20150914203254@codesynthesis.com>
References: <CAGTJuHgHmmq_CpkAoSrSQBmgZ=X4=VcGijFXArNJ2_7ExGzQWw@mail.gmail.com>
	<boris.20150914203254@codesynthesis.com>
Message-ID: <CAGTJuHhv6MbBomjAZTQJu8Goa7cgfENT43nR6e0fAvOHQ3GpQw@mail.gmail.com>

Actually, I **just** tracked down the problem. It was my fault, but the
symptoms are interesting.

I am using SQLite with "deferred" foreign keys.
My data structure is a medium number (100's) of subsystems each pointing to
"thousands" of elements.
A small number of the elements may also point to an auxiliary object ( this
object is typically NULL in the element data structure).

The sudden SQLite  performance degradation happened when one of my
"elements"  was persisted but was pointing to a non-persistent "auxiliary"
object.

This is a violation of a foreign key constraint,and would have caused an
exception in immediate mode, or on "commit" in deferred mode -  but I never
got to that commit stage as it would have taken 24 hours at least.

The key point I want to make here is that although this foreign key
constraint was violated for only a handful of "elements" the performance of
SQLite was degraded 1000x from that point on.




On Mon, Sep 14, 2015 at 11:36 AM, Boris Kolpackov <boris@codesynthesis.com>
wrote:

> Hi Andrew,
>
> These two pages list various tuning options for using SQLite with
> very large databases. In particular, I would try the WAL journal.
>
>
> http://stackoverflow.com/questions/10209438/sqlite-end-transaction-is-taking-too-long
>
> http://stackoverflow.com/a/6533930
>
> Boris
>
From boris at codesynthesis.com  Tue Sep 15 05:56:05 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Tue Sep 15 05:56:16 2015
Subject: [odb-users] Problem with QDate and Sqlite when it's mapped with
	TEXT type
In-Reply-To: <27bcd53e22fc41819a170195d668732e@exch.crcis.local>
References: <9c80cadbfcf747068d7a391042134ea4@exch.crcis.local>
	<boris.20150907182731@codesynthesis.com>
	<a322c6dec57b495b95e9a352d1d8a62f@exch.crcis.local>
	<boris.20150910162523@codesynthesis.com>
	<27bcd53e22fc41819a170195d668732e@exch.crcis.local>
Message-ID: <boris.20150915115434@codesynthesis.com>

Hi Ali,

> Using QLocate::c() static function, the problem of the proposed
> fix can be solved.
> 
> std::string s(QLocale::c().toString(v, "yyyy-MM-dd").toLatin1().constData());

Ok, that looks better. Will apply the fix for the next release.

Boris

From christian at gsvitec.com  Tue Sep 15 06:21:16 2015
From: christian at gsvitec.com (Christian Sell)
Date: Tue Sep 15 06:21:20 2015
Subject: [odb-users] Schema evolution and upward compatibility
Message-ID: <508878647.7268330.1442312476864.JavaMail.open-xchange@omgreatgod.store>

Hello,
 
we have a requirement that our application not only provide forward schema
migration (i.e., a newer version of the app will be able to access and migrate a
schema created with an older version), but also upward schema compatibility. By
this I mean that an older application will be able to access a newer schema and
still behave gracefully.
 
Now, I understand that this will only work if the application has been designed
with that compatibility in mind, but there are a few aspects that rely on ODB,
namely:
 
1. if we enable schema versioning and forward migration, can we tell ODB to also
accept newer schemas (without migrating of course) or will QDB simply refuse to
cooperate
2. if 1. is true, would we be able to traverse polymorphic relationships where
new classes have been defined and are referenced in the relationship? Our
expectation would be that non-mapped classes would simply be ignored.
 
example:
 
schema 1:   classA 1---N-> classB
                     classB ---subclass--> classB1
 
schema 2:   classA 1---N-> classB
                     classB ---subclass--> classB1
                     classB ---subclass--> classB2
 
now, when I traverse the relationship a->b in the app based on schema1, and the
relationship contains, say, {classB1, classB2}, I would only see {classB1}
 
thanks,
Christian
From finjulhich at gmail.com  Tue Sep 15 16:43:31 2015
From: finjulhich at gmail.com (MM)
Date: Tue Sep 15 16:43:39 2015
Subject: [odb-users] Re: 2-table query
In-Reply-To: <CADsB1iBD-MxARHcSwtoPH3FoF-UJr4mxszr5Ea1FOkn1jFUiVA@mail.gmail.com>
References: <CADsB1iDM5J7FcTLwuKD5uij_rNi=i249bjkpnTHUGZmAxvWivQ@mail.gmail.com>
	<CADsB1iAicGKh5OhitdSSpDe1BS7Djk11HbXoy2D_Y9fLT+r4iQ@mail.gmail.com>
	<boris.20150907183432@codesynthesis.com>
	<CADsB1iCJVCGQ2VnJadsdv=dDVZcxrccar-_24Zuqd8Pqii0erA@mail.gmail.com>
	<boris.20150914164211@codesynthesis.com>
	<CADsB1iBD-MxARHcSwtoPH3FoF-UJr4mxszr5Ea1FOkn1jFUiVA@mail.gmail.com>
Message-ID: <CADsB1iCpHxsw=rZscKUXMe-b9PzbKCSmB6avJLo+DYue8B=XNg@mail.gmail.com>

On 14 September 2015 at 21:27, MM <finjulhich@gmail.com> wrote:

>
>
> On 14 September 2015 at 15:45, Boris Kolpackov <boris@codesynthesis.com>
> wrote:
>
>> MM <finjulhich@gmail.com> writes:
>>
>> > odb generates the ?xx files, but it doesn't compile
>>
>> What is the error?!
>>
>>
>> > #pragma db view object(F) object(M: F::get_m)\
>> >   query( F::month.in (M::month1, M::month2) )
>>
>> ODB in() function only works on values, not members. I guess that's
>> the error but I shouldn't be guessing.
>>
>> Boris
>>
>
> I see. I changed that. Thanks.
>
> For the pragma syntax:
>
> *object(**name* [*=* *alias*] [*join-type*] [*:* *join-condition*]*)*
>
> In my case, I have 2 associated objects. F and M.
>
> One can obtain the M object from the F object through this public member
> function
>
> struct F : public Base {
> ...
>    const M* getM() const override;
> };
>
> I couldn't find out how to express the *join-condition.*
> I tried (and used 2 aliases as well) this:
>
> #pragma db view object(::NS::NS1::F = fctrct)   object(::NS::NS2::M =
> fmkt: fctrct::getM)
>
>  The error by the odb compiler is:
>
> error: unable to resolve data member 'getM' specified with db pragma
> object
>
> Basically, I don't understand the syntax of the join-condition. The
> example in the manual show actual data members. I couldn't find an example
> with a function member .
>
> Rds,
>
> getM() in fact, only returns a protected data member     const M* m_;
from struct Base

I change the pragma to

#pragma db view object(::NS::NS1::F = fctrct)   object(::NS::NS2::M = fmkt:
fctrct::m_)
struct Fcurrent {
  ::NS::NS1::F* fctrct;
};

the odb compiler generates the cxx code, but compiling that code fails.

Is the join-condition limited to data members?
From a.pasternak at mt-robot.com  Wed Sep 16 05:15:22 2015
From: a.pasternak at mt-robot.com (Andreas Pasternak MT-Robot AG)
Date: Wed Sep 16 05:15:26 2015
Subject: [odb-users] Custom pointer as in pointer-traits.hxx
Message-ID: <20150916111522.EGroupware.UPAXkY5Gv6ceA9ro9tM6_A1@mail.mt-robot.com>

Hello all,

A third party framework uses a custom pointer. So I tried to create a  
simple example:

template
class mypointer {
public:
  T* ptr;
};

I tried then according to the manual but also as a hack directly in  
pointer-traits.hxx to insert:

namespace odb
{
template
  class pointer_traits >
  {
  public:
  static const pointer_kind kind = pk_unique;
  static const bool lazy = false;

  typedef T element_type;
  typedef mypointer pointer_type;
  typedef mypointer const_pointer_type;
  typedef smart_ptr_guard
  guard;
  };
} //end odb

I then get the error message:

  error: unable to map C++ type '::mypointer' used in data member  
'myptr' to a MySQL database type

Other pointers (std::shared_ptr etc.) work. I also tried to copy&past  
the std::shared_ptr code and adapt it, leaving it as close to the  
original as possible. I just have no clue why it does not find my  
template spezialisation.

But it does compile it, if I have an error in the code it complains.  
Also my container specialisation and string specialisation I made work  
flawlessly. I just don't get the pointers to work.

Thank you,

Andreas
From fredrik.strand at qliro.com  Wed Sep 16 05:06:24 2015
From: fredrik.strand at qliro.com (Fredrik Strand)
Date: Wed Sep 16 10:08:25 2015
Subject: [odb-users] Set Query timeout
In-Reply-To: <boris.20150914155111@codesynthesis.com>
References: <AM3PR02MB03358CA068950C0C90B9782895D0@AM3PR02MB033.eurprd02.prod.outlook.com>
	<boris.20150914155111@codesynthesis.com>
Message-ID: <AM3PR02MB0336634851349A7F7B65913895B0@AM3PR02MB033.eurprd02.prod.outlook.com>

Hi,

I'm currently using a query object to call a stored procedure in SQL server.

So this approach is only usable with prepared queries? Would be really nice to have a generic way to set query timeouts since it's probably a must have in enterprise applications.

/Fredrik

-----Original Message-----
From: Boris Kolpackov [mailto:boris@codesynthesis.com] 
Sent: Monday, September 14, 2015 4:11 PM
To: Fredrik Strand
Cc: odb-users@codesynthesis.com
Subject: Re: [odb-users] Set Query timeout

Hi Fredrik,

Fredrik Strand <fredrik.strand@qliro.com> writes:

> It's possible to set query timeout using raw odbc calls. However I 
> don't find a way to set query timeout using ODB. Nor do I see any 
> support for this in the source code.

I assume you are interesting in the SQL_ATTR_QUERY_TIMEOUT statement attribute. While ODB normally hides underlying database statements, you can get hold of it for a prepared query (Section 4.5 in the manual). Here is an outline based on the example from this section:

#include <odb/mssql/error.hxx>
#include <odb/mssql/statement.hxx>

using query = odb::query<person>;
using prep_query = odb::prepared_query<person>; using result = odb::result<person>;

transaction t (db.begin ());

prep_query pq (db.prepare_query<person> ("age-query", query::age > 18));

// Set timeout.
//
{
  auto& s (static_cast<odb::mssql::statement&> (pq.statement ()));

  SQLHSTMT h (s.handle ());
  SQLULEN t (10); // In seconds.
  SQLRETURN r (SQLSetStmtAttr (h, 
                               SQL_ATTR_QUERY_TIMEOUT, 
                               (SQLPOINTER) t,
                               nullptr));

  // Translate errors to exceptions using ODB mechanisms.
  //
  if (!SQL_SUCCEEDED (r))
    odb::mssql::translate_error (r, s.connection (), s); }

// Run the query.
//
result r (pq.execute ());
...

t.commit ();

Let me know if there are any problems with this approach.

Boris

From a.pasternak at mt-robot.com  Wed Sep 16 10:57:50 2015
From: a.pasternak at mt-robot.com (Andreas Pasternak MT-Robot AG)
Date: Wed Sep 16 10:57:53 2015
Subject: [odb-users] Custom pointer as in pointer-traits.hxx
Message-ID: <20150916165750.EGroupware.23OV33oLKtSKOMzYEFDGkg1@mail.mt-robot.com>

Ok, I figured it out. I tried to create a pointer of a db value (an
int exactly) and not an db object.
Now everything works without any problems.

----------------original message-----------------
From: Andreas Pasternak MT-Robot AG [a.pasternak@mt-robot.com  ]
To: odb-users@codesynthesis.com  Date: Wed, 16 Sep 2015 11:15:22 +0200
-------------------------------------------------


> Hello all,
>
> A third party framework uses a custom pointer. So I tried to create   
> a simple example:
>
> template
> class mypointer {
> public:
>   T* ptr;
> };
>
> I tried then according to the manual but also as a hack directly in   
> pointer-traits.hxx to insert:
>
> namespace odb
> {
> template
>   class pointer_traits >
>   {
>   public:
>   static const pointer_kind kind = pk_unique;
>   static const bool lazy = false;
>
>   typedef T element_type;
>   typedef mypointer pointer_type;
>   typedef mypointer const_pointer_type;
>   typedef smart_ptr_guard
>   guard;
>   };
> } //end odb
>
> I then get the error message:
>
>   error: unable to map C++ type '::mypointer' used in data member    
> 'myptr' to a MySQL database type
>
> Other pointers (std::shared_ptr etc.) work. I also tried to   
> copy&past the std::shared_ptr code and adapt it, leaving it as close  
>  to the  original as possible. I just have no clue why it does not   
> find my  template spezialisation.
>
> But it does compile it, if I have an error in the code it complains.  
>  Also my container specialisation and string specialisation I made   
> work flawlessly. I just don't get the pointers to work.
>
> Thank you,
>
> Andreas
>


-- 


			MT Robot AG

			Riedstrasse 16

			CH-4222 Zwingen
			Tel. - +41 (0)61 775 20 20
			Direkt - +41 (0)61 775 20 26
			Mobile - +41 (0)76 489 21 16
			Fax - +41 (0)61 775 20 21
			Email - [ mailto:a.pasternak@mt-robot.com  %20%20 ->
a.pasternak@mt-robot.com  ]
			Web - www.mt-robot.com
Diese E-Mail k?nnte vertrauliche und/oder rechtlich gesch?tzte
Informationen enthalten.Wenn Sie nicht der richtige Adressat sind oder
diese E-Mail irrt?mlich erhalten haben, informieren Sie bitte sofort
den Absender und vernichten Sie diese Mail. Das unerlaubte Kopieren
sowie die unbefugte Weitergabe dieser E-Mail sind nicht gestattet.

This e-mail may contain confidential and/or privileged information. If
you are not the intended recipien



From boris at codesynthesis.com  Wed Sep 16 11:10:24 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Wed Sep 16 11:10:31 2015
Subject: [odb-users] Re: 2-table query
In-Reply-To: <CADsB1iCpHxsw=rZscKUXMe-b9PzbKCSmB6avJLo+DYue8B=XNg@mail.gmail.com>
References: <CADsB1iDM5J7FcTLwuKD5uij_rNi=i249bjkpnTHUGZmAxvWivQ@mail.gmail.com>
	<CADsB1iAicGKh5OhitdSSpDe1BS7Djk11HbXoy2D_Y9fLT+r4iQ@mail.gmail.com>
	<boris.20150907183432@codesynthesis.com>
	<CADsB1iCJVCGQ2VnJadsdv=dDVZcxrccar-_24Zuqd8Pqii0erA@mail.gmail.com>
	<boris.20150914164211@codesynthesis.com>
	<CADsB1iBD-MxARHcSwtoPH3FoF-UJr4mxszr5Ea1FOkn1jFUiVA@mail.gmail.com>
	<CADsB1iCpHxsw=rZscKUXMe-b9PzbKCSmB6avJLo+DYue8B=XNg@mail.gmail.com>
Message-ID: <boris.20150916170341@codesynthesis.com>

MM <finjulhich@gmail.com> writes:

> Is the join-condition limited to data members?

Just think a little: how would a C++ member function be mapped to
a relational model?

In ODB views, the JOIN condition can be specified as either a data
member, which should be a pointer to object, or you can spell the
condition literally, as a comparison of the relevant columns. In
other words, the JOIN condition in ODB views is a very thin
abstraction of the SQL JOIN condition. So next time when you are
try to "stick" something in there, find out it doesn't work, and
start writing another email to the mailing list, pause and think
how would something like this be mapped to SQL..?

Boris

From boris at codesynthesis.com  Wed Sep 16 11:13:19 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Wed Sep 16 11:13:26 2015
Subject: [odb-users] SQLIte "INSERT" slows dramatically...
In-Reply-To: <CAGTJuHhv6MbBomjAZTQJu8Goa7cgfENT43nR6e0fAvOHQ3GpQw@mail.gmail.com>
References: <CAGTJuHgHmmq_CpkAoSrSQBmgZ=X4=VcGijFXArNJ2_7ExGzQWw@mail.gmail.com>
	<boris.20150914203254@codesynthesis.com>
	<CAGTJuHhv6MbBomjAZTQJu8Goa7cgfENT43nR6e0fAvOHQ3GpQw@mail.gmail.com>
Message-ID: <boris.20150916171040@codesynthesis.com>

Hi Andrew,

Andrew Cunningham <andrew@a-cunningham.com> writes:

> The key point I want to make here is that although this foreign key
> constraint was violated for only a handful of "elements" the performance of
> SQLite was degraded 1000x from that point on.

Yes, that's indeed quite interesting, thanks for sharing! I found that
SQLite has such "dark corners" here and there.

I would suggest that you re-send your findings to the SQLite mailing
list and see what the developers say. They are usually very helpful.

Boris

From boris at codesynthesis.com  Wed Sep 16 11:20:18 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Wed Sep 16 11:20:25 2015
Subject: [odb-users] Custom pointer as in pointer-traits.hxx
In-Reply-To: <20150916165750.EGroupware.23OV33oLKtSKOMzYEFDGkg1@mail.mt-robot.com>
References: <20150916165750.EGroupware.23OV33oLKtSKOMzYEFDGkg1@mail.mt-robot.com>
Message-ID: <boris.20150916171712@codesynthesis.com>

Andreas Pasternak MT-Robot AG <a.pasternak@mt-robot.com> writes:

> Now everything works without any problems.

Great, glad to hear that. Just for reference, to add support for a 
custom smart pointer in ODB:

1. You will need to implement a pointer_traits specialization. See
   libodb/odb/pointer-traits.hxx for inspirations.

2a. The easiest way to integrate this traits implementation into
    ODB is to simply add it at the end of the header that defines
    your smart pointer.

2b. If you want to make it a bit less intrusive (i.e., you may
    be using the same smart pointer in other parts of your
    application that have nothing to do with ODB), then you can
    place this traits specialization into a separate file and
    then:

    1) Include it into the ODB compilation with the --odb-epilogue
       option, for example:

       --odb-epilogue '#include "ptr-traits.h"'

    2) Include it into the generated source code with --hxx-prologue:

       --hxx-epilogue '#include "ptr-traits.h"'

    For an example on how to do this, see libodb-boost, specifically,
    the smart-ptr.options file.


> I tried to create a pointer of a db value (an int exactly) and
> not an db object.

You can make that work as well: see wrapper_traits.

Boris

From boris at codesynthesis.com  Wed Sep 16 11:59:24 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Wed Sep 16 11:59:32 2015
Subject: [odb-users] Schema evolution and upward compatibility
In-Reply-To: <508878647.7268330.1442312476864.JavaMail.open-xchange@omgreatgod.store>
References: <508878647.7268330.1442312476864.JavaMail.open-xchange@omgreatgod.store>
Message-ID: <boris.20150916173538@codesynthesis.com>

Hi Christian,

Christian Sell <christian@gsvitec.com> writes:

> By this I mean that an older application will be able to access a
> newer schema and still behave gracefully.

Just to make it clear, ODB doesn't "officially" support this mode
and probably never will; the "normal" direction is already complex
enough.


> Now, I understand that this will only work if the application has
> been designed with that compatibility in mind [...]

Right, you will at the least have to make sure that the new schema
if always a superset of the old one. That is, not remove any objects,
data members, etc. Even then, you can run into breakages. For example,
the new version of the object might contain a non-NULLable pointer
that the old code will fail to persist. So it seems like besides
not removing anything, whatever new data members you add, they all
should have default values (see 'default' pragma).


> 1. if we enable schema versioning and forward migration, can we
>    tell ODB to also accept newer schemas (without migrating of
>    course) or will QDB simply refuse to cooperate

All ODB statements always explicitly list all the columns in SELECT-
lists, etc. In other words, we don't do "SELECT *". So, provided
the new schema is a proper superset (see above) of the old one,
the old code should theoretically work.


> 2. if 1. is true, would we be able to traverse polymorphic 
> relationships where new classes have been defined and are
> referenced in the relationship? Our expectation would be that
> non-mapped classes would simply be ignored.
>  
> example:
>  
> schema 1:   classA 1---N-> classB
>                      classB ---subclass--> classB1
>  
> schema 2:   classA 1---N-> classB
>                      classB ---subclass--> classB1
>                      classB ---subclass--> classB2
>  
> now, when I traverse the relationship a->b in the app based on
> schema1, and the relationship contains, say, {classB1, classB2},
> I would only see {classB1}

This is a good one. No, unfortunately, it won't work. What will
happen is schema1-based application will load the data from the
classB table. It will examine the type_id stored in the classB
table (root or the polymorphic hierarchy). It will see "classB1"
id for the first entry, which it knows about and will load fine.
For the second entry it will see "classB2" which it has no idea
about. So it will throw the odb::no_type_info exception.

The only way I can think of to make this work is to query the
relationship manually (i.e., with a query that returns odb::result)
and then, while iterating over the result, catch no_type_info and
skip over such unknown entries. You could probably even do this
inside a callback so that it works just like an ordinary
relationship.

Good luck.

Boris

From christian at gsvitec.com  Wed Sep 16 12:23:47 2015
From: christian at gsvitec.com (Christian Sell)
Date: Wed Sep 16 12:23:51 2015
Subject: [odb-users] Schema evolution and upward compatibility
In-Reply-To: <boris.20150916173538@codesynthesis.com>
References: <508878647.7268330.1442312476864.JavaMail.open-xchange@omgreatgod.store>
	<boris.20150916173538@codesynthesis.com>
Message-ID: <683709131.7612559.1442420628117.JavaMail.open-xchange@omgreatgod.store>

ok, thanks for the reply. So I take it that if my application contains a
 
#pragma db model version(1, 1)
 
and then encounters a model with, say, version (2, 2), ODB will not call exit()
on my application.
 
The rest of your answer is as expected. I really don't like this requirement
myself, but wasn't able to talk them out of it..
 
thanks,
Chris

> Boris Kolpackov <boris@codesynthesis.com> hat am 16. September 2015 um 17:59
> geschrieben:
>
>
> Hi Christian,
>
> Christian Sell <christian@gsvitec.com> writes:
>
> > By this I mean that an older application will be able to access a
> > newer schema and still behave gracefully.
>
> Just to make it clear, ODB doesn't "officially" support this mode
> and probably never will; the "normal" direction is already complex
> enough.
>
>
> > Now, I understand that this will only work if the application has
> > been designed with that compatibility in mind [...]
>
> Right, you will at the least have to make sure that the new schema
> if always a superset of the old one. That is, not remove any objects,
> data members, etc. Even then, you can run into breakages. For example,
> the new version of the object might contain a non-NULLable pointer
> that the old code will fail to persist. So it seems like besides
> not removing anything, whatever new data members you add, they all
> should have default values (see 'default' pragma).
>
>
> > 1. if we enable schema versioning and forward migration, can we
> > tell ODB to also accept newer schemas (without migrating of
> > course) or will QDB simply refuse to cooperate
>
> All ODB statements always explicitly list all the columns in SELECT-
> lists, etc. In other words, we don't do "SELECT *". So, provided
> the new schema is a proper superset (see above) of the old one,
> the old code should theoretically work.
>
>
> > 2. if 1. is true, would we be able to traverse polymorphic
> > relationships where new classes have been defined and are
> > referenced in the relationship? Our expectation would be that
> > non-mapped classes would simply be ignored.
> >
> > example:
> >
> > schema 1: classA 1---N-> classB
> > classB ---subclass--> classB1
> >
> > schema 2: classA 1---N-> classB
> > classB ---subclass--> classB1
> > classB ---subclass--> classB2
> >
> > now, when I traverse the relationship a->b in the app based on
> > schema1, and the relationship contains, say, {classB1, classB2},
> > I would only see {classB1}
>
> This is a good one. No, unfortunately, it won't work. What will
> happen is schema1-based application will load the data from the
> classB table. It will examine the type_id stored in the classB
> table (root or the polymorphic hierarchy). It will see "classB1"
> id for the first entry, which it knows about and will load fine.
> For the second entry it will see "classB2" which it has no idea
> about. So it will throw the odb::no_type_info exception.
>
> The only way I can think of to make this work is to query the
> relationship manually (i.e., with a query that returns odb::result)
> and then, while iterating over the result, catch no_type_info and
> skip over such unknown entries. You could probably even do this
> inside a callback so that it works just like an ordinary
> relationship.
>
> Good luck.
>
> Boris
Christian Sell

GS Vitec GmbH
Im Ziegelhaus 6-8
D-63571 Gelnhausen

mail: christian@gsvitec.com
mobil: +49 (0) 173 5384289

Tel: +49 (0) 6051 601.26-90
Fax: +49 (0) 6051 601.26-91
From boris at codesynthesis.com  Wed Sep 16 12:47:39 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Wed Sep 16 12:47:45 2015
Subject: [odb-users] Schema evolution and upward compatibility
In-Reply-To: <683709131.7612559.1442420628117.JavaMail.open-xchange@omgreatgod.store>
References: <508878647.7268330.1442312476864.JavaMail.open-xchange@omgreatgod.store>
	<boris.20150916173538@codesynthesis.com>
	<683709131.7612559.1442420628117.JavaMail.open-xchange@omgreatgod.store>
Message-ID: <boris.20150916184155@codesynthesis.com>

Hi Christian,

Christian Sell <christian@gsvitec.com> writes:

> So I take it that if my application contains a
>  
> #pragma db model version(1, 1)
>  
> and then encounters a model with, say, version (2, 2), ODB will not
> call exit() on my application.

No, it won't. You, however, can detect this situation (schema_version()
function of the database) and, perhaps, warn the user that things are
about to get rocky ;-).

Boris

From finjulhich at gmail.com  Wed Sep 16 17:22:11 2015
From: finjulhich at gmail.com (MM)
Date: Wed Sep 16 17:22:19 2015
Subject: [odb-users] Re: 2-table query
In-Reply-To: <boris.20150916170341@codesynthesis.com>
References: <CADsB1iDM5J7FcTLwuKD5uij_rNi=i249bjkpnTHUGZmAxvWivQ@mail.gmail.com>
	<CADsB1iAicGKh5OhitdSSpDe1BS7Djk11HbXoy2D_Y9fLT+r4iQ@mail.gmail.com>
	<boris.20150907183432@codesynthesis.com>
	<CADsB1iCJVCGQ2VnJadsdv=dDVZcxrccar-_24Zuqd8Pqii0erA@mail.gmail.com>
	<boris.20150914164211@codesynthesis.com>
	<CADsB1iBD-MxARHcSwtoPH3FoF-UJr4mxszr5Ea1FOkn1jFUiVA@mail.gmail.com>
	<CADsB1iCpHxsw=rZscKUXMe-b9PzbKCSmB6avJLo+DYue8B=XNg@mail.gmail.com>
	<boris.20150916170341@codesynthesis.com>
Message-ID: <CADsB1iBhTxg_aKb3BCkEmJqY3YaeJ297hB2Q_rstPQi0NNS92g@mail.gmail.com>

On 16 September 2015 at 16:10, Boris Kolpackov <boris@codesynthesis.com>
wrote:

> MM <finjulhich@gmail.com> writes:
>
> > Is the join-condition limited to data members?
>
> Just think a little: how would a C++ member function be mapped to
> a relational model?
>
> In ODB views, the JOIN condition can be specified as either a data
> member, which should be a pointer to object, or you can spell the
> condition literally, as a comparison of the relevant columns. In
> other words, the JOIN condition in ODB views is a very thin
> abstraction of the SQL JOIN condition. So next time when you are
> try to "stick" something in there, find out it doesn't work, and
> start writing another email to the mailing list, pause and think
> how would something like this be mapped to SQL..?
>
> Boris
>
>
Yes indeed it makes sense now. I suppose I assumed odb could magically make
the association given the member function  (and the fact that it returns
the associated object pointer)... But  that doesn't make sense.

Is it possible to add a little paragraph to clarify this part for the
join-condition in the case of?
 JOIN ON ( F::mid = M::id )
I could write something up and send a "pull request", I suppose, if that is
how you merge things in.

Thanks, I have made the changes, and the generated code compiles correctly.
Runtime works as well, except that I have object ownership issues that
cause a crash on exit,
which I am trying to understand.

Here is the types and pragmas again:

struct M {
...
  std::uint32_t id;
...
};

struct F {
... const M* m;   /// gets stored/loaded as a mid in the table, setting m
searchs for the right M object
  std::uint32_t  id;
...
};

#pragma db view object(F = fctrct) object(M = fmkt: fctrct::mid==fmkt::id)\
  query( fctrct::month==fmkt::activemonth || fctrct::month==fmkt::nextmonth
)
struct F_current {
  std::shared_ptr<F> fctrct;
};

and the query code is:

odb::transaction t{ db.begin() };
odb::session s;  /// as required mentioned in the manual
auto res = db.query<F_current>();
for (const auto& fc: res) {
  /// I would like to use the F* (take ownership of it)
  fc.fctrct.get();  ///this is probably wrong to take ownership of this
pointer
}

As it happens, the M objects have manually been loaded prior to this query.

After the crash, I printed the object F's address. It's the same address
for each iteration of the loop. So, odb allocates a single F_current
instance, and a single F instance with this code above, and populates it
with the result of the query for each iteration.  Is this correct?

When does the F instance get deallocated? Obviously when the last
shared_ptr pointing to it is destroyed, but when is that? Ie, in the code
above, what is the scope of the fctrct shared pointer?
When does the F_current instance get destroyed?

This is similar to the employee/employer examples... But I manually load
all the employers ahead of time, then load either all or some of the
employees later. And the employed_by_ is not a shared pointer but a raw
pointer in my case. The employers are stored in a boost ptr container, so
are the employees.

Can I get odb to allocate a new instance of F for each result from the
query?
I could just clone the F instance atevery iteration in the for loop, and
take ownership of that instead.

I appreciate all the help I can get. It is a difficult problem.

Rds,
From mne at qosmotec.com  Thu Sep 17 05:38:12 2015
From: mne at qosmotec.com (Marcel Nehring)
Date: Thu Sep 17 05:39:17 2015
Subject: [odb-users] many-to-many with two FKs
Message-ID: <fedd80d865914a2d8bf07e2a1c8b9c3d@QEX.qosmotec.com>

Hi all,

when modeling a many-to-many relationship in ODB the corresponding mapping table will have only one foreign key. Is there any way to make it two? Is there a reason not to make it two that I don't see?

Instead of

CREATE TABLE project (
name VARCHAR (255) NOT NULL PRIMARY KEY);

CREATE TABLE employee (
id BIGINT UNSIGNED NOT NULL PRIMARY KEY);

CREATE TABLE employee_projects (
object_id BIGINT UNSIGNED NOT NULL,
value VARCHAR (255) NOT NULL REFERENCES project (name));

I would like

CREATE TABLE project (
name VARCHAR (255) NOT NULL PRIMARY KEY);

CREATE TABLE employee (
id BIGINT UNSIGNED NOT NULL PRIMARY KEY);

CREATE TABLE employee_projects (
object_id BIGINT UNSIGNED NOT NULL REFERENCES employee (id),
value VARCHAR (255) NOT NULL REFERENCES project (name));

Regards,
      Marcel
From boris at codesynthesis.com  Thu Sep 17 10:57:04 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Thu Sep 17 10:57:11 2015
Subject: [odb-users] many-to-many with two FKs
In-Reply-To: <fedd80d865914a2d8bf07e2a1c8b9c3d@QEX.qosmotec.com>
References: <fedd80d865914a2d8bf07e2a1c8b9c3d@QEX.qosmotec.com>
Message-ID: <boris.20150917165345@codesynthesis.com>

Hi Marcel,

Marcel Nehring <mne@qosmotec.com> writes:

> Instead of
> 
> CREATE TABLE project (
> name VARCHAR (255) NOT NULL PRIMARY KEY);
> 
> CREATE TABLE employee (
> id BIGINT UNSIGNED NOT NULL PRIMARY KEY);
> 
> CREATE TABLE employee_projects (
> object_id BIGINT UNSIGNED NOT NULL,
> value VARCHAR (255) NOT NULL REFERENCES project (name));
> 
> I would like
> 
> [...]
> 
> CREATE TABLE employee_projects (
> object_id BIGINT UNSIGNED NOT NULL REFERENCES employee (id),
> value VARCHAR (255) NOT NULL REFERENCES project (name));

And that's pretty much what I get:

$ cat test.hxx

#include <vector>

#pragma db object
struct project
{
  #pragma db id
  int id;
};

#pragma db object
struct employee
{
  #pragma db id
  int id;

  std::vector<project*> projects;
};

$ odb -d oracle -s test.hxx

$ cat test.sql

CREATE TABLE "project" (
  "id" NUMBER(10) NOT NULL PRIMARY KEY);

CREATE TABLE "employee" (
  "id" NUMBER(10) NOT NULL PRIMARY KEY);

CREATE TABLE "employee_projects" (
  "object_id" NUMBER(10) NOT NULL,
  "index" NUMBER(20) NOT NULL,
  "value" NUMBER(10) NULL,
  CONSTRAINT "employee_projects_object_id_fk"
    FOREIGN KEY ("object_id")
    REFERENCES "employee" ("id")
    ON DELETE CASCADE,
  CONSTRAINT "employee_projects_value_fk"
    FOREIGN KEY ("value")
    REFERENCES "project" ("id")
    DEFERRABLE INITIALLY DEFERRED);

Boris

From erick.jeronimo at aeronav.avianet.cu  Thu Sep 17 11:24:16 2015
From: erick.jeronimo at aeronav.avianet.cu (Erick Jeronimo)
Date: Thu Sep 17 11:26:16 2015
Subject: [odb-users] how to include compiles odb files in Subprojects of
	QtCreator
Message-ID: <55FADB20.1070200@aeronav.avianet.cu>

How can i import/include compiled odb files (-odb.hxx,..) to Subprojects 
in QtCreator without compile again the same files already compiled in 
other subproject. I followed the tutorial 
http://wiki.codesynthesis.com/Using_ODB_with_Qt_Creator_on_Linux and 
works ok, but... how can i use the compiled files in other subproject.

thanks in advance

From adanesh at noornet.net  Fri Sep 18 00:19:57 2015
From: adanesh at noornet.net (=?utf-8?B?2LnZhNuMINiv2KfZhti0?=)
Date: Fri Sep 18 02:07:22 2015
Subject: [odb-users] BaseDaoEnabled for ODB
In-Reply-To: <b3a69b3398a04f019b9fc9310076f79b@exch.crcis.local>
References: <b3a69b3398a04f019b9fc9310076f79b@exch.crcis.local>
Message-ID: <b7b4ce7fbf0c45568d4918011333fae5@exch.crcis.local>

Hi Boris,
Is there a feature in ODB similar to 'Class BaseDaoEnabled' in OrmLite?
(http://ormlite.com/javadoc/ormlite-core/com/j256/ormlite/misc/BaseDaoEnabled.html)
I need this feature in my project, I changed the odb code to add this feature. It works up to now and I have not any problem yet, but I am not sure is it a good solution or not? Is it possible for you to review it and inform me about possible problems in it?
Steps to add this feature:

1-      Add 'BaseDaoEnabled .hxx':
#ifndef ODB_BASEDAOENABLED_HXX
#define ODB_BASEDAOENABLED_HXX
namespace odb
{
   // Base of odb objects and views:
   class BaseDaoEnabled
   {
      template <typename T, bool isBaseDaoEnabled> friend struct SetDaoEnabled;
   public:
      inline database* getDB() const { return db_; };
   protected:
      inline void setDB(database* db){ db_ = db; };
   private:
      database* db_ = nullptr;
   };
   template <typename T, bool = std::is_base_of<BaseDaoEnabled, T>::value> struct SetDaoEnabled
   {
      static inline void setDB(T*, odb::database*) {}
   };
   template <typename T> struct SetDaoEnabled < T, true >
   {
      static inline void setDB(T*, odb::database*);
   };
   template <typename T> /*static*/ inline void SetDaoEnabled<T, true>::setDB(T* p, odb::database* db)
   {
      p->BaseDaoEnabled::setDB(db);
   }
}
#endif //ODB_BASEDAOENABLED_HXX

2-      odb\forward.hxx, Line 174 (before #include <odb/post.hxx>) add:

a.  #include "odb/BaseDaoEnabled.hxx"

3-      odb\no_id_object_result.hxx, Line 17, change (a to b):

a.  pointer_type p (object_traits::create ());

b.  pointer_type p (object_traits::create (&db_));

4-      odb\simple-object-result.txx, Line 25, change (a to b):

a.       p = object_traits::create ();

b.      p = object_traits::create (&db_);

5-      odb\view_result.txx, Line 18, change (a to b):

a.       pointer_type p (view_traits::create ());

b.      pointer_type p (view_traits::create (&db_));

6-      odb\traits.hxx

a.  Line 44 & 60 & 76, change (i to ii):

                                                               i.   create ()

                                                               ii.   create (odb::database* db = nullptr)

b.      Line 48 & 64, change (i to ii):

                                                               i.      return pointer_factory<T, P>::create ();

                                                             ii.      return pointer_factory<T, P>::create (db);

c.       Line 81, add:

                                                               i.      SetDaoEnabled<T>::setDB(p, db);

7-      Generated file "%(Filename)-odb.cxx"; replace (a) with (b) by using a tool after running odb.exe:

a.       access::object_factory<object_type, pointer_type>::create ()

b.      access::object_factory<object_type, pointer_type>::create (&db)
That?s it. And now I can easily access to db in my models if they are inherited from odb::BaseDaoEnabled. If the model is not inherited from odb::BaseDaoEnabled all things happens like before. In my project each instance of my model may be from different sqlite database and in the model I need to access to db for example to do a query or load another related object and so on.
Best regards,
Ali
From boris at codesynthesis.com  Fri Sep 18 02:14:38 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Fri Sep 18 02:14:44 2015
Subject: [odb-users] BaseDaoEnabled for ODB
In-Reply-To: <b7b4ce7fbf0c45568d4918011333fae5@exch.crcis.local>
References: <b3a69b3398a04f019b9fc9310076f79b@exch.crcis.local>
	<b7b4ce7fbf0c45568d4918011333fae5@exch.crcis.local>
Message-ID: <boris.20150918081105@codesynthesis.com>

Hi Ali,

> I changed the odb code to add this feature. It works up to now and
> I have not any problem yet, but I am not sure is it a good solution
> or not? Is it possible for you to review it and inform me about
> possible problems in it?

Sorry, can't review it, it's just too messy. But I can suggest a
much cleaner solution: The idea is to use ODB database operation
callbacks (Section 14.1.7, "callback"). All that hacking you've
done to make various parts of ODB call you setDB() -- there is
actually support for this in ODB. Sometimes it pays to read the
manual ;-).

In any case, here is the outline (I hate that "DAO" term, so using
a more meaningful names):

#include <odb/core.hxx>
#include <odb/callback.hxx>

struct database_object
{
  mutable odb::database* db_ = nullptr;

  void
  init (odb::callback_event, odb::database& db) const
  {
    db_ = &db;
  }
};

#pragma db object callback(init)
struct object1: database_object
{
  ...
};

#pragma db object callback(init)
struct object2: database_object
{
  ...
};

The really cool thing about this approach is that you can set the
database not only on loads but also on persists and updates. And
you can use the callback_event argument to detect erases and reset
db_ back to nullptr, since the object is no longer persistent.

Now, if you have a persistent common base for all your objects (for
example, that defines the "standard" object id for your model), then
you won't even need to repear that callback() pragma for every
persistent class:

#pragma db object abstract callback(init)
struct database_object
{
  #pragma db id auto
  std::uint64_t id;

  #pragma db transient
  mutable odb::database* db_ = nullptr;

  void
  init (odb::callback_event, odb::database& db) const
  {
    db_ = &db;
  }
};

#pragma db object 
struct object1: database_object
{
  ...
};

#pragma db object
struct object2: database_object
{
  ...
};

Boris

From boris at codesynthesis.com  Fri Sep 18 02:18:44 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Fri Sep 18 02:18:50 2015
Subject: [odb-users] how to include compiles odb files in Subprojects of
	QtCreator
In-Reply-To: <55FADB20.1070200@aeronav.avianet.cu>
References: <55FADB20.1070200@aeronav.avianet.cu>
Message-ID: <boris.20150918081500@codesynthesis.com>

Hi Erick,

Erick Jeronimo <erick.jeronimo@aeronav.avianet.cu> writes:

> How can i import/include compiled odb files (-odb.hxx,..) to
> Subprojects in QtCreator without compile again the same files
> already compiled in other subproject.

I think you will get better help with this on QtCreator's forums
or mailing lists. Having said that, have you considered putting
the generated files into a (static) library and then linking it
to the various "Subprojects"? Surely QtCreator should have support
for something like that.

Boris

From erick.jeronimo at aeronav.avianet.cu  Fri Sep 18 09:24:19 2015
From: erick.jeronimo at aeronav.avianet.cu (Erick Jeronimo)
Date: Fri Sep 18 11:12:47 2015
Subject: [odb-users] how to include compiles odb files in Subprojects of
	QtCreator
In-Reply-To: <boris.20150918081500@codesynthesis.com>
References: <55FADB20.1070200@aeronav.avianet.cu>
	<boris.20150918081500@codesynthesis.com>
Message-ID: <55FC1083.3010006@aeronav.avianet.cu>

Hi Boris, thanks for your answer. I'll try with shared libraries instead 
of static libraries..


Thanks


On 09/18/2015 02:18 AM, Boris Kolpackov wrote:
> Hi Erick,
>
> Erick Jeronimo <erick.jeronimo@aeronav.avianet.cu> writes:
>
>> How can i import/include compiled odb files (-odb.hxx,..) to
>> Subprojects in QtCreator without compile again the same files
>> already compiled in other subproject.
> I think you will get better help with this on QtCreator's forums
> or mailing lists. Having said that, have you considered putting
> the generated files into a (static) library and then linking it
> to the various "Subprojects"? Surely QtCreator should have support
> for something like that.
>
> Boris
>
>


From christian at gsvitec.com  Sat Sep 19 06:38:57 2015
From: christian at gsvitec.com (Christian Sell)
Date: Sat Sep 19 06:39:01 2015
Subject: [odb-users] mapping complex, templatized value types
Message-ID: <1934133047.8280601.1442659137534.JavaMail.open-xchange@patina.store>

Hello,
 
I am trying to map custom value types using the value_traits approach mentioned
in the manual and exemplified in the mapping example from odb-examples. I can
succefully map a simple type like:

class Data {
string something;
}

but I am failing to map complex templatized types like:
::std::shared_ptr< ::FlexisBase::SimplePropertyImp< ::std::basic_string< char >,
::FlexisBase::NoValidation< ::std::basic_string< char >, 2 >, 2 > >

I have defined value_traits for both types, and as I said, the first type is
picked up correctly. The second already fails in the

#pragma db value(X) type("VARCHAR")

line.

Question: what restrictions exist with regard to the types that can be mapped
using this approach? What can I do to map the second type?

thanks,
 
Christian

From a.pasternak at mt-robot.com  Sat Sep 19 07:53:33 2015
From: a.pasternak at mt-robot.com (Andreas Pasternak MT-Robot AG)
Date: Sat Sep 19 07:53:34 2015
Subject: [odb-users] mapping complex, templatized value types
In-Reply-To: <1934133047.8280601.1442659137534.JavaMail.open-xchange@patina.store>
References: <1934133047.8280601.1442659137534.JavaMail.open-xchange@patina.store>
Message-ID: <20150919135333.EGroupware.3OBj12lW3Dc-phnDnrlsrbE@mail.mt-robot.com>

Did you try an C++11 type alias?

using SPI = ::std::shared_ptr< ::FlexisBase::SimplePr...
#pragma db value(SPI ) type("VARCHAR")

----------------original message-----------------
From: Christian Sell [christian@gsvitec.com  ]
To: odb-users@codesynthesis.com  Date: Sat, 19 Sep 2015 12:38:57 +0200 (CEST)
-------------------------------------------------


> Hello,
>  I am trying to map custom value types using the value_traits  
> approach mentioned
> in the manual and exemplified in the mapping example from odb-examples. I can
> succefully map a simple type like:
>
> class Data {
> string something;
> }
>
> but I am failing to map complex templatized types like:
> ::std::shared_ptr< ::FlexisBase::SimplePropertyImp<  
> ::std::basic_string< char >,
> ::FlexisBase::NoValidation< ::std::basic_string< char >, 2 >, 2 > >
>
> I have defined value_traits for both types, and as I said, the first type is
> picked up correctly. The second already fails in the
>
> #pragma db value(X) type("VARCHAR")
>
> line.
>
> Question: what restrictions exist with regard to the types that can be mapped
> using this approach? What can I do to map the second type?
>
> thanks,
>  Christian
>
>


-- 


			MT Robot AG

			Riedstrasse 16

			CH-4222 Zwingen
			Tel. - +41 (0)61 775 20 20
			Direkt - +41 (0)61 775 20 26
			Mobile - +41 (0)76 489 21 16
			Fax - +41 (0)61 775 20 21
			Email - [ mailto:a.pasternak@mt-robot.com  %20%20 ->  
a.pasternak@mt-robot.com  ]
			Web - www.mt-robot.com
Diese E-Mail k?nnte vertrauliche und/oder rechtlich gesch?tzte  
Informationen enthalten.Wenn Sie nicht der richtige Adressat sind oder  
diese E-Mail irrt?mlich erhalten haben, informieren Sie bitte sofort  
den Absender und vernichten Sie diese Mail. Das unerlaubte Kopieren  
sowie die unbefugte Weitergabe dieser E-Mail sind nicht gestattet.

This e-mail may contain confidential and/or privileged information. If  
you are not the intended recipien

From christian at gsvitec.com  Sat Sep 19 08:21:35 2015
From: christian at gsvitec.com (Christian Sell)
Date: Sat Sep 19 08:21:37 2015
Subject: [odb-users] mapping complex, templatized value types
In-Reply-To: <20150919135333.EGroupware.3OBj12lW3Dc-phnDnrlsrbE@mail.mt-robot.com>
References: <1934133047.8280601.1442659137534.JavaMail.open-xchange@patina.store>
	<20150919135333.EGroupware.3OBj12lW3Dc-phnDnrlsrbE@mail.mt-robot.com>
Message-ID: <2039437477.8287905.1442665295294.JavaMail.open-xchange@patina.store>

sure did:
 
unable to map C++ type '::DTA' used in data member 'data3' to a SQLite database
type

> Andreas Pasternak MT-Robot AG <a.pasternak@mt-robot.com> hat am 19. September
> 2015 um 13:53 geschrieben:
>
>
> Did you try an C++11 type alias?
>
> using SPI = ::std::shared_ptr< ::FlexisBase::SimplePr...
> #pragma db value(SPI ) type("VARCHAR")
>
> ----------------original message-----------------
> From: Christian Sell [christian@gsvitec.com ]
> To: odb-users@codesynthesis.com Date: Sat, 19 Sep 2015 12:38:57 +0200 (CEST)
> -------------------------------------------------
>
>
> > Hello,
> > I am trying to map custom value types using the value_traits
> > approach mentioned
> > in the manual and exemplified in the mapping example from odb-examples. I
> > can
> > succefully map a simple type like:
> >
> > class Data {
> > string something;
> > }
> >
> > but I am failing to map complex templatized types like:
> > ::std::shared_ptr< ::FlexisBase::SimplePropertyImp<
> > ::std::basic_string< char >,
> > ::FlexisBase::NoValidation< ::std::basic_string< char >, 2 >, 2 > >
> >
> > I have defined value_traits for both types, and as I said, the first type is
> > picked up correctly. The second already fails in the
> >
> > #pragma db value(X) type("VARCHAR")
> >
> > line.
> >
> > Question: what restrictions exist with regard to the types that can be
> > mapped
> > using this approach? What can I do to map the second type?
> >
> > thanks,
> > Christian
> >
> >
>
>
> --
>
>
> MT Robot AG
>
> Riedstrasse 16
>
> CH-4222 Zwingen
> Tel. - +41 (0)61 775 20 20
> Direkt - +41 (0)61 775 20 26
> Mobile - +41 (0)76 489 21 16
> Fax - +41 (0)61 775 20 21
> Email - [ mailto:a.pasternak@mt-robot.com %20%20 ->
> a.pasternak@mt-robot.com ]
> Web - www.mt-robot.com
> Diese E-Mail k?nnte vertrauliche und/oder rechtlich gesch?tzte
> Informationen enthalten.Wenn Sie nicht der richtige Adressat sind oder
> diese E-Mail irrt?mlich erhalten haben, informieren Sie bitte sofort
> den Absender und vernichten Sie diese Mail. Das unerlaubte Kopieren
> sowie die unbefugte Weitergabe dieser E-Mail sind nicht gestattet.
>
> This e-mail may contain confidential and/or privileged information. If
> you are not the intended recipien
>
Christian Sell

GS Vitec GmbH
Im Ziegelhaus 6-8
D-63571 Gelnhausen

mail: christian@gsvitec.com
mobil: +49 (0) 173 5384289

Tel: +49 (0) 6051 601.26-90
Fax: +49 (0) 6051 601.26-91
From christian at gsvitec.com  Sat Sep 19 09:01:48 2015
From: christian at gsvitec.com (Christian Sell)
Date: Sat Sep 19 09:01:52 2015
Subject: [odb-users] mapping complex, templatized value types
In-Reply-To: <20150919135333.EGroupware.3OBj12lW3Dc-phnDnrlsrbE@mail.mt-robot.com>
References: <1934133047.8280601.1442659137534.JavaMail.open-xchange@patina.store>
	<20150919135333.EGroupware.3OBj12lW3Dc-phnDnrlsrbE@mail.mt-robot.com>
Message-ID: <1313242852.8290421.1442667708750.JavaMail.open-xchange@patina.store>

an observation:
 
template instances do not work, neither directly nor through alias or typedef,
e.g.:

using DTA = ::std::shared_ptr<SimpleProperty<std::string>>; //NOT usable

subclassing the template class does work:

class DTA : public ::std::shared_ptr<SimpleProperty<std::string>> {}

this is of course no real solution. I still hope there is a workaround..

Christian

> Andreas Pasternak MT-Robot AG <a.pasternak@mt-robot.com> hat am 19. September
> 2015 um 13:53 geschrieben:
>
>
> Did you try an C++11 type alias?
>
> using SPI = ::std::shared_ptr< ::FlexisBase::SimplePr...
> #pragma db value(SPI ) type("VARCHAR")
>
> ----------------original message-----------------
> From: Christian Sell [christian@gsvitec.com ]
> To: odb-users@codesynthesis.com Date: Sat, 19 Sep 2015 12:38:57 +0200 (CEST)
> -------------------------------------------------
>
>
> > Hello,
> > I am trying to map custom value types using the value_traits
> > approach mentioned
> > in the manual and exemplified in the mapping example from odb-examples. I
> > can
> > succefully map a simple type like:
> >
> > class Data {
> > string something;
> > }
> >
> > but I am failing to map complex templatized types like:
> > ::std::shared_ptr< ::FlexisBase::SimplePropertyImp<
> > ::std::basic_string< char >,
> > ::FlexisBase::NoValidation< ::std::basic_string< char >, 2 >, 2 > >
> >
> > I have defined value_traits for both types, and as I said, the first type is
> > picked up correctly. The second already fails in the
> >
> > #pragma db value(X) type("VARCHAR")
> >
> > line.
> >
> > Question: what restrictions exist with regard to the types that can be
> > mapped
> > using this approach? What can I do to map the second type?
> >
> > thanks,
> > Christian
> >
> >
>
>
> --
>
>
> MT Robot AG
>
> Riedstrasse 16
>
> CH-4222 Zwingen
> Tel. - +41 (0)61 775 20 20
> Direkt - +41 (0)61 775 20 26
> Mobile - +41 (0)76 489 21 16
> Fax - +41 (0)61 775 20 21
> Email - [ mailto:a.pasternak@mt-robot.com %20%20 ->
> a.pasternak@mt-robot.com ]
> Web - www.mt-robot.com
> Diese E-Mail k?nnte vertrauliche und/oder rechtlich gesch?tzte
> Informationen enthalten.Wenn Sie nicht der richtige Adressat sind oder
> diese E-Mail irrt?mlich erhalten haben, informieren Sie bitte sofort
> den Absender und vernichten Sie diese Mail. Das unerlaubte Kopieren
> sowie die unbefugte Weitergabe dieser E-Mail sind nicht gestattet.
>
> This e-mail may contain confidential and/or privileged information. If
> you are not the intended recipien
>
Christian Sell

GS Vitec GmbH
Im Ziegelhaus 6-8
D-63571 Gelnhausen

mail: christian@gsvitec.com
mobil: +49 (0) 173 5384289

Tel: +49 (0) 6051 601.26-90
Fax: +49 (0) 6051 601.26-91

From boris at codesynthesis.com  Sat Sep 19 09:55:06 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Sat Sep 19 09:55:11 2015
Subject: [odb-users] mapping complex, templatized value types
In-Reply-To: <1313242852.8290421.1442667708750.JavaMail.open-xchange@patina.store>
References: <1934133047.8280601.1442659137534.JavaMail.open-xchange@patina.store>
	<20150919135333.EGroupware.3OBj12lW3Dc-phnDnrlsrbE@mail.mt-robot.com>
	<1313242852.8290421.1442667708750.JavaMail.open-xchange@patina.store>
Message-ID: <boris.20150919155222@codesynthesis.com>

Hi Christian,

Christian Sell <christian@gsvitec.com> writes:

> template instances do not work, neither directly nor through 
> alias or typedef

Strange, this should work and appear to work for me:

$ cat test.hxx

#include <memory>
#include <utility>

using int_pair = std::pair<int, int>;
#pragma db value(int_pair) type("TEXT")

using shared_short_pair = std::shared_ptr<std::pair<short, short>>;
#pragma db value(shared_short_pair) type("TEXT")

#pragma db object
struct object
{
  #pragma db id
  int id;

  int_pair ip;
  std::pair<int, int> ip1;

  std::shared_ptr<int_pair> sip; // Ok, shared_ptr is a wrapper (Section 7.3).

  shared_short_pair ssp;
  std::shared_ptr<std::pair<short, short>> ssp1;
};

$ odb -d sqlite -s --std c++11 test.hxx

Can you try to adjust this example to reproduce your problem?

Boris

From christian at gsvitec.com  Sat Sep 19 11:21:09 2015
From: christian at gsvitec.com (Christian Sell)
Date: Sat Sep 19 11:21:13 2015
Subject: [odb-users] mapping complex, templatized value types
In-Reply-To: <boris.20150919155222@codesynthesis.com>
References: <1934133047.8280601.1442659137534.JavaMail.open-xchange@patina.store>
	<20150919135333.EGroupware.3OBj12lW3Dc-phnDnrlsrbE@mail.mt-robot.com>
	<1313242852.8290421.1442667708750.JavaMail.open-xchange@patina.store>
	<boris.20150919155222@codesynthesis.com>
Message-ID: <562079123.8298655.1442676070066.JavaMail.open-xchange@patina.store>

will try.. however, I just realized that I still was on 2.3 and therefore
decided to quickly upgrade.. quickly.. first, I was not able to install
libodb-qt, because it didnt find my valid Qt 5.5 install. All attempts to fix
this via command line options failed. Then I installed the rest, ran my
makefile, and got

In file included from /usr/local/include/odb/forward.hxx:10:0,
                 from /usr/local/include/odb/nullable.hxx:10,
                 from /usr/local/include/odb/wrapper-traits.hxx:12,
                 from <standard-odb-epilogue>:5:
/opt/odb-2.4.0-x86_64-linux-gnu/lib/odb/x86_64-linux-gnu/include/c++/4.9.3/cstddef:51:11:
error: ?::max_align_t? has not been declared

so I am stuck for now..

> Boris Kolpackov <boris@codesynthesis.com> hat am 19. September 2015 um 15:55
> geschrieben:
>
>
> Hi Christian,
>
> Christian Sell <christian@gsvitec.com> writes:
>
> > template instances do not work, neither directly nor through
> > alias or typedef
>
> Strange, this should work and appear to work for me:
>
> $ cat test.hxx
>
> #include <memory>
> #include <utility>
>
> using int_pair = std::pair<int, int>;
> #pragma db value(int_pair) type("TEXT")
>
> using shared_short_pair = std::shared_ptr<std::pair<short, short>>;
> #pragma db value(shared_short_pair) type("TEXT")
>
> #pragma db object
> struct object
> {
> #pragma db id
> int id;
>
> int_pair ip;
> std::pair<int, int> ip1;
>
> std::shared_ptr<int_pair> sip; // Ok, shared_ptr is a wrapper (Section 7.3).
>
> shared_short_pair ssp;
> std::shared_ptr<std::pair<short, short>> ssp1;
> };
>
> $ odb -d sqlite -s --std c++11 test.hxx
>
> Can you try to adjust this example to reproduce your problem?
>
> Boris
Christian Sell

From christian at gsvitec.com  Sat Sep 19 12:39:11 2015
From: christian at gsvitec.com (Christian Sell)
Date: Sat Sep 19 12:39:14 2015
Subject: [odb-users] mapping complex, templatized value types
In-Reply-To: <boris.20150919155222@codesynthesis.com>
References: <1934133047.8280601.1442659137534.JavaMail.open-xchange@patina.store>
	<20150919135333.EGroupware.3OBj12lW3Dc-phnDnrlsrbE@mail.mt-robot.com>
	<1313242852.8290421.1442667708750.JavaMail.open-xchange@patina.store>
	<boris.20150919155222@codesynthesis.com>
Message-ID: <1608086312.8303521.1442680751599.JavaMail.open-xchange@patina.store>

Hello Boris,
 
I have been able to work around the max_align error by adding a #include
<cstddef> at the top of the header file containing the mapped classes. This
seems to be an issue with gcc versions.
 
I then tried the stuff you gave, and it worked. I have looked into the issues
with my classes, and it seems to me that the problem lies with the code
generated by odb, which is not compilable for my classes (the set_value method
is generated with 4 instead of 3 parameters). I have everything packaged in a
small sample app, which I could mail to you.

regards,
Christian


> Boris Kolpackov <boris@codesynthesis.com> hat am 19. September 2015 um 15:55
> geschrieben:
>
>
> Hi Christian,
>
> Christian Sell <christian@gsvitec.com> writes:
>
> > template instances do not work, neither directly nor through
> > alias or typedef
>
> Strange, this should work and appear to work for me:
>
> $ cat test.hxx
>
> #include <memory>
> #include <utility>
>
> using int_pair = std::pair<int, int>;
> #pragma db value(int_pair) type("TEXT")
>
> using shared_short_pair = std::shared_ptr<std::pair<short, short>>;
> #pragma db value(shared_short_pair) type("TEXT")
>
> #pragma db object
> struct object
> {
> #pragma db id
> int id;
>
> int_pair ip;
> std::pair<int, int> ip1;
>
> std::shared_ptr<int_pair> sip; // Ok, shared_ptr is a wrapper (Section 7.3).
>
> shared_short_pair ssp;
> std::shared_ptr<std::pair<short, short>> ssp1;
> };
>
> $ odb -d sqlite -s --std c++11 test.hxx
>
> Can you try to adjust this example to reproduce your problem?
>
> Boris
Christian Sell

GS Vitec GmbH
Im Ziegelhaus 6-8
D-63571 Gelnhausen

mail: christian@gsvitec.com
mobil: +49 (0) 173 5384289

Tel: +49 (0) 6051 601.26-90
Fax: +49 (0) 6051 601.26-91

From christian at gsvitec.com  Sat Sep 19 13:03:40 2015
From: christian at gsvitec.com (Christian Sell)
Date: Sat Sep 19 13:03:42 2015
Subject: [odb-users] mapping complex, templatized value types
In-Reply-To: <1608086312.8303521.1442680751599.JavaMail.open-xchange@patina.store>
References: <1934133047.8280601.1442659137534.JavaMail.open-xchange@patina.store>
	<20150919135333.EGroupware.3OBj12lW3Dc-phnDnrlsrbE@mail.mt-robot.com>
	<1313242852.8290421.1442667708750.JavaMail.open-xchange@patina.store>
	<boris.20150919155222@codesynthesis.com>
	<1608086312.8303521.1442680751599.JavaMail.open-xchange@patina.store>
Message-ID: <1581296639.8304951.1442682220364.JavaMail.open-xchange@patina.store>

never mind - I had the value_traits wrongly implemented. Template stuff can get
quite confusing at times..
 
So the real problem lay with version 2.3 I presume and was fixed by the upgrade
 
thanks

> Christian Sell <christian@gsvitec.com> hat am 19. September 2015 um 18:39
> geschrieben:
>
>
> Hello Boris,
>
> I have been able to work around the max_align error by adding a #include
> <cstddef> at the top of the header file containing the mapped classes. This
> seems to be an issue with gcc versions.
>
> I then tried the stuff you gave, and it worked. I have looked into the issues
> with my classes, and it seems to me that the problem lies with the code
> generated by odb, which is not compilable for my classes (the set_value method
> is generated with 4 instead of 3 parameters). I have everything packaged in a
> small sample app, which I could mail to you.
>
> regards,
> Christian
>
>
> > Boris Kolpackov <boris@codesynthesis.com> hat am 19. September 2015 um 15:55
> > geschrieben:
> >
> >
> > Hi Christian,
> >
> > Christian Sell <christian@gsvitec.com> writes:
> >
> > > template instances do not work, neither directly nor through
> > > alias or typedef
> >
> > Strange, this should work and appear to work for me:
> >
> > $ cat test.hxx
> >
> > #include <memory>
> > #include <utility>
> >
> > using int_pair = std::pair<int, int>;
> > #pragma db value(int_pair) type("TEXT")
> >
> > using shared_short_pair = std::shared_ptr<std::pair<short, short>>;
> > #pragma db value(shared_short_pair) type("TEXT")
> >
> > #pragma db object
> > struct object
> > {
> > #pragma db id
> > int id;
> >
> > int_pair ip;
> > std::pair<int, int> ip1;
> >
> > std::shared_ptr<int_pair> sip; // Ok, shared_ptr is a wrapper (Section 7.3).
> >
> > shared_short_pair ssp;
> > std::shared_ptr<std::pair<short, short>> ssp1;
> > };
> >
> > $ odb -d sqlite -s --std c++11 test.hxx
> >
> > Can you try to adjust this example to reproduce your problem?
> >
> > Boris
> Christian Sell
>
> GS Vitec GmbH
> Im Ziegelhaus 6-8
> D-63571 Gelnhausen
>
> mail: christian@gsvitec.com
> mobil: +49 (0) 173 5384289
>
> Tel: +49 (0) 6051 601.26-90
> Fax: +49 (0) 6051 601.26-91
>
Christian Sell

GS Vitec GmbH
Im Ziegelhaus 6-8
D-63571 Gelnhausen

mail: christian@gsvitec.com
mobil: +49 (0) 173 5384289

Tel: +49 (0) 6051 601.26-90
Fax: +49 (0) 6051 601.26-91
From adanesh at noornet.net  Sat Sep 19 13:29:07 2015
From: adanesh at noornet.net (=?utf-8?B?2LnZhNuMINiv2KfZhti0?=)
Date: Sat Sep 19 13:29:24 2015
Subject: [odb-users] BaseDaoEnabled for ODB
In-Reply-To: <boris.20150918081105@codesynthesis.com>
References: <b3a69b3398a04f019b9fc9310076f79b@exch.crcis.local>
	<b7b4ce7fbf0c45568d4918011333fae5@exch.crcis.local>,
	<boris.20150918081105@codesynthesis.com>
Message-ID: <780031f9ffa84dadbd7ff39813f6cab0@exch.crcis.local>

Hi Boris,
Your solution is much better than I did. I used and enjoyed. :)
Thanks,
Ali
________________________________________
From: Boris Kolpackov <boris@codesynthesis.com>
Sent: Friday, September 18, 2015 10:44 AM
To: ??? ????
Cc: odb-users@codesynthesis.com
Subject: Re: [odb-users] BaseDaoEnabled for ODB

Hi Ali,

> I changed the odb code to add this feature. It works up to now and
> I have not any problem yet, but I am not sure is it a good solution
> or not? Is it possible for you to review it and inform me about
> possible problems in it?

Sorry, can't review it, it's just too messy. But I can suggest a
much cleaner solution: The idea is to use ODB database operation
callbacks (Section 14.1.7, "callback"). All that hacking you've
done to make various parts of ODB call you setDB() -- there is
actually support for this in ODB. Sometimes it pays to read the
manual ;-).

In any case, here is the outline (I hate that "DAO" term, so using
a more meaningful names):

#include <odb/core.hxx>
#include <odb/callback.hxx>

struct database_object
{
  mutable odb::database* db_ = nullptr;

  void
  init (odb::callback_event, odb::database& db) const
  {
    db_ = &db;
  }
};

#pragma db object callback(init)
struct object1: database_object
{
  ...
};

#pragma db object callback(init)
struct object2: database_object
{
  ...
};

The really cool thing about this approach is that you can set the
database not only on loads but also on persists and updates. And
you can use the callback_event argument to detect erases and reset
db_ back to nullptr, since the object is no longer persistent.

Now, if you have a persistent common base for all your objects (for
example, that defines the "standard" object id for your model), then
you won't even need to repear that callback() pragma for every
persistent class:

#pragma db object abstract callback(init)
struct database_object
{
  #pragma db id auto
  std::uint64_t id;

  #pragma db transient
  mutable odb::database* db_ = nullptr;

  void
  init (odb::callback_event, odb::database& db) const
  {
    db_ = &db;
  }
};

#pragma db object
struct object1: database_object
{
  ...
};

#pragma db object
struct object2: database_object
{
  ...
};

Boris

From christian at gsvitec.com  Sun Sep 20 06:31:48 2015
From: christian at gsvitec.com (Christian Sell)
Date: Sun Sep 20 06:31:52 2015
Subject: [odb-users] Lazy Relationships?
Message-ID: <696336520.8341178.1442745108397.JavaMail.open-xchange@patina.store>

Hello,
 
just now I realize that when I load an object that has relationships, all
relatioships are loaded as well (and that using plain select statements without
<outer> join.. duh). This can obviously be a real performance issue. The only
solution I have found so far is using a section. However, that requires a
section instance variable in the mapped class, which is not easily done for me,
since I am mapping library classes.
 
We do, however, already have our own collection classes which I intend to make
known to ODB using the collection_traits mechanism. These containers are also
used to implement relationships (i.e., containers with shared pointers to mapped
classes).
 
My questions:
 
- does the collection_traits mechanism allow us to implement relationships in
the way described above?
- if it does, is there a way to make relationship resolution lazy, such that the
rlationship is only loaded on demand (e.g., when I say department->employees)?
 
thanks,
Christian
From christian at gsvitec.com  Mon Sep 21 05:30:20 2015
From: christian at gsvitec.com (Christian Sell)
Date: Mon Sep 21 05:30:24 2015
Subject: [odb-users] mapping complex, templatized value types
In-Reply-To: <1581296639.8304951.1442682220364.JavaMail.open-xchange@patina.store>
References: <1934133047.8280601.1442659137534.JavaMail.open-xchange@patina.store>
	<20150919135333.EGroupware.3OBj12lW3Dc-phnDnrlsrbE@mail.mt-robot.com>
	<1313242852.8290421.1442667708750.JavaMail.open-xchange@patina.store>
	<boris.20150919155222@codesynthesis.com>
	<1608086312.8303521.1442680751599.JavaMail.open-xchange@patina.store>
	<1581296639.8304951.1442682220364.JavaMail.open-xchange@patina.store>
Message-ID: <1028539932.8637773.1442827820786.JavaMail.open-xchange@ptangptang.store>

I have to again correct myself. The problem was that I was mapping long values,
and ODB generates different method signatures for those (no buffer and size but
only a value). The mapping example only contains code for string, so one has to
employ investigative effort to find the solution
 
christian
 

> Christian Sell <christian@gsvitec.com> hat am 19. September 2015 um 19:03
> geschrieben:
> 
>  never mind - I had the value_traits wrongly implemented. Template stuff can
> get quite confusing at times..
>   
>  So the real problem lay with version 2.3 I presume and was fixed by the
> upgrade
>   
>  thanks
> 
>  > Christian Sell <christian@gsvitec.com> hat am 19. September 2015 um 18:39
>  > geschrieben:
>  >
>  >
>  > Hello Boris,
>  >
>  > I have been able to work around the max_align error by adding a #include
>  > <cstddef> at the top of the header file containing the mapped classes. This
>  > seems to be an issue with gcc versions.
>  >
>  > I then tried the stuff you gave, and it worked. I have looked into the
>  > issues
>  > with my classes, and it seems to me that the problem lies with the code
>  > generated by odb, which is not compilable for my classes (the set_value
>  > method
>  > is generated with 4 instead of 3 parameters). I have everything packaged in
>  > a
>  > small sample app, which I could mail to you.
>  >
>  > regards,
>  > Christian
>  >
>  >
>  > > Boris Kolpackov <boris@codesynthesis.com> hat am 19. September 2015 um
>  > > 15:55
>  > > geschrieben:
>  > >
>  > >
>  > > Hi Christian,
>  > >
>  > > Christian Sell <christian@gsvitec.com> writes:
>  > >
>  > > > template instances do not work, neither directly nor through
>  > > > alias or typedef
>  > >
>  > > Strange, this should work and appear to work for me:
>  > >
>  > > $ cat test.hxx
>  > >
>  > > #include <memory>
>  > > #include <utility>
>  > >
>  > > using int_pair = std::pair<int, int>;
>  > > #pragma db value(int_pair) type("TEXT")
>  > >
>  > > using shared_short_pair = std::shared_ptr<std::pair<short, short>>;
>  > > #pragma db value(shared_short_pair) type("TEXT")
>  > >
>  > > #pragma db object
>  > > struct object
>  > > {
>  > > #pragma db id
>  > > int id;
>  > >
>  > > int_pair ip;
>  > > std::pair<int, int> ip1;
>  > >
>  > > std::shared_ptr<int_pair> sip; // Ok, shared_ptr is a wrapper (Section
>  > > 7.3).
>  > >
>  > > shared_short_pair ssp;
>  > > std::shared_ptr<std::pair<short, short>> ssp1;
>  > > };
>  > >
>  > > $ odb -d sqlite -s --std c++11 test.hxx
>  > >
>  > > Can you try to adjust this example to reproduce your problem?
>  > >
>  > > Boris
>  > Christian Sell
>  >
>  > GS Vitec GmbH
>  > Im Ziegelhaus 6-8
>  > D-63571 Gelnhausen
>  >
>  > mail: christian@gsvitec.com
>  > mobil: +49 (0) 173 5384289
>  >
>  > Tel: +49 (0) 6051 601.26-90
>  > Fax: +49 (0) 6051 601.26-91
>  >
>  Christian Sell
> 
>  GS Vitec GmbH
>  Im Ziegelhaus 6-8
>  D-63571 Gelnhausen
> 
>  mail: christian@gsvitec.com
>  mobil: +49 (0) 173 5384289
> 
>  Tel: +49 (0) 6051 601.26-90
>  Fax: +49 (0) 6051 601.26-91
> 

 
Christian Sell

GS Vitec GmbH
Im Ziegelhaus 6-8
D-63571 Gelnhausen

mail: christian@gsvitec.com
mobil: +49 (0) 173 5384289

Tel: +49 (0) 6051 601.26-90
Fax: +49 (0) 6051 601.26-91
From christian at gsvitec.com  Mon Sep 21 08:01:58 2015
From: christian at gsvitec.com (Christian Sell)
Date: Mon Sep 21 08:02:02 2015
Subject: [odb-users] multi-file schema?
Message-ID: <1102530626.8685727.1442836918203.JavaMail.open-xchange@ptangptang.store>

Hello,
 
I am trying to map a schema that is distributed over several header files. Right
now I have one central application header which #includes a library header file.
There is one class from that library header file which I want to map:
 
main.h
   LocalObject
  #include <library.h>
        LibraryObject
 
now, since I cannot modify the library code, I am using a pragma-only epilogue
file, which looks like:

#include "main.h"
#include "library.h"

#pragma db model version(1, 1)
#pragma db object(LibraryObject)
#pragma db member(LibraryObject::name)


I am invoking odb with the following command line: 

odb --std c++11 -x -fPIC -d sqlite --generate-query --generate-session
--generate-schema --changelog changelog --schema-format separate --schema-name
flexis --odb-epilogue-file pragamafile


as a result I get a main_odb.cpp file which #includes a non-existent
"library_odb.h" file and an empty schema

Question: does ODB support mapping classes from different headers into one
common schema? If yes, how? What am I doing wrong?

thanks,
Christian

From christian at gsvitec.com  Mon Sep 21 10:48:41 2015
From: christian at gsvitec.com (Christian Sell)
Date: Mon Sep 21 10:48:47 2015
Subject: [odb-users] multi-file schema?
In-Reply-To: <1102530626.8685727.1442836918203.JavaMail.open-xchange@ptangptang.store>
References: <1102530626.8685727.1442836918203.JavaMail.open-xchange@ptangptang.store>
Message-ID: <53276272.8736152.1442846921979.JavaMail.open-xchange@ptangptang.store>

ok, I've gotten a litle further by adding the --at-once option to the odb
command line:
 
odb --std c++11 -x -fPIC -d sqlite --generate-query --generate-session
--generate-schema --changelog changelog --schema-format separate --schema-name
--at-once main-h
 
this does generate code for LibraryObject into the cpp and h files, and
everything is compilable. But still no tables for LibraryObject in the schema.
 
?
Chris

> Christian Sell <christian@gsvitec.com> hat am 21. September 2015 um 14:01
> geschrieben:
>
>
> Hello,
>
> I am trying to map a schema that is distributed over several header files.
> Right
> now I have one central application header which #includes a library header
> file.
> There is one class from that library header file which I want to map:
>
> main.h
> LocalObject
> #include <library.h>
> LibraryObject
>
> now, since I cannot modify the library code, I am using a pragma-only epilogue
> file, which looks like:
>
> #include "main.h"
> #include "library.h"
>
> #pragma db model version(1, 1)
> #pragma db object(LibraryObject)
> #pragma db member(LibraryObject::name)
>
>
> I am invoking odb with the following command line:
>
> odb --std c++11 -x -fPIC -d sqlite --generate-query --generate-session
> --generate-schema --changelog changelog --schema-format separate --schema-name
> flexis --odb-epilogue-file pragamafile
>
>
> as a result I get a main_odb.cpp file which #includes a non-existent
> "library_odb.h" file and an empty schema
>
> Question: does ODB support mapping classes from different headers into one
> common schema? If yes, how? What am I doing wrong?
>
> thanks,
> Christian
>
From boris at codesynthesis.com  Mon Sep 21 10:51:15 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Sep 21 10:51:17 2015
Subject: [odb-users] mapping complex, templatized value types
In-Reply-To: <1028539932.8637773.1442827820786.JavaMail.open-xchange@ptangptang.store>
References: <1934133047.8280601.1442659137534.JavaMail.open-xchange@patina.store>
	<20150919135333.EGroupware.3OBj12lW3Dc-phnDnrlsrbE@mail.mt-robot.com>
	<1313242852.8290421.1442667708750.JavaMail.open-xchange@patina.store>
	<boris.20150919155222@codesynthesis.com>
	<1608086312.8303521.1442680751599.JavaMail.open-xchange@patina.store>
	<1581296639.8304951.1442682220364.JavaMail.open-xchange@patina.store>
	<1028539932.8637773.1442827820786.JavaMail.open-xchange@ptangptang.store>
Message-ID: <boris.20150921165000@codesynthesis.com>

Christian Sell <christian@gsvitec.com> writes:

> The mapping example only contains code for string, so one has to employ
> investigative effort to find the solution

... or just look in <odb/<database>/trait.hxx> for examples.

Boris

From Tim.Tao at akunacapital.com  Mon Sep 21 09:09:50 2015
From: Tim.Tao at akunacapital.com (Tim Tao)
Date: Mon Sep 21 10:52:16 2015
Subject: [odb-users] Index on subtable column "index"
Message-ID: <BLUPR07MB6744B9570D1EEF2BE75D14983460@BLUPR07MB674.namprd07.prod.outlook.com>

Hi ODB support,


I got a question about ODB subtable, so when we declare a container, e.g vector, in an ODB object, it will create a subtable for this container and a new column "index" with it, which is a very good design I think, but I am wondering why ODB will also add an index on this "index" column, is there any performance consideration? To me this index seems useless since the subtable already has an index on its parent table's primary key.


Looking forward to your reply


Thanks!

________________________________
[http://www.akunacapital.com/images/akuna.png]
Tim Tao | Junior Developer

| www.akunacapital.com <http://www.akunacapital.com>
p: | m: | f: | tim.tao@akunacapital.com

Please consider the environment, before printing this email.

This electronic message contains information from Akuna Capital LLC that may be confidential, legally privileged or otherwise protected from disclosure. This information is intended for the use of the addressee only and is not offered as investment advice to be relied upon for personal or professional use. Additionally, all electronic messages are recorded and stored in compliance pursuant to applicable SEC rules. If you are not the intended recipient, you are hereby notified that any disclosure, copying, distribution, printing or any other use of, or any action in reliance on, the contents of this electronic message is strictly prohibited. If you have received this communication in error, please notify us by telephone at (312)994-4640 and destroy the original message.
From boris at codesynthesis.com  Mon Sep 21 11:01:36 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Sep 21 11:01:39 2015
Subject: [odb-users] Index on subtable column "index"
In-Reply-To: <BLUPR07MB6744B9570D1EEF2BE75D14983460@BLUPR07MB674.namprd07.prod.outlook.com>
References: <BLUPR07MB6744B9570D1EEF2BE75D14983460@BLUPR07MB674.namprd07.prod.outlook.com>
Message-ID: <boris.20150921165937@codesynthesis.com>

Hi Tim,

Tim Tao <Tim.Tao@akunacapital.com> writes:

> I got a question about ODB subtable, so when we declare a container, 
> e.g vector, in an ODB object, it will create a subtable for this 
> container and a new column "index" with it, which is a very good
> design I think, but I am wondering why ODB will also add an index
> on this "index" column, is there any performance consideration?

Yes, the SELECT statement that loads the elements of the container,
has the 'ORDER BY index' clause.

Boris

From boris at codesynthesis.com  Mon Sep 21 11:07:13 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Sep 21 11:07:16 2015
Subject: [odb-users] multi-file schema?
In-Reply-To: <53276272.8736152.1442846921979.JavaMail.open-xchange@ptangptang.store>
References: <1102530626.8685727.1442836918203.JavaMail.open-xchange@ptangptang.store>
	<53276272.8736152.1442846921979.JavaMail.open-xchange@ptangptang.store>
Message-ID: <boris.20150921170209@codesynthesis.com>

Hi Christian,

Christian Sell <christian@gsvitec.com> writes:

> ok, I've gotten a litle further by adding the --at-once option to the odb
> command line:

Take a look also at the 'definition' pragma (Section 4.1.11). Probably
a more appropriate mechanism in this case.


> odb --std c++11 -x -fPIC -d sqlite --generate-query --generate-session
> --generate-schema --changelog changelog --schema-format separate --schema-name
> --at-once main-h
>  
> this does generate code for LibraryObject into the cpp and h files, and
> everything is compilable. But still no tables for LibraryObject in the 
> schema.

What *exactly* happens? You specified '--schema-format separate' which
means the schema should be generated into a separate file, in your
case, main-h-schema.cxx. Is this file not geneated? Is it generated
but doesn't contain the schema?

Boris

From boris at codesynthesis.com  Mon Sep 21 11:16:35 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Sep 21 11:16:37 2015
Subject: [odb-users] Lazy Relationships?
In-Reply-To: <696336520.8341178.1442745108397.JavaMail.open-xchange@patina.store>
References: <696336520.8341178.1442745108397.JavaMail.open-xchange@patina.store>
Message-ID: <boris.20150921170912@codesynthesis.com>

Hi Christian,

Christian Sell <christian@gsvitec.com> writes:

> just now I realize that when I load an object that has relationships, all
> relatioships are loaded as well (and that using plain select statements 
> without <outer> join.. duh).

See object loading views (Section 10.2) for the "duh" part.


> This can obviously be a real performance issue.

No kidding. How come we never thought about it..? Oh, wait a second,
there is this Section 6.4, "Lazy Pointers" in Chapter 6, "Relatioships".
Sounds almost exactly like the subject of this email...


> We do, however, already have our own collection classes which I intend to
> make known to ODB using the collection_traits mechanism.

It's called container_traits, not collection_traits.

Boris

From christian at gsvitec.com  Mon Sep 21 11:32:33 2015
From: christian at gsvitec.com (Christian Sell)
Date: Mon Sep 21 11:32:37 2015
Subject: [odb-users] multi-file schema?
In-Reply-To: <boris.20150921170209@codesynthesis.com>
References: <1102530626.8685727.1442836918203.JavaMail.open-xchange@ptangptang.store>
	<53276272.8736152.1442846921979.JavaMail.open-xchange@ptangptang.store>
	<boris.20150921170209@codesynthesis.com>
Message-ID: <364452775.8747101.1442849553178.JavaMail.open-xchange@ptangptang.store>

 
> Take a look also at the 'definition' pragma (Section 4.1.11). Probably
> a more appropriate mechanism in this case.
 
the docs here talk about "value types", but I have a fully mapped class that
resides in the library
 
> What *exactly* happens? You specified '--schema-format separate' which
> means the schema should be generated into a separate file, in your
> case, main-h-schema.cxx. Is this file not geneated? Is it generated
> but doesn't contain the schema?
 
it doesn't matter what schema-format I choose. There are no CREATE TABLE
statements anywhere for the mapped library class, although there is support code
that refers to it in the generated C++ files.
 
chris
From christian at gsvitec.com  Mon Sep 21 11:48:29 2015
From: christian at gsvitec.com (Christian Sell)
Date: Mon Sep 21 11:48:33 2015
Subject: [odb-users] Lazy Relationships?
In-Reply-To: <boris.20150921170912@codesynthesis.com>
References: <696336520.8341178.1442745108397.JavaMail.open-xchange@patina.store>
	<boris.20150921170912@codesynthesis.com>
Message-ID: <972643647.8750693.1442850509343.JavaMail.open-xchange@ptangptang.store>

Hello Boris,

> See object loading views (Section 10.2) for the "duh" part.
 
the duh refers to the fact that a standard load operation which pulls data from
multiple related tables does not use the obvious SQL optimization meachnisms.
There may be ways to get the optimized SQL, but I see no reason why they
shouldn't be employed transparently in the standard case as well. As an aside: I
have spent roughly a decade of my worklife writing and using object/relational
mapping tools for other languages (Smalltalk and Java), and that was the
behaviour I got used to.
 
> No kidding. How come we never thought about it..? Oh, wait a second,
> there is this Section 6.4, "Lazy Pointers" in Chapter 6, "Relatioships".
> Sounds almost exactly like the subject of this email...
 
no, not kidding. I have of course read that chapter, and even after re-reading
it I don't see how it (directly) relates to my request. I am talking about a
to-N relationship which is lazily resolved, not a pointer (which, in my book,
points to one object). Let me know if I overlooked something
 
Christian
From Tim.Tao at akunacapital.com  Mon Sep 21 11:41:23 2015
From: Tim.Tao at akunacapital.com (Tim Tao)
Date: Mon Sep 21 11:52:58 2015
Subject: [odb-users] Index on subtable column "index"
In-Reply-To: <boris.20150921165937@codesynthesis.com>
References: <BLUPR07MB6744B9570D1EEF2BE75D14983460@BLUPR07MB674.namprd07.prod.outlook.com>,
	<boris.20150921165937@codesynthesis.com>
Message-ID: <BLUPR07MB674898C644EA265D0A5092983460@BLUPR07MB674.namprd07.prod.outlook.com>

Hi Boris,

Thanks for your reply.

Another question about the subtable.

I notice all the subtables don't have primary key. I don't know if I can use the combination of parent table's primary key and "index" as the primary key of the subtable? Is there any reason we don't want to do that.

Thanks !

Tim
________________________________________
From: Boris Kolpackov <boris@codesynthesis.com>
Sent: Monday, September 21, 2015 10:01 AM
To: Tim Tao
Cc: odb-users@codesynthesis.com
Subject: Re: [odb-users] Index on subtable column "index"

Hi Tim,

Tim Tao <Tim.Tao@akunacapital.com> writes:

> I got a question about ODB subtable, so when we declare a container,
> e.g vector, in an ODB object, it will create a subtable for this
> container and a new column "index" with it, which is a very good
> design I think, but I am wondering why ODB will also add an index
> on this "index" column, is there any performance consideration?

Yes, the SELECT statement that loads the elements of the container,
has the 'ORDER BY index' clause.

Boris
________________________________
 [http://www.akunacapital.com/images/akuna.png]
Tim Tao | Junior Developer

    | www.akunacapital.com <http://www.akunacapital.com>
p:  | m:  | f:  | tim.tao@akunacapital.com

Please consider the environment, before printing this email.

This electronic message contains information from Akuna Capital LLC that may be confidential, legally privileged or otherwise protected from disclosure. This information is intended for the use of the addressee only and is not offered as investment advice to be relied upon for personal or professional use. Additionally, all electronic messages are recorded and stored in compliance pursuant to applicable SEC rules. If you are not the intended recipient, you are hereby notified that any disclosure, copying, distribution, printing or any other use of, or any action in reliance on, the contents of this electronic message is strictly prohibited. If you have received this communication in error, please notify us by telephone at (312)994-4640 and destroy the original message.

From christian at gsvitec.com  Mon Sep 21 11:54:11 2015
From: christian at gsvitec.com (Christian Sell)
Date: Mon Sep 21 11:54:13 2015
Subject: [odb-users] multi-file schema?
In-Reply-To: <boris.20150921170209@codesynthesis.com>
References: <1102530626.8685727.1442836918203.JavaMail.open-xchange@ptangptang.store>
	<53276272.8736152.1442846921979.JavaMail.open-xchange@ptangptang.store>
	<boris.20150921170209@codesynthesis.com>
Message-ID: <1736438104.8752077.1442850851639.JavaMail.open-xchange@ptangptang.store>

> What *exactly* happens? You specified '--schema-format separate' which
> means the schema should be generated into a separate file, in your
> case, main-h-schema.cxx. Is this file not geneated? Is it generated
> but doesn't contain the schema?
 
to be more specific with regard to your question: the schema file is generated
(if I use separate), and it contains CREATE TABLE statements (e.g. for the
schema_version table), but none for the library class
 
chris
From boris at codesynthesis.com  Mon Sep 21 12:04:18 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Sep 21 12:20:03 2015
Subject: [odb-users] multi-file schema?
In-Reply-To: <1736438104.8752077.1442850851639.JavaMail.open-xchange@ptangptang.store>
References: <1102530626.8685727.1442836918203.JavaMail.open-xchange@ptangptang.store>
	<53276272.8736152.1442846921979.JavaMail.open-xchange@ptangptang.store>
	<boris.20150921170209@codesynthesis.com>
	<1736438104.8752077.1442850851639.JavaMail.open-xchange@ptangptang.store>
Message-ID: <boris.20150921180100@codesynthesis.com>

Hi Christian,

Christian Sell <christian@gsvitec.com> writes:

> to be more specific with regard to your question: the schema file is 
> generated (if I use separate), and it contains CREATE TABLE statements 
> (e.g. for the schema_version table), but none for the library class

Normally, it should be you who provides the test that reproduces the
problem, so this is the last time:

$ cat test.hxx 
struct object
{
  int id;
  int value;
};

$ cat test-mapping.hxx 
#include <test.hxx>

#pragma db object(object) definition
#pragma db member(object::id) id auto

$ odb -d sqlite -s --schema-format separate --std c++11 -I. --input-name test test-mapping.hxx
$ cat test-schema.cxx

[...]

db.execute ("CREATE TABLE \"object\" (\n"
                      "  \"id\" INTEGER NOT NULL PRIMARY KEY AUTOINCREMENT,\n"
                      "  \"value\" INTEGER NOT NULL)");
[...]

Boris

From a.pasternak at mt-robot.com  Tue Sep 22 10:04:54 2015
From: a.pasternak at mt-robot.com (Andreas Pasternak MT-Robot AG)
Date: Tue Sep 22 10:04:54 2015
Subject: [odb-users] Automatically persist / update relations
Message-ID: <20150922160454.EGroupware.i7k-hBqYpI46nJDYIPlm17D@10.0.0.38>

Hello all,

I have:

#pragma db object
struct Val {
  short id;
  string value;
};
#pragma db object
struct Cont{
  std::vector> vals_rel;
}

To persist all of Cont I have to write:

Cont c;
db.persist(c);
for(val : vals_rel)
  db.persist(val);

But I'd like that ODB checks all relations automatically and inserts /  
updates them. The background is that I get the Cont class via network  
inclusive all data referenced by pointers and the recieving client  
might not have the related database entries yet.

Best regards,

Andreas
From boris at codesynthesis.com  Tue Sep 22 10:39:15 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Tue Sep 22 10:39:18 2015
Subject: [odb-users] Automatically persist / update relations
In-Reply-To: <20150922160454.EGroupware.i7k-hBqYpI46nJDYIPlm17D@10.0.0.38>
References: <20150922160454.EGroupware.i7k-hBqYpI46nJDYIPlm17D@10.0.0.38>
Message-ID: <boris.20150922163030@codesynthesis.com>

Hi Andreas,

Andreas Pasternak MT-Robot AG <a.pasternak@mt-robot.com> writes:

> But I'd like that ODB checks all relations automatically and inserts
> / updates them.

No, ODB doesn't do this. It would be too "magicy" to work reliably and
with a reasonably-good performance. For example, how does ODB know that
the object is not yet persistent? Similarly, how would it know whether
the object has changed (and thus needs to be updated)?

On the other hand, if you make a few application-specific assumptions, 
this should be pretty easy to implement yourself. For example, you can
use 0 as a special "not persistent" object id (auto ids always start
from 1). You can also have a flag that keeps track of whether the object
has changed. The final ingredient is the database operation callback
(Section 14.1.7 in the manual). In it, you can check and update/persist
all the relationships of an object, if necessary.

Boris

From boris at codesynthesis.com  Tue Sep 22 10:42:24 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Tue Sep 22 10:42:25 2015
Subject: [odb-users] Index on subtable column "index"
In-Reply-To: <BLUPR07MB674898C644EA265D0A5092983460@BLUPR07MB674.namprd07.prod.outlook.com>
References: <BLUPR07MB6744B9570D1EEF2BE75D14983460@BLUPR07MB674.namprd07.prod.outlook.com>
	<boris.20150921165937@codesynthesis.com>
	<BLUPR07MB674898C644EA265D0A5092983460@BLUPR07MB674.namprd07.prod.outlook.com>
Message-ID: <boris.20150922163936@codesynthesis.com>

Hi Tim,

Tim Tao <Tim.Tao@akunacapital.com> writes:

> Another question about the subtable.

Please in the future start another thread with proper subject
for "another question". See the posting guidelines for details:

http://www.codesynthesis.com/support/posting-guidelines.xhtml


> I notice all the subtables don't have primary key. I don't 
> know if I can use the combination of parent table's primary
> key and "index" as the primary key of the subtable? Is there
> any reason we don't want to do that.

No, the object id and index pair would be unique so could be a
primary key.

Boris

From boris at codesynthesis.com  Tue Sep 22 10:54:56 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Tue Sep 22 10:54:59 2015
Subject: [odb-users] Lazy Relationships?
In-Reply-To: <972643647.8750693.1442850509343.JavaMail.open-xchange@ptangptang.store>
References: <696336520.8341178.1442745108397.JavaMail.open-xchange@patina.store>
	<boris.20150921170912@codesynthesis.com>
	<972643647.8750693.1442850509343.JavaMail.open-xchange@ptangptang.store>
Message-ID: <boris.20150922164242@codesynthesis.com>

Hi Christian,

Christian Sell <christian@gsvitec.com> writes:

> the duh refers to the fact that a standard load operation which pulls data from
> multiple related tables does not use the obvious SQL optimization meachnisms.
> There may be ways to get the optimized SQL, but I see no reason why they
> shouldn't be employed transparently in the standard case as well.

There are reasons. Firstly, in most non-trivial applications most
relationships will be lazy, at least this is what we advocate.
Secondly, consider a to-many relationship: If you JOIN the 
relationship data to the same SELECT statement, you are going to
fetch N copies of the object. Maybe not a big deal unless N is
large or you have a couple of massive BLOBs in the object. So,
as you can see, this is not as clear cut as you make it sound.
And the "ODB way", in such situations, is not to make any
assumptions and instead give you the mechanisms and control to do
it either way (See the "Object Loading Views" Section for how
to get the behavior that you advocate).


> I am talking about a to-N relationship which is lazily resolved, not
> a pointer (which, in my book, points to one object). Let me know if
> I overlooked something

Say you have:

std::vector<odb::lazy_ptr<object>> objects;

In this setup the object ids for all the pointed-to objects will be
loaded eagerly, but not the objects themselves. To load the objects,
you will have to load them explicitly by calling load() on each 
element (lazy_ptr<object>).

Now, you may want to get even lazier and not even load object ids.
This you can only do by placing 'objects' into a lazy-loaded  section.
In fact, the chapter on sections talks about this two-level laziness.

Boris

From christian at gsvitec.com  Tue Sep 22 11:32:17 2015
From: christian at gsvitec.com (Christian Sell)
Date: Tue Sep 22 11:32:21 2015
Subject: [odb-users] Lazy Relationships?
In-Reply-To: <boris.20150922164242@codesynthesis.com>
References: <696336520.8341178.1442745108397.JavaMail.open-xchange@patina.store>
	<boris.20150921170912@codesynthesis.com>
	<972643647.8750693.1442850509343.JavaMail.open-xchange@ptangptang.store>
	<boris.20150922164242@codesynthesis.com>
Message-ID: <1890784003.8968373.1442935938142.JavaMail.open-xchange@ptangptang.store>

> Secondly, consider a to-many relationship: If you JOIN the
> relationship data to the same SELECT statement, you are going to
> fetch N copies of the object.
 
outer join. It's been done before. Fetching ids only is another performance
killer (N+1 problem). Of course there must be switches to allow adaptation to
the concrete data model (BLOBs etc.). But hey, it is as it is - my application
is not a big data cruncher.

> Now, you may want to get even lazier and not even load object ids.
> This you can only do by placing 'objects' into a lazy-loaded section.
> In fact, the chapter on sections talks about this two-level laziness.
 
yes, and in fact, I mentioned sections as the only way I had found to achieve
"full lazyness". However, this is not feasible because, as I said, I am mapping
classes that I cannot modify. Therefore my question was if I could achieve the
same with custom relationship (container) and accompanying container_traits

chris

From Tim.Tao at akunacapital.com  Tue Sep 22 13:33:16 2015
From: Tim.Tao at akunacapital.com (Tim Tao)
Date: Tue Sep 22 14:43:31 2015
Subject: [odb-users] ODB query
Message-ID: <BLUPR07MB674FF702CD36BF389E74CEB83450@BLUPR07MB674.namprd07.prod.outlook.com>

Hi Boris,


I got a question about ODB query statement, suppose I have the following odb objects:


#pragma db object

struct A{

    string name;

}


#pragma db object

struct B{

    std::shared_ptr<A> ptr;

}


Is there any way we can query table B by using condition ptr->name == "Tom" ?

Since A is a pointer in table B...I dont know how to write the query statement, I tried db->query<B>(query::name == "Tom") but got no lucky.


Thanks,


Tim



________________________________
[http://www.akunacapital.com/images/akuna.png]
Tim Tao | Junior Developer

| www.akunacapital.com <http://www.akunacapital.com>
p: | m: | f: | tim.tao@akunacapital.com

Please consider the environment, before printing this email.

This electronic message contains information from Akuna Capital LLC that may be confidential, legally privileged or otherwise protected from disclosure. This information is intended for the use of the addressee only and is not offered as investment advice to be relied upon for personal or professional use. Additionally, all electronic messages are recorded and stored in compliance pursuant to applicable SEC rules. If you are not the intended recipient, you are hereby notified that any disclosure, copying, distribution, printing or any other use of, or any action in reliance on, the contents of this electronic message is strictly prohibited. If you have received this communication in error, please notify us by telephone at (312)994-4640 and destroy the original message.
From markus at markusklemm.net  Wed Sep 23 02:50:46 2015
From: markus at markusklemm.net (Markus Klemm)
Date: Wed Sep 23 02:50:59 2015
Subject: [odb-users] ODB query
In-Reply-To: <EA347220-FEC2-4D89-BEC3-6303A1C25180@markusklemm.net>
References: <BLUPR07MB674FF702CD36BF389E74CEB83450@BLUPR07MB674.namprd07.prod.outlook.com>
	<EA347220-FEC2-4D89-BEC3-6303A1C25180@markusklemm.net>
Message-ID: <629BD919-A49F-4976-B9BF-BF45523A8352@markusklemm.net>

Forgot to send this to the list:

> Am 22.09.2015 um 23:27 schrieb Markus Klemm <markus@markusklemm.net>:
> 
> I'm not sure, since I'm writing this in the tram, but (query<B>::A::name == "Tom") should do the trick.
> 
> Regards/Mit freundlichen Gr??en
> 
> Markus Klemm
> 
> Gesendet via Mobiltelefon
> 
>> Am 22.09.2015 um 19:33 schrieb Tim Tao <Tim.Tao@akunacapital.com>:
>> 
>> (query::name == "Tom")

From Tim.Tao at akunacapital.com  Tue Sep 22 19:38:41 2015
From: Tim.Tao at akunacapital.com (Tim Tao)
Date: Wed Sep 23 11:28:17 2015
Subject: [odb-users] update object without caching the query result
Message-ID: <BLUPR07MB67479270BE40C67DBBE797783450@BLUPR07MB674.namprd07.prod.outlook.com>

Hi ODB users,


Is there any way we can update an object without caching the query result ?


I query the database and load() each result, it seems like this will cache the result in the application memory which is unnecessary.....


Any solutions?


Thanks,


Tim

________________________________
[http://www.akunacapital.com/images/akuna.png]
Tim Tao | Junior Developer

| www.akunacapital.com <http://www.akunacapital.com>
p: | m: | f: | tim.tao@akunacapital.com

Please consider the environment, before printing this email.

This electronic message contains information from Akuna Capital LLC that may be confidential, legally privileged or otherwise protected from disclosure. This information is intended for the use of the addressee only and is not offered as investment advice to be relied upon for personal or professional use. Additionally, all electronic messages are recorded and stored in compliance pursuant to applicable SEC rules. If you are not the intended recipient, you are hereby notified that any disclosure, copying, distribution, printing or any other use of, or any action in reliance on, the contents of this electronic message is strictly prohibited. If you have received this communication in error, please notify us by telephone at (312)994-4640 and destroy the original message.
From boris at codesynthesis.com  Wed Sep 23 11:34:08 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Wed Sep 23 11:34:09 2015
Subject: [odb-users] ODB query
In-Reply-To: <BLUPR07MB674FF702CD36BF389E74CEB83450@BLUPR07MB674.namprd07.prod.outlook.com>
References: <BLUPR07MB674FF702CD36BF389E74CEB83450@BLUPR07MB674.namprd07.prod.outlook.com>
Message-ID: <boris.20150923173111@codesynthesis.com>

Hi Tim,

Tim Tao <Tim.Tao@akunacapital.com> writes:

> Is there any way we can query table B by using condition ptr->name == "Tom"?

Yes, that's exactly how you would do it: A::ptr->name == "Tom".

As always, the manual is your friend. Quoting from Chapter 6, 
"Relationships":

"

We can also use data members from pointed-to objects in database queries
(Chapter 4, "Querying the Database"). For each pointer in a persistent class,
the query class defines a smart pointer-like member that contains members
corresponding to the data members in the pointed-to object. We can then use
the access via a pointer syntax (->) to refer to data members in pointed-to
objects. For example, the query class for the employee object contains the
employer member (its name is derived from the employer_ pointer) which in turn
contains the name member (its name is derived from the employer::name_ data
member of the pointed-to object). As a result, we can use the
query::employer->name expression while querying the database for the employee
objects. For example, the following transaction finds all the employees of
Example Inc that have the Doe last name:


typedef odb::query<employee> query;
typedef odb::result<employee> result;

session s;
transaction t (db.begin ());

result r (db.query<employee> (
  query::employer->name == "Example Inc" && query::last == "Doe"));

for (result::iterator i (r.begin ()); i != r.end (); ++i)
  cout << i->first_ << " " << i->last_ << endl;

t.commit ();
  

A query class member corresponding to a non-inverse (Section 6.2,
"Bidirectional Relationships") object pointer can also be used as a normal
member that has the id type of the pointed-to object. For example, the
following query locates all the employee objects that don't have an associated
employer object:

result r (db.query<employee> (query::employer.is_null ()));

"

Boris

From christian at gsvitec.com  Wed Sep 23 11:51:41 2015
From: christian at gsvitec.com (Christian Sell)
Date: Wed Sep 23 11:51:44 2015
Subject: [odb-users] custom relationship containers?
Message-ID: <540420223.9211626.1443023501886.JavaMail.open-xchange@ptangptang.store>

hello,
 
after having successfully introduced my own custom container class by
implementing the appropriate value_trais template, I would now like to also use
this container in relationships (i.e. the container holds shared_ptrs to a
mapped class). However, I am getting the meanwhile well-known error
 
error: unable to map C++ type '::FlexisBase::SharedPtrVectorProperty<
::flexis_data::Recording >' used in data member 'recordings' to a SQLite
database type
info: use '#pragma db type' to specify the database type

just to make sure: is it possible to use custom containers in relationships?
What might I be doing wrong?

thanks,
Christian

From boris at codesynthesis.com  Wed Sep 23 12:05:05 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Wed Sep 23 12:05:07 2015
Subject: [odb-users] custom relationship containers?
In-Reply-To: <540420223.9211626.1443023501886.JavaMail.open-xchange@ptangptang.store>
References: <540420223.9211626.1443023501886.JavaMail.open-xchange@ptangptang.store>
Message-ID: <boris.20150923175828@codesynthesis.com>

Hi Christian,

Christian Sell <christian@gsvitec.com> writes:

> after having successfully introduced my own custom container class by
> implementing the appropriate value_trais template, I would now like to
> also use this container in relationships (i.e. the container holds
> shared_ptrs to a mapped class).

The containers and relationships features are orthogonal, you should
be able to use a container to store pointers to objects without doing
anything special.


> However, I am getting the meanwhile well-known error
>  
> error: unable to map C++ type '::FlexisBase::SharedPtrVectorProperty<
> ::flexis_data::Recording >' used in data member 'recordings' to a SQLite
> database type

Well, this looks more like a specialized container that implicitly
stores its elements as shared_ptr's. But I think you should still
be able to use it with ODB. Just make sure that value_type in the
container_traits specialization is shared_ptr<T>, not just T.

Boris

From boris at codesynthesis.com  Wed Sep 23 12:20:54 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Wed Sep 23 12:20:56 2015
Subject: [odb-users] Lazy Relationships?
In-Reply-To: <1890784003.8968373.1442935938142.JavaMail.open-xchange@ptangptang.store>
References: <696336520.8341178.1442745108397.JavaMail.open-xchange@patina.store>
	<boris.20150921170912@codesynthesis.com>
	<972643647.8750693.1442850509343.JavaMail.open-xchange@ptangptang.store>
	<boris.20150922164242@codesynthesis.com>
	<1890784003.8968373.1442935938142.JavaMail.open-xchange@ptangptang.store>
Message-ID: <boris.20150923181632@codesynthesis.com>

Hi Christian,

Christian Sell <christian@gsvitec.com> writes:

> yes, and in fact, I mentioned sections as the only way I had found to achieve
> "full lazyness". However, this is not feasible because, as I said, I am mapping
> classes that I cannot modify. Therefore my question was if I could achieve the
> same with custom relationship (container) and accompanying container_traits

ODB needs a place to store the state of the section somewhere (is it
loaded? etc). What you could try to do is hide the section data member
inside the container. Here is an outline:

#include <odb/section.hxx>

template <typename T>
struct sectioned_vector
{
  odb::section s;
};

#pragma db object
struct object
{
  #pragma db id
  int id;

  #pragma db member(s) virtual(odb::section) load(lazy) access(v.s)

  #pragma db section(s)
  sectioned_vector<int> v;
};

Kind of cool, actually.

Boris

From christian at gsvitec.com  Wed Sep 23 12:22:59 2015
From: christian at gsvitec.com (Christian Sell)
Date: Wed Sep 23 12:23:02 2015
Subject: [odb-users] custom relationship containers?
In-Reply-To: <boris.20150923175828@codesynthesis.com>
References: <540420223.9211626.1443023501886.JavaMail.open-xchange@ptangptang.store>
	<boris.20150923175828@codesynthesis.com>
Message-ID: <54375196.9217765.1443025379954.JavaMail.open-xchange@ptangptang.store>

> Well, this looks more like a specialized container that implicitly
> stores its elements as shared_ptr's. But I think you should still
> be able to use it with ODB. Just make sure that value_type in the
> container_traits specialization is shared_ptr<T>, not just T.
 
correct, that's what it is. However, I now realize that SharedPtrVectorProperty
is a class (should have been an alias) for which I have no container_traits. I
have now switched to

VectorProperty<std::shared_ptr<Recording>>

where VectorProperty is known via container_traits, and Recording is mapped. As
a result, I get a load of compile errors from ODB generated code, starting with

error: 'index_type' in 'odb::access::object_traits_impl<flexis_data::Project,
(odb::database_id)1u>::recordings_traits::container_traits_type {aka class
odb::access::container_traits<FlexisBase::VectorProperty<std::shared_ptr<flexis_data::Recording>
> >}' does not name a type
       typedef container_traits_type::index_type index_type;
               ^
error: 'value_type' in 'odb::access::object_traits_impl<flexis_data::Project,
(odb::database_id)1u>::recordings_traits::container_traits_type {aka class
odb::access::container_traits<FlexisBase::VectorProperty<std::shared_ptr<flexis_data::Recording>
> >}' does not name a type
       typedef container_traits_type::value_type value_type;

 
any hints?
Christian

From christian at gsvitec.com  Wed Sep 23 12:26:37 2015
From: christian at gsvitec.com (Christian Sell)
Date: Wed Sep 23 12:26:40 2015
Subject: [odb-users] Lazy Relationships?
In-Reply-To: <boris.20150923181632@codesynthesis.com>
References: <696336520.8341178.1442745108397.JavaMail.open-xchange@patina.store>
	<boris.20150921170912@codesynthesis.com>
	<972643647.8750693.1442850509343.JavaMail.open-xchange@ptangptang.store>
	<boris.20150922164242@codesynthesis.com>
	<1890784003.8968373.1442935938142.JavaMail.open-xchange@ptangptang.store>
	<boris.20150923181632@codesynthesis.com>
Message-ID: <1884972911.9218519.1443025597553.JavaMail.open-xchange@ptangptang.store>

> Kind of cool, actually.
 
cool, indeed! I'll give it a try asap
 
Christian
From boris at codesynthesis.com  Wed Sep 23 12:33:08 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Wed Sep 23 12:33:10 2015
Subject: [odb-users] custom relationship containers?
In-Reply-To: <54375196.9217765.1443025379954.JavaMail.open-xchange@ptangptang.store>
References: <540420223.9211626.1443023501886.JavaMail.open-xchange@ptangptang.store>
	<boris.20150923175828@codesynthesis.com>
	<54375196.9217765.1443025379954.JavaMail.open-xchange@ptangptang.store>
Message-ID: <boris.20150923183045@codesynthesis.com>

Hi Christian,

Christian Sell <christian@gsvitec.com> writes:

>        typedef container_traits_type::index_type index_type;

Looks like your container_traits specialization does not define
types that it should. Check libodb/odb/std-vector-traits.hxx for
an example. Also make sure that the header that defines tit is
actually included into the generated header files (--hxx-prologue).

Boris

From finjulhich at gmail.com  Thu Sep 24 06:45:32 2015
From: finjulhich at gmail.com (MM)
Date: Thu Sep 24 06:45:40 2015
Subject: [odb-users] aggregate or dis-aggregate a data member
Message-ID: <CADsB1iC3w6AH1muhyyoU-HEurORZUxxrcuLkZdbOhLoXC0bueA@mail.gmail.com>

Hello, I have a need similar to this case from the manual:

consider the need to aggregate or dis-aggregate a data member:
#pragma db object
class person
{
  ...
  #pragma db transient
  std::pair<std::string, std::string> name_;
  #pragma db member(first) virtual(std::string) access(name_.first)
  #pragma db member(last) virtual(std::string) access(name_.second)

};

except that in my case, I have:
class D
{ public:
const std::string& get_type() const;
const std::uint64_t& get_id() const;
}; class C {
public:
const D* d() const;
void d(const D&);
private:
const D* d_;
};
const D* get_D( const std::string& type, std::uint64_t id );

so, I need the d member to map to 2 columns in the table, and the 2 columns
in the table to map back to a single d object, obtained through the get_D()
free function.

I apologize for asking on the list, and I think the answer is quite
straight forward, but I just can't see it :-(

Currently I have:

#pragma db object(C) abstract definition
#pragma db member(C::d) transient
#pragma db member(C::dtype) virtual(std::string)\
  get(this.d()? this.d()->get_type(): std::string())\
  set()
#pragma db member(C::did) virtual(std::uint64_t)\
  get(this.d()? this.d()->get_id(): std::uint64_t())\
  set()

For the setting part, I would need both columns to then do the get_D() call.

Rds,
MM
From mne at qosmotec.com  Thu Sep 24 07:12:43 2015
From: mne at qosmotec.com (Marcel Nehring)
Date: Thu Sep 24 07:13:32 2015
Subject: AW: [odb-users] many-to-many with two FKs
In-Reply-To: <boris.20150917165345@codesynthesis.com>
References: <fedd80d865914a2d8bf07e2a1c8b9c3d@QEX.qosmotec.com>
	<boris.20150917165345@codesynthesis.com>
Message-ID: <46674fc64d224d5095c01d15f645e953@QEX.qosmotec.com>

Hi Boris,

sorry, I didn't test that. Because the example in section 6.2.3 of the ODB manual does not create two FKs (is this a bug in the documentation?) and because for my code the CREATE TABLE command also does not I thought that is not done by default.
However, while trying to generate a small code example that reproduces my problem I discovered that ODB later uses an ALTER TABLE command to add the second FK. This is because of the order in which tables get created.
So the resulting table structure does indeed use two FKs.

Sorry for any inconvenience.

Regards,
Marcel

-----Urspr?ngliche Nachricht-----
Von: Boris Kolpackov [mailto:boris@codesynthesis.com] 
Gesendet: Donnerstag, 17. September 2015 16:57
An: Marcel Nehring <mne@qosmotec.com>
Cc: odb-users@codesynthesis.com
Betreff: Re: [odb-users] many-to-many with two FKs

Hi Marcel,

Marcel Nehring <mne@qosmotec.com> writes:

> Instead of
> 
> CREATE TABLE project (
> name VARCHAR (255) NOT NULL PRIMARY KEY);
> 
> CREATE TABLE employee (
> id BIGINT UNSIGNED NOT NULL PRIMARY KEY);
> 
> CREATE TABLE employee_projects (
> object_id BIGINT UNSIGNED NOT NULL,
> value VARCHAR (255) NOT NULL REFERENCES project (name));
> 
> I would like
> 
> [...]
> 
> CREATE TABLE employee_projects (
> object_id BIGINT UNSIGNED NOT NULL REFERENCES employee (id), value 
> VARCHAR (255) NOT NULL REFERENCES project (name));

And that's pretty much what I get:

$ cat test.hxx

#include <vector>

#pragma db object
struct project
{
  #pragma db id
  int id;
};

#pragma db object
struct employee
{
  #pragma db id
  int id;

  std::vector<project*> projects;
};

$ odb -d oracle -s test.hxx

$ cat test.sql

CREATE TABLE "project" (
  "id" NUMBER(10) NOT NULL PRIMARY KEY);

CREATE TABLE "employee" (
  "id" NUMBER(10) NOT NULL PRIMARY KEY);

CREATE TABLE "employee_projects" (
  "object_id" NUMBER(10) NOT NULL,
  "index" NUMBER(20) NOT NULL,
  "value" NUMBER(10) NULL,
  CONSTRAINT "employee_projects_object_id_fk"
    FOREIGN KEY ("object_id")
    REFERENCES "employee" ("id")
    ON DELETE CASCADE,
  CONSTRAINT "employee_projects_value_fk"
    FOREIGN KEY ("value")
    REFERENCES "project" ("id")
    DEFERRABLE INITIALLY DEFERRED);

Boris

From christian at gsvitec.com  Thu Sep 24 10:26:08 2015
From: christian at gsvitec.com (Christian Sell)
Date: Thu Sep 24 10:26:10 2015
Subject: [odb-users] access cached instances of a certain type?
Message-ID: <714566761.9410136.1443104768215.JavaMail.open-xchange@ptangptang.store>

Hello,
 
for reason of not breaking existing API  I cannot hand out shared_ptrs to my
(lightweight) mapped objects. I therefore need to determine the persistence
state explicitly. Here's the sequence
 
vector<MyClass> v = persistenceManager.getMyClasses();
 
//change objects in v, add new ones
 
persistenceManager.saveMyClasses(v);
 
in order to detect newly added objects, I would like to ask the object cache
(rather than maintaining my own). I think would need a way to get all cached
MyClass objects and check by id whether the passed-in objects match or not... I
did not find such API on odb::session, so I thought I'd ask
 
thanks,
Christian

From boris at codesynthesis.com  Thu Sep 24 11:12:36 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Thu Sep 24 11:12:37 2015
Subject: [odb-users] many-to-many with two FKs
In-Reply-To: <46674fc64d224d5095c01d15f645e953@QEX.qosmotec.com>
References: <fedd80d865914a2d8bf07e2a1c8b9c3d@QEX.qosmotec.com>
	<boris.20150917165345@codesynthesis.com>
	<46674fc64d224d5095c01d15f645e953@QEX.qosmotec.com>
Message-ID: <boris.20150924171146@codesynthesis.com>

Hi Marcel,

Marcel Nehring <mne@qosmotec.com> writes:

> Because the example in section 6.2.3 of the ODB manual does not
> create two FKs (is this a bug in the documentation?)

Yes, fixed. Thanks for pointing it out.

Boris

From picserez at gmail.com  Thu Sep 24 08:29:12 2015
From: picserez at gmail.com (Erez Pics)
Date: Thu Sep 24 11:14:56 2015
Subject: [odb-users] Query Result Problem
Message-ID: <CADk8h+Pc1aGYJ1YRLkT2whFq=Sb_M05t86wkwwDQaZKbkY3_Pg@mail.gmail.com>

 Hi Boris,

I am performing an aggregated count grouped query, somehow I get an
extra first row with invalid data, the rest of the result is as
expected.

//////////////////////////////////////////////////////
// this is the expected result from SQLite
//////////////////////////////////////////////////////
0|2
1|3
2|1
//////////////////////////////////////////////////////
// this is the result I get from ODB
//////////////////////////////////////////////////////
0|1 // <<< this is the problem
0|2
1|3
2|1

We have seen the result of 0|1 in different combinations of data, even
if the first row should be 0|1 we get it twice, not sure why it
happens, maybe something is wrong with my syntax.

I will be very happy to hear what you think.

[all information to reproduce the problem is in the text attached]

Thank you very very much,
Erez.
-------------- next part --------------

//////////////////////////////////////////////////////
// these are the actual SQL tables
//////////////////////////////////////////////////////

CREATE TABLE person_data (
	person_id INTEGER PRIMARY KEY NOT NULL,
	gender char(1) default 'U'
);

CREATE TABLE person_set (
	person_set_id INTEGER PRIMARY KEY AUTOINCREMENT,
	delete_flag tinyint UNSIGNED default '0',  -- soft erase flag
    person_id INT NOT NULL,	
	
	FOREIGN KEY(person_id) REFERENCES person_data(person_id) ON DELETE CASCADE ON UPDATE CASCADE 
);

CREATE TABLE person_lang_data (
    person_lang_data_id INTEGER PRIMARY KEY AUTOINCREMENT,
	person_set_id INT NOT NULL,
	data_language TINYINT UNSIGNED NOT NULL default '0',
	first_name varchar(255) default '',
	last_name varchar(255) default '',

	FOREIGN KEY(person_set_id) REFERENCES person_set(person_set_id) ON DELETE CASCADE ON UPDATE CASCADE
);

//////////////////////////////////////////////////////
// these are the ODB classes for the above tables
//////////////////////////////////////////////////////

#pragma db object table("person_data") pointer(shared_ptr) session
class CPersonData
{
public:

protected:

	// default empty constructor for ODB
	CPersonData()
	{
	}

	#pragma db id column("person_id") type("INT")
	long m_person_id;

	#pragma db column("gender") type("char(1)") default("U")
	char m_gender;
	
	// this is the inversed relationship side of the one to many
	#pragma db value_not_null inverse(m_person_id)
	std::vector<shared_ptr<CPersonSet> > m_vecPersonSet;
};

#pragma db object table("person_set") pointer(shared_ptr) session
class CPersonSet
{
public:

protected:

	// default empty constructor for ODB
	CPersonSet(void)
	{
	}

	friend class odb::access;

	#pragma db id auto column("person_set_id") type("INT")
	long m_person_set_id;

	#pragma db column("delete_flag") type("TINYINT UNSIGNED") default(0)
	short m_delete_flag;

	// this is the many side of the relationship with CPersonData
	#pragma db not_null
	weak_ptr<CPersonData> m_person_id;
	
	// this is the inversed relationship side of the one to many
	#pragma db value_not_null inverse(m_person_set_id)
	std::vector<shared_ptr<CPersonLangData> > m_vecLangData;
};

#pragma db object table("person_lang_data") pointer(shared_ptr) session
class CPersonLangData
{
public:

protected:

	// default empty constructor for ODB
	CPersonLangData()
	{
	}

	friend class odb::access;

	#pragma db id auto column("person_lang_data_id") type("INT")
	long m_person_lang_data_id;

	#pragma db not_null
	weak_ptr<CPersonSet> m_person_set_id;

	#pragma db column("data_language") type("TINYINT UNSIGNED") default(0)
	int m_data_language;

	#pragma db column("first_name") type("VARCHAR(255)") default("")
	string m_first_name;

	#pragma db column("last_name") type("VARCHAR(255)") default("")
	string m_last_name;
};

//////////////////////////////////////////////////////
// this is the data in the tables
//////////////////////////////////////////////////////

INSERT INTO "person_data" VALUES(1,'M');
INSERT INTO "person_data" VALUES(2,'F');
INSERT INTO "person_data" VALUES(3,'M');
INSERT INTO "person_data" VALUES(4,'F');

INSERT INTO "person_set" VALUES(1,0,1);
INSERT INTO "person_set" VALUES(2,0,2);
INSERT INTO "person_set" VALUES(3,0,3);
INSERT INTO "person_set" VALUES(4,0,4);
INSERT INTO "person_set" VALUES(5,0,1);

INSERT INTO "person_lang_data" VALUES(1,1,0,'eee','');
INSERT INTO "person_lang_data" VALUES(2,1,1,'eee','');
INSERT INTO "person_lang_data" VALUES(3,3,1,'eee','');
INSERT INTO "person_lang_data" VALUES(4,4,1,'eee','');
INSERT INTO "person_lang_data" VALUES(5,1,2,'eee','');
INSERT INTO "person_lang_data" VALUES(6,5,0,'eee','');
INSERT INTO "person_lang_data" VALUES(7,4,0,'eee','');

//////////////////////////////////////////////////////
// this is the query I am executing
//////////////////////////////////////////////////////
SELECT person_lang_data.data_language,COUNT(DISTINCT person_set.person_id) FROM person_lang_data JOIN person_set ON person_lang_data.person_set_id = person_set.person_set_id GROUP BY data_language;

//////////////////////////////////////////////////////
// this is the ODB syntax for the query above
//////////////////////////////////////////////////////
#pragma db view object(CPersonSet) \
	object(CPersonLangData: CPersonSet::m_person_set_id == CPersonLangData::m_person_set_id) \
	query((?) + "GROUP BY" + CPersonLangData::m_data_language)
struct view_persons_per_langauges
{
#pragma db column(CPersonLangData::m_data_language)
	int data_language;

#pragma db column("count (distinct " + CPersonSet::m_person_id + ")")
	long count_persons;
};

//////////////////////////////////////////////////////
// this is the expected result from SQLite
//////////////////////////////////////////////////////
0|2
1|3
2|1

//////////////////////////////////////////////////////
// this is the result I get from ODB
//////////////////////////////////////////////////////
0|1 // <<< this is the problem
0|2
1|3
2|1
From boris at codesynthesis.com  Thu Sep 24 11:26:26 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Thu Sep 24 11:26:26 2015
Subject: [odb-users] update object without caching the query result
In-Reply-To: <BLUPR07MB67479270BE40C67DBBE797783450@BLUPR07MB674.namprd07.prod.outlook.com>
References: <BLUPR07MB67479270BE40C67DBBE797783450@BLUPR07MB674.namprd07.prod.outlook.com>
Message-ID: <boris.20150924172135@codesynthesis.com>

Hi Tim,

Tim Tao <Tim.Tao@akunacapital.com> writes:

> Is there any way we can update an object without caching the 
> query result ?
> 
> I query the database and load() each result, it seems like this
> will cache the result in the application memory which is unnecessary.

There is a "query result", i.e., what's returned by the call to
query(), which, for some databases, may involve caching of the raw
data. When you call load() (presumably on result::iterator), you
load the "object". When you say "will cache the result", which
one do you refer to?

Also, it is always a good idea to show the code fragment.

Boris

From boris at codesynthesis.com  Thu Sep 24 11:36:34 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Thu Sep 24 11:36:35 2015
Subject: [odb-users] aggregate or dis-aggregate a data member
In-Reply-To: <CADsB1iC3w6AH1muhyyoU-HEurORZUxxrcuLkZdbOhLoXC0bueA@mail.gmail.com>
References: <CADsB1iC3w6AH1muhyyoU-HEurORZUxxrcuLkZdbOhLoXC0bueA@mail.gmail.com>
Message-ID: <boris.20150924172945@codesynthesis.com>

Hi,

MM <finjulhich@gmail.com> writes:

> class D
> { public:
> const std::string& get_type() const;
> const std::uint64_t& get_id() const;
> }; class C {
> public:
> const D* d() const;
> void d(const D&);
> private:
> const D* d_;
> };
> const D* get_D( const std::string& type, std::uint64_t id );
> 
> so, I need the d member to map to 2 columns in the table, and the 2 columns
> in the table to map back to a single d object, obtained through the get_D()
> free function.
> 
> Currently I have:
> 
> #pragma db object(C) abstract definition
> #pragma db member(C::d) transient
> #pragma db member(C::dtype) virtual(std::string)\
>   get(this.d()? this.d()->get_type(): std::string())\
>   set()
> #pragma db member(C::did) virtual(std::uint64_t)\
>   get(this.d()? this.d()->get_id(): std::uint64_t())\
>   set()
> 
> For the setting part, I would need both columns to then do the 
> get_D() call.

Right, and that's the reason why your approach will not work: ODB
will supply you the columns separately but you need them both at
the same time.

The way to handle this is with a composite value. Something along
these lines:

#pragma db value
struct D_Image
{
  #pragma db column("dtype")
  std::string type;

  #pragma db column("did")
  std::uint64_t id;
};

#pragma db member(C::d) virtual(D_Image) column("")          \
  get(D_Image{this.d ()->get_type (), this.d ()->get_id ()}) \
  set(this.d (get_D ((?).type, (?).id))

Boris

From Tim.Tao at akunacapital.com  Thu Sep 24 11:35:38 2015
From: Tim.Tao at akunacapital.com (Tim Tao)
Date: Thu Sep 24 11:37:29 2015
Subject: [odb-users] update object without caching the query result
In-Reply-To: <boris.20150924172135@codesynthesis.com>
References: <BLUPR07MB67479270BE40C67DBBE797783450@BLUPR07MB674.namprd07.prod.outlook.com>,
	<boris.20150924172135@codesynthesis.com>
Message-ID: <BLUPR07MB6746E8FF25C65ED99025F5F83430@BLUPR07MB674.namprd07.prod.outlook.com>

Hi Boris,

Thanks for your reply.

The following is the code fragment:

using query = odb::query<A>;
using result = odb::result<A>;

result r(db->query<A>(some conditions here));

for(result::iterator::iterator i (r.begin()); i != r.end(); ++i){
            shared_ptr<A> ptr = i.load();
            ptr->field = new_value;
            db->update(ptr);
}

So in this way I have to cache each result which seems unnecessary to me since I don't need them in the future once I finished updating

I wonder if there is any way I can update the results without loading them to cache ?

By the way, I am using postgres

Thanks,

Tim
________________________________________
From: Boris Kolpackov <boris@codesynthesis.com>
Sent: Thursday, September 24, 2015 10:26 AM
To: Tim Tao
Cc: odb-users@codesynthesis.com
Subject: Re: [odb-users] update object without caching the query result

Hi Tim,

Tim Tao <Tim.Tao@akunacapital.com> writes:

> Is there any way we can update an object without caching the
> query result ?
>
> I query the database and load() each result, it seems like this
> will cache the result in the application memory which is unnecessary.

There is a "query result", i.e., what's returned by the call to
query(), which, for some databases, may involve caching of the raw
data. When you call load() (presumably on result::iterator), you
load the "object". When you say "will cache the result", which
one do you refer to?

Also, it is always a good idea to show the code fragment.

Boris
________________________________
 [http://www.akunacapital.com/images/akuna.png]
Tim Tao | Junior Developer

    | www.akunacapital.com <http://www.akunacapital.com>
p:  | m:  | f:  | tim.tao@akunacapital.com

Please consider the environment, before printing this email.

This electronic message contains information from Akuna Capital LLC that may be confidential, legally privileged or otherwise protected from disclosure. This information is intended for the use of the addressee only and is not offered as investment advice to be relied upon for personal or professional use. Additionally, all electronic messages are recorded and stored in compliance pursuant to applicable SEC rules. If you are not the intended recipient, you are hereby notified that any disclosure, copying, distribution, printing or any other use of, or any action in reliance on, the contents of this electronic message is strictly prohibited. If you have received this communication in error, please notify us by telephone at (312)994-4640 and destroy the original message.

From boris at codesynthesis.com  Thu Sep 24 11:43:44 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Thu Sep 24 11:43:45 2015
Subject: [odb-users] Query Result Problem
In-Reply-To: <CADk8h+Pc1aGYJ1YRLkT2whFq=Sb_M05t86wkwwDQaZKbkY3_Pg@mail.gmail.com>
References: <CADk8h+Pc1aGYJ1YRLkT2whFq=Sb_M05t86wkwwDQaZKbkY3_Pg@mail.gmail.com>
Message-ID: <boris.20150924173931@codesynthesis.com>

Hi Erez,

Erez Pics <picserez@gmail.com> writes:

> SELECT person_lang_data.data_language,COUNT(DISTINCT person_set.person_id)
> FROM person_lang_data JOIN person_set ON person_lang_data.person_set_id =
> person_set.person_set_id GROUP BY data_language;
> 
> #pragma db view object(CPersonSet) \
> 	object(CPersonLangData: CPersonSet::m_person_set_id == CPersonLangData::m_person_set_id) \
> 	query((?) + "GROUP BY" + CPersonLangData::m_data_language)

One thing I noticed is that in the SELECT statement you JOIN person_set
on person_lang_data while in the view, you join CPersonLangData on
CPersonSet. Could this be the reason for the different results?

Also, I suggest that you enable statement tracing and check the actual
statement that is being executed for this view.

Boris

From boris at codesynthesis.com  Thu Sep 24 11:50:48 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Thu Sep 24 11:50:49 2015
Subject: [odb-users] access cached instances of a certain type?
In-Reply-To: <714566761.9410136.1443104768215.JavaMail.open-xchange@ptangptang.store>
References: <714566761.9410136.1443104768215.JavaMail.open-xchange@ptangptang.store>
Message-ID: <boris.20150924174642@codesynthesis.com>

Hi Christian,

Christian Sell <christian@gsvitec.com> writes:

> I think would need a way to get all cached MyClass objects and check
> by id whether the passed-in objects match or not... I did not find
> such API on odb::session, so I thought I'd ask

Did you see this?

namespace odb
{
  class session
  {
    // Low-level object cache access (iteration, etc).
    //
  public:
    typedef std::map<const std::type_info*,
                     details::shared_ptr<object_map_base>,
                     details::type_info_comparator> type_map;

    typedef std::map<database_type*, type_map> database_map;

    database_map&
    map () {return db_map_;}

    const database_map&
    map () const {return db_map_;}

  };
}

This is the start of the rabbit hole and you can go as deep as you
would like.

Also note that you can provide a custom session implementation (see
odb-tests/common/session/custom/).

Boris

From boris at codesynthesis.com  Thu Sep 24 11:58:01 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Thu Sep 24 11:58:01 2015
Subject: [odb-users] update object without caching the query result
In-Reply-To: <BLUPR07MB6746E8FF25C65ED99025F5F83430@BLUPR07MB674.namprd07.prod.outlook.com>
References: <BLUPR07MB67479270BE40C67DBBE797783450@BLUPR07MB674.namprd07.prod.outlook.com>
	<boris.20150924172135@codesynthesis.com>
	<BLUPR07MB6746E8FF25C65ED99025F5F83430@BLUPR07MB674.namprd07.prod.outlook.com>
Message-ID: <boris.20150924175107@codesynthesis.com>

Hi Tim,

Tim Tao <Tim.Tao@akunacapital.com> writes:

> using query = odb::query<A>;
> using result = odb::result<A>;
> 
> result r(db->query<A>(some conditions here));
> 
> for(result::iterator::iterator i (r.begin()); i != r.end(); ++i){
>             shared_ptr<A> ptr = i.load();
>             ptr->field = new_value;
>             db->update(ptr);
> }
> 
> So in this way I have to cache each result which seems unnecessary 
> to me since I don't need them in the future once I finished updating

Assuming A is not a session-enabled object and you don't have active
odb::session somewhere up the stack, nothing is cached here. On each
iteration, one object is loaded, updated, and freed.

If you do have a session and A is session-enabled but you don't
want to cache just this set of objects, then there are a couple
of ways you can do it. First is to load the data into a stack-
allocated object:

for(result::iterator::iterator i (r.begin()); i != r.end(); ++i){
            A a;
            i.load(a);
            a.field = new_value;
            db->update(a);
}

The other way is to temporarily "suspend" the session:

odb::session& s (odb::session::current ());
odb::session::reset_current (); // "Suspend".

for(result::iterator::iterator i (r.begin()); i != r.end(); ++i){
  ...
}

odb::session::current (s); // "Resume".

Boris

From picserez at gmail.com  Thu Sep 24 12:13:02 2015
From: picserez at gmail.com (Erez Pics)
Date: Fri Sep 25 01:24:39 2015
Subject: [odb-users] Query Result Problem
In-Reply-To: <boris.20150924173931@codesynthesis.com>
References: <CADk8h+Pc1aGYJ1YRLkT2whFq=Sb_M05t86wkwwDQaZKbkY3_Pg@mail.gmail.com>
	<boris.20150924173931@codesynthesis.com>
Message-ID: <CADk8h+NyeFVeXj_Z39-itjRZK7QO-KGY8YTMvMR7h6L21Dd0ew@mail.gmail.com>

 Hi Boris,

Wow, you are right :-)

Thank you very very much, as always you have great advice for any question.

I am unsure how to make this an inner join instead of the default left
join, the ODB compiler does not approve the syntax I use, can you
please advice how to make the view to be an inner join ?

Thank you,
Erez.

On Thu, Sep 24, 2015 at 6:43 PM, Boris Kolpackov
<boris@codesynthesis.com> wrote:
> Hi Erez,
>
> Erez Pics <picserez@gmail.com> writes:
>
>> SELECT person_lang_data.data_language,COUNT(DISTINCT person_set.person_id)
>> FROM person_lang_data JOIN person_set ON person_lang_data.person_set_id =
>> person_set.person_set_id GROUP BY data_language;
>>
>> #pragma db view object(CPersonSet) \
>>       object(CPersonLangData: CPersonSet::m_person_set_id == CPersonLangData::m_person_set_id) \
>>       query((?) + "GROUP BY" + CPersonLangData::m_data_language)
>
> One thing I noticed is that in the SELECT statement you JOIN person_set
> on person_lang_data while in the view, you join CPersonLangData on
> CPersonSet. Could this be the reason for the different results?
>
> Also, I suggest that you enable statement tracing and check the actual
> statement that is being executed for this view.
>
> Boris

From finjulhich at gmail.com  Fri Sep 25 06:11:58 2015
From: finjulhich at gmail.com (MM)
Date: Fri Sep 25 06:12:06 2015
Subject: [odb-users] aggregate or dis-aggregate a data member
In-Reply-To: <boris.20150924172945@codesynthesis.com>
References: <CADsB1iC3w6AH1muhyyoU-HEurORZUxxrcuLkZdbOhLoXC0bueA@mail.gmail.com>
	<boris.20150924172945@codesynthesis.com>
Message-ID: <CADsB1iAHJrXBP989W=+d-zax1e_JmJ+v9vn=oTLMc1gKwLLd=Q@mail.gmail.com>

On 24 September 2015 at 16:36, Boris Kolpackov <boris@codesynthesis.com>
wrote:

> Hi,
>
> MM <finjulhich@gmail.com> writes:
>
> > class D
> > { public:
> > const std::string& get_type() const;
> > const std::uint64_t& get_id() const;
> > }; class C {
> > public:
> > const D* d() const;
> > void d(const D&);
> > private:
> > const D* d_;
> > };
> > const D* get_D( const std::string& type, std::uint64_t id );
> >
> > so, I need the d member to map to 2 columns in the table, and the 2
> columns
> > in the table to map back to a single d object, obtained through the
> get_D()
> > free function.
> >
> > Currently I have:
> >
> > #pragma db object(C) abstract definition
> > #pragma db member(C::d) transient
> > #pragma db member(C::dtype) virtual(std::string)\
> >   get(this.d()? this.d()->get_type(): std::string())\
> >   set()
> > #pragma db member(C::did) virtual(std::uint64_t)\
> >   get(this.d()? this.d()->get_id(): std::uint64_t())\
> >   set()
> >
> > For the setting part, I would need both columns to then do the
> > get_D() call.
>
> Right, and that's the reason why your approach will not work: ODB
> will supply you the columns separately but you need them both at
> the same time.
>
> The way to handle this is with a composite value. Something along
> these lines:
>
> #pragma db value
> struct D_Image
> {
>   #pragma db column("dtype")
>   std::string type;
>
>   #pragma db column("did")
>   std::uint64_t id;
> };
>
> #pragma db member(C::d) virtual(D_Image) column("")          \
>   get(D_Image{this.d ()->get_type (), this.d ()->get_id ()}) \
>   set(this.d (get_D ((?).type, (?).id))
>
> Boris
>

Thanks, the composite value solution does it.

Rds,
From finjulhich at gmail.com  Fri Sep 25 06:16:18 2015
From: finjulhich at gmail.com (MM)
Date: Fri Sep 25 06:16:25 2015
Subject: [odb-users] persist non-member containers
Message-ID: <CADsB1iDhggSZJAsAZGT3=Amj+x2Aikp75iO8+OoavU3t3HisTA@mail.gmail.com>

I have a global single process-wide std::unordered_map<const D*, double>.

>From the manual, i see the container needs to be a member of a persisted
class for odb to handle it. Also, that the key cannot be an object pointer.

The D instances are stored in a global boost::ptr_xxx container.
There is a 1-to-1 equivalence between a D instance and a
std::pair<std::string, unsigned>. Given a D I get a unique pair, given the
pair I can get a D uniquely.

What sort of work around is possible if I cannot change my global container?

Rds,
From christian at gsvitec.com  Fri Sep 25 06:42:12 2015
From: christian at gsvitec.com (Christian Sell)
Date: Fri Sep 25 06:42:16 2015
Subject: [odb-users] access cached instances of a certain type?
In-Reply-To: <boris.20150924174642@codesynthesis.com>
References: <714566761.9410136.1443104768215.JavaMail.open-xchange@ptangptang.store>
	<boris.20150924174642@codesynthesis.com>
Message-ID: <192648142.9544550.1443177732703.JavaMail.open-xchange@ptangptang.store>

> > I think would need a way to get all cached MyClass objects and check
> > by id whether the passed-in objects match or not... I did not find
> > such API on odb::session, so I thought I'd ask
>
> Did you see this?
 
I surely saw it when I inspected the session API, but was appalled by the
low-level nature. After having fumbled for an hour or two, I can confirm that
initial impression ;).  At least I know that once I am done nobody will question
my indispensability..
 
My currently remaining problem is: how do I obtain a key from an object?
 
thanks,
Christian
From christian at gsvitec.com  Fri Sep 25 06:58:30 2015
From: christian at gsvitec.com (Christian Sell)
Date: Fri Sep 25 06:58:33 2015
Subject: [odb-users] access cached instances of a certain type?
In-Reply-To: <192648142.9544550.1443177732703.JavaMail.open-xchange@ptangptang.store>
References: <714566761.9410136.1443104768215.JavaMail.open-xchange@ptangptang.store>
	<boris.20150924174642@codesynthesis.com>
	<192648142.9544550.1443177732703.JavaMail.open-xchange@ptangptang.store>
Message-ID: <426002012.9548340.1443178710852.JavaMail.open-xchange@ptangptang.store>

> My currently remaining problem is: how do I obtain a key from an object?
 
ok, got it. It's in the generated code, obviously:
 
::odb::access::object_traits<MyType>::id(myobj)
 
thanks,
Chris
From boris at codesynthesis.com  Fri Sep 25 09:58:36 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Fri Sep 25 09:58:35 2015
Subject: [odb-users] Query Result Problem
In-Reply-To: <CADk8h+NyeFVeXj_Z39-itjRZK7QO-KGY8YTMvMR7h6L21Dd0ew@mail.gmail.com>
References: <CADk8h+Pc1aGYJ1YRLkT2whFq=Sb_M05t86wkwwDQaZKbkY3_Pg@mail.gmail.com>
	<boris.20150924173931@codesynthesis.com>
	<CADk8h+NyeFVeXj_Z39-itjRZK7QO-KGY8YTMvMR7h6L21Dd0ew@mail.gmail.com>
Message-ID: <boris.20150925155534@codesynthesis.com>

Hi Erez,

Erez Pics <picserez@gmail.com> writes:

> I am unsure how to make this an inner join instead of the default left
> join, the ODB compiler does not approve the syntax I use, can you
> please advice how to make the view to be an inner join ?

Just search the manual. Is it really easier to write an email than
open the manual, hit Ctrl-F, and type "inner join"? There is even
an example:

#pragma db view object(employer) \
  object(country inner: employer::name == country::name)
struct employer_named_country
{
  shared_ptr<employer> e;
  shared_ptr<country> c;
};

Boris

From boris at codesynthesis.com  Fri Sep 25 10:04:19 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Fri Sep 25 10:04:19 2015
Subject: [odb-users] persist non-member containers
In-Reply-To: <CADsB1iDhggSZJAsAZGT3=Amj+x2Aikp75iO8+OoavU3t3HisTA@mail.gmail.com>
References: <CADsB1iDhggSZJAsAZGT3=Amj+x2Aikp75iO8+OoavU3t3HisTA@mail.gmail.com>
Message-ID: <boris.20150925160052@codesynthesis.com>

Hi,

MM <finjulhich@gmail.com> writes:

> What sort of work around is possible if I cannot change my global
> container?

The same workaround that you've been using for a while now: virtual
members. With virtual members, the actual data can be anywhere as
long as you can provide suitable accessors/modifiers. So if you
create an object with a virtual data member that always returns
your global container, and you only persist a single instance of
this object, then that should do the trick.

Boris

From christian at gsvitec.com  Fri Sep 25 14:05:28 2015
From: christian at gsvitec.com (Christian Sell)
Date: Fri Sep 25 14:05:31 2015
Subject: [odb-users] query for composite value?
Message-ID: <763869791.9271410.1443204328825.JavaMail.open-xchange@omgreatgod.store>

Hello,
 
I have a mapped class with a composite key, like:
 

struct OverlayGroup
{
  struct Key {
    unsigned sourceId;
    string name;

    Key(unsigned sourceId=0, string name="") : sourceId(sourceId), name(name) {}

    bool operator<(const Key &other) const {
       return sourceId == other.sourceId ? name < other.name : sourceId <
other.sourceId;
    }
   } key;

   string description;
}

and I would like to query for the name (which is embedd inside the Key value
class). How?

thanks,
Chris

From christian at gsvitec.com  Fri Sep 25 14:13:22 2015
From: christian at gsvitec.com (Christian Sell)
Date: Fri Sep 25 14:13:24 2015
Subject: [odb-users] query for composite value?
In-Reply-To: <763869791.9271410.1443204328825.JavaMail.open-xchange@omgreatgod.store>
References: <763869791.9271410.1443204328825.JavaMail.open-xchange@omgreatgod.store>
Message-ID: <1753914102.9272241.1443204802223.JavaMail.open-xchange@omgreatgod.store>

never mind. A look at the "composite" example shows that the syntax is
 
query_t::key.name == name

> Christian Sell <christian@gsvitec.com> hat am 25. September 2015 um 20:05
> geschrieben:
>
>
> Hello,
>
> I have a mapped class with a composite key, like:
>
>
> struct OverlayGroup
> {
> struct Key {
> unsigned sourceId;
> string name;
>
> Key(unsigned sourceId=0, string name="") : sourceId(sourceId), name(name) {}
>
> bool operator<(const Key &other) const {
> return sourceId == other.sourceId ? name < other.name : sourceId <
> other.sourceId;
> }
> } key;
>
> string description;
> }
>
> and I would like to query for the name (which is embedd inside the Key value
> class). How?
>
> thanks,
> Chris
>
Christian Sell

GS Vitec GmbH
Im Ziegelhaus 6-8
D-63571 Gelnhausen

mail: christian@gsvitec.com
mobil: +49 (0) 173 5384289

Tel: +49 (0) 6051 601.26-90
Fax: +49 (0) 6051 601.26-91

From christian at gsvitec.com  Mon Sep 28 05:23:19 2015
From: christian at gsvitec.com (Christian Sell)
Date: Mon Sep 28 05:23:23 2015
Subject: [odb-users] odb persistence in a library
Message-ID: <1793361386.9709959.1443432199513.JavaMail.open-xchange@patina.store>

Hello,
 
we would like to implement persistence functionality as a shared library module
which would hold mappings for the common classes and also provide an extension
interface through which new mappings can be introduced. The idea is that these
application-specific classes would extend the schema of the library module, such
that the application classes get stored in the same database.

My question is whether this is considered a supported use case, or if there are
issues we may be running into. Specifically, I am uncertain about the
implications for schema creation and schema migration (can 2 schemas that reside
in the same database be migrated independently)
 
Christian

From picserez at gmail.com  Fri Sep 25 10:27:54 2015
From: picserez at gmail.com (Erez Pics)
Date: Mon Sep 28 11:52:50 2015
Subject: [odb-users] Query Result Problem
In-Reply-To: <boris.20150925155534@codesynthesis.com>
References: <CADk8h+Pc1aGYJ1YRLkT2whFq=Sb_M05t86wkwwDQaZKbkY3_Pg@mail.gmail.com>
	<boris.20150924173931@codesynthesis.com>
	<CADk8h+NyeFVeXj_Z39-itjRZK7QO-KGY8YTMvMR7h6L21Dd0ew@mail.gmail.com>
	<boris.20150925155534@codesynthesis.com>
Message-ID: <CADk8h+NzEMu-kXfUn+fWCTK5M+4h9dE9g-A==cG5-JM9aGrS6A@mail.gmail.com>

 Hi Boris,

Thank you for your answer, I did research the manual, it seems that to
perform an inner join I need to add the "inner" right after the second
table and before the ":"

But I got an error : 'error: ')' expected at the end of db pragma
object', maybe I am missing something,

The only change I did was adding the inner directive, this is what I mean:

//////////////////////////////////////////////////////
// this is the original ODB syntax [works by ODB compiler]
//////////////////////////////////////////////////////
#pragma db view object(CPersonSet) \
object(CPersonLangData: CPersonSet::m_person_set_id ==
CPersonLangData::m_person_set_id) \
query((?) + "GROUP BY" + CPersonLangData::m_data_language)
struct view_persons_per_langauges
{
#pragma db column(CPersonLangData::m_data_language)
int data_language;

#pragma db column("count (distinct " + CPersonSet::m_person_id + ")")
long count_persons;
};

//////////////////////////////////////////////////////
// this is the original ODB syntax [with inner directive, ODB compiler
gives an error]
//////////////////////////////////////////////////////
#pragma db view object(CPersonSet) \
object(CPersonLangData inner: CPersonSet::m_person_set_id ==
CPersonLangData::m_person_set_id) \
query((?) + "GROUP BY" + CPersonLangData::m_data_language)
struct view_persons_per_langauges
{
#pragma db column(CPersonLangData::m_data_language)
int data_language;

#pragma db column("count (distinct " + CPersonSet::m_person_id + ")")
long count_persons;
};

Thank you,
Erez.


On Fri, Sep 25, 2015 at 4:58 PM, Boris Kolpackov
<boris@codesynthesis.com> wrote:
> Hi Erez,
>
> Erez Pics <picserez@gmail.com> writes:
>
>> I am unsure how to make this an inner join instead of the default left
>> join, the ODB compiler does not approve the syntax I use, can you
>> please advice how to make the view to be an inner join ?
>
> Just search the manual. Is it really easier to write an email than
> open the manual, hit Ctrl-F, and type "inner join"? There is even
> an example:
>
> #pragma db view object(employer) \
>   object(country inner: employer::name == country::name)
> struct employer_named_country
> {
>   shared_ptr<employer> e;
>   shared_ptr<country> c;
> };
>
> Boris

From boris at codesynthesis.com  Mon Sep 28 11:55:15 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Sep 28 11:55:11 2015
Subject: [odb-users] Query Result Problem
In-Reply-To: <CADk8h+NzEMu-kXfUn+fWCTK5M+4h9dE9g-A==cG5-JM9aGrS6A@mail.gmail.com>
References: <CADk8h+Pc1aGYJ1YRLkT2whFq=Sb_M05t86wkwwDQaZKbkY3_Pg@mail.gmail.com>
	<boris.20150924173931@codesynthesis.com>
	<CADk8h+NyeFVeXj_Z39-itjRZK7QO-KGY8YTMvMR7h6L21Dd0ew@mail.gmail.com>
	<boris.20150925155534@codesynthesis.com>
	<CADk8h+NzEMu-kXfUn+fWCTK5M+4h9dE9g-A==cG5-JM9aGrS6A@mail.gmail.com>
Message-ID: <boris.20150928175357@codesynthesis.com>

Hi Erez,

Erez Pics <picserez@gmail.com> writes:

> But I got an error : 'error: ')' expected at the end of db pragma
> object', maybe I am missing something,

JOIN type support is only available in ODB 2.4.0. Which version are
you using?

Boris

From boris at codesynthesis.com  Mon Sep 28 12:35:03 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Sep 28 12:35:00 2015
Subject: [odb-users] Set Query timeout
In-Reply-To: <AM3PR02MB0336634851349A7F7B65913895B0@AM3PR02MB033.eurprd02.prod.outlook.com>
References: <AM3PR02MB03358CA068950C0C90B9782895D0@AM3PR02MB033.eurprd02.prod.outlook.com>
	<boris.20150914155111@codesynthesis.com>
	<AM3PR02MB0336634851349A7F7B65913895B0@AM3PR02MB033.eurprd02.prod.outlook.com>
Message-ID: <boris.20150928181300@codesynthesis.com>

Hi Fredrik,

Fredrik Strand <fredrik.strand@qliro.com> writes:

> I'm currently using a query object to call a stored procedure in SQL server.
> 
> So this approach is only usable with prepared queries?

Yes, that's correct. Note, however, that a non-prepared query is a
special case of the more general prepared query mechanism (i.e, it
behaves as if the query was prepared, executed, and then released).
So anything you can do with a non-prepared query you can do with a
prepared one.


> Would be really nice to have a generic way to set query timeouts
> since it's probably a must have in enterprise applications.

I agree, if there was (1) a natural place to specify it in the ODB
API and (2) uniform support for it across databases.

Specifically, regarding (1), for a simple (i.e., non-prepared) query
there is no notion of a (user-visible) statement. There is just a
query condition and a call to execute the query. Specifying the
timeout as part of the query condition doesn't feel right. So that
leaves us to passing the timeout to the query() call. Which brings
us to (2): what do we do for databases that don't support timeouts
(like SQLite). Also, timeout is just one "tunable" of a query. What
if tomorrow someone else wants something different? Do we add another
overload of query()?

On the other hand, with a prepared query, you get the notion of a
statement which you can resolve all the way down to the underlying
database handle and tune it in any way you want. And if you want
the simple query()-like interface, it is pretty easy to wrap into
a function, for example:

#include <odb/query.hxx>
#include <odb/result.hxx>

#include <odb/mssql/error.hxx>
#include <odb/mssql/statement.hxx>

template <typename T>
odb::result<T>
query_with_timeout (const odb::query<T>& q, 
                    odb::mssql::database& db,
                    unsigned short timeout)
{

  using query = odb::query<T>;
  using prep_query = odb::prepared_query<T>;  

  prep_query pq (db.prepare_query<T> ("temp-query", q));

  // Set timeout.
  //

  // ... as in the previous email.

  return pq.execute ();
}

Then, in client code, instead of:

result<person> r = db.query<person> (query::age > 18);

You do:

result<person> r = query_with_timeout<person> (query::age > 18, db, 10);

Boris

From finjulhich at gmail.com  Mon Sep 28 14:10:54 2015
From: finjulhich at gmail.com (MM)
Date: Mon Sep 28 14:11:10 2015
Subject: [odb-users] libcutl repo not exported
Message-ID: <CADsB1iBPhmBRdcHagYER7m01kHm5b=cwdcy9gyAMnGXeeMa44Q@mail.gmail.com>

fatal: remote error: access denied or repository not exported:
/odb/libcutl/libcutl.git

I am trying to build the odb compiler from git rather than from distrib
(tar file) by reading the INSTALL-GIT. libcutl is required (version 1.9.0)
but I have only 1.8.3 on my fedora.

Rds,
From finjulhich at gmail.com  Mon Sep 28 14:57:54 2015
From: finjulhich at gmail.com (MM)
Date: Mon Sep 28 14:58:01 2015
Subject: [odb-users] Re: libcutl repo not exported
In-Reply-To: <CADsB1iBPhmBRdcHagYER7m01kHm5b=cwdcy9gyAMnGXeeMa44Q@mail.gmail.com>
References: <CADsB1iBPhmBRdcHagYER7m01kHm5b=cwdcy9gyAMnGXeeMa44Q@mail.gmail.com>
Message-ID: <CADsB1iACL3+cZWrxH6EwV2Gw3Fj998e+EUhY-SJNX3VpScEVRw@mail.gmail.com>

On 28 September 2015 at 19:10, MM <finjulhich@gmail.com> wrote:

> fatal: remote error: access denied or repository not exported:
> /odb/libcutl/libcutl.git
>
> I am trying to build the odb compiler from git rather than from distrib
> (tar file) by reading the INSTALL-GIT. libcutl is required (version 1.9.0)
> but I have only 1.8.3 on my fedora.
>
> Rds,
>
I managed to install libcutl 1.9 through fedora copr. I have the 'build'
and 'cli' installed in /usr/local/ and all the other deps specififed in
INSTALL and INSTALL-GIT.

I now follow INSTALL-GIT and just do a make inside the odb directory. It
seems this file options.hxx is not generated, even

odb/odb/odb.cxx:38:27: fatal error: odb/options.hxx: No such file or
directory

so I just ran

  cli options.cli

myself, but then that fails:

cannot make target 'cli'

Rds,
From picserez at gmail.com  Mon Sep 28 14:28:30 2015
From: picserez at gmail.com (Erez Pics)
Date: Tue Sep 29 10:16:17 2015
Subject: [odb-users] Query Result Problem
In-Reply-To: <boris.20150928175357@codesynthesis.com>
References: <CADk8h+Pc1aGYJ1YRLkT2whFq=Sb_M05t86wkwwDQaZKbkY3_Pg@mail.gmail.com>
	<boris.20150924173931@codesynthesis.com>
	<CADk8h+NyeFVeXj_Z39-itjRZK7QO-KGY8YTMvMR7h6L21Dd0ew@mail.gmail.com>
	<boris.20150925155534@codesynthesis.com>
	<CADk8h+NzEMu-kXfUn+fWCTK5M+4h9dE9g-A==cG5-JM9aGrS6A@mail.gmail.com>
	<boris.20150928175357@codesynthesis.com>
Message-ID: <CADk8h+MjVn=BTpAOOZiGMdC48Xaignj3U_wmtSc3CCtHgxa-yg@mail.gmail.com>

Hi Boris,

We are using ODB 2.3.0.

1) Does upgrading to ODB 2.4.0 is the only fix to get inner join ?

2) Assuming we will upgrade, do we need to pay special attention or be
ready to handle specific issues with our existing code that use ODB
2.3.0 ?

Thank you,
Erez.

On Mon, Sep 28, 2015 at 6:55 PM, Boris Kolpackov
<boris@codesynthesis.com> wrote:
> Hi Erez,
>
> Erez Pics <picserez@gmail.com> writes:
>
>> But I got an error : 'error: ')' expected at the end of db pragma
>> object', maybe I am missing something,
>
> JOIN type support is only available in ODB 2.4.0. Which version are
> you using?
>
> Boris

From boris at codesynthesis.com  Tue Sep 29 10:38:59 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Tue Sep 29 10:38:57 2015
Subject: [odb-users] odb persistence in a library
In-Reply-To: <1793361386.9709959.1443432199513.JavaMail.open-xchange@patina.store>
References: <1793361386.9709959.1443432199513.JavaMail.open-xchange@patina.store>
Message-ID: <boris.20150929163052@codesynthesis.com>

Hi Christian,

Christian Sell <christian@gsvitec.com> writes:

> My question is whether this is considered a supported use case, or if 
> there are issues we may be running into. Specifically, I am uncertain
> about the implications for schema creation and schema migration (can
> 2 schemas that reside in the same database be migrated independently)

Yes, if you give them different names (--schema-name), they will end
up completely independently versioned. If you take closer look, you
will see that the schema creation/migration API has the optional
scheme name argument throughout. The only restriction is that if you
want to use different versions for different schemas, then the
corresponding  object models should be independent (i.e., you won't
be able to #include headers that have '#pragma db model version' for
one object model into another). In fact, Section 13.1 has this to say
on the subject:

"It is also possible to have multiple object models within the same
application that have different versions. Such models must be independent,
that is, no headers from one model shall include a header from another.
You will also need to assign different schema names to each model with 
the --schema-name ODB compiler option."

Boris

From boris at codesynthesis.com  Tue Sep 29 10:50:29 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Tue Sep 29 10:50:26 2015
Subject: [odb-users] Re: libcutl repo not exported
In-Reply-To: <CADsB1iACL3+cZWrxH6EwV2Gw3Fj998e+EUhY-SJNX3VpScEVRw@mail.gmail.com>
References: <CADsB1iBPhmBRdcHagYER7m01kHm5b=cwdcy9gyAMnGXeeMa44Q@mail.gmail.com>
	<CADsB1iACL3+cZWrxH6EwV2Gw3Fj998e+EUhY-SJNX3VpScEVRw@mail.gmail.com>
Message-ID: <boris.20150929164143@codesynthesis.com>

Hi,

MM <finjulhich@gmail.com> writes:

> fatal: remote error: access denied or repository not exported:
> /odb/libcutl/libcutl.git

I wonder when will you learn to tell what you have actually done
(i.e. the *exact* commands that you ran) that triggered the error?
In fact, enough is enough. From now on, I will be ignoring your
questions that don't include this information.

In any case, both:

git clone git://scm.codesynthesis.com/libcutl/libcutl.git
git clone http://scm.codesynthesis.com/libcutl/libcutl.git

As suggested on:

http://scm.codesynthesis.com/?p=libcutl/libcutl.git;a=summary

Work for me.


> I now follow INSTALL-GIT and just do a make inside the odb directory. It
> seems this file options.hxx is not generated, even
> 
> odb/odb/odb.cxx:38:27: fatal error: odb/options.hxx: No such file or
> directory

When you ran make for the first time, you should have been asked
a bunch of questions, including whether you would like to use the
development build of cli (you should answer 'no') and where cli
is (you should answer '/usr/local/bin/cli'). After that, the
build process should generate options.?xx without problems.

If this still doesn't work, provide *complete* and *precise* log
of what you have done and the output you have observed.

Boris

From christian at gsvitec.com  Tue Sep 29 11:01:18 2015
From: christian at gsvitec.com (Christian Sell)
Date: Tue Sep 29 11:01:21 2015
Subject: [odb-users] odb persistence in a library
In-Reply-To: <boris.20150929163052@codesynthesis.com>
References: <1793361386.9709959.1443432199513.JavaMail.open-xchange@patina.store>
	<boris.20150929163052@codesynthesis.com>
Message-ID: <1050331374.178577.1443538878594.JavaMail.open-xchange@ptangptang.store>

> "It is also possible to have multiple object models within the same
> application that have different versions. Such models must be independent,
> that is, no headers from one model shall include a header from another.
> You will also need to assign different schema names to each model with
> the --schema-name ODB compiler option."
 
ah, thanks, thats what I needed to hear (rsp. read). I always had the plan to
read the manual completely, but haven't gotten that far yet. Right now I am
instead reading the qur'an in order to be prepared for the chaos that's
impending in our country ;)
 
Christian
From finjulhich at gmail.com  Tue Sep 29 16:13:54 2015
From: finjulhich at gmail.com (MM)
Date: Tue Sep 29 16:14:01 2015
Subject: [odb-users] persist non-member containers
In-Reply-To: <boris.20150925160052@codesynthesis.com>
References: <CADsB1iDhggSZJAsAZGT3=Amj+x2Aikp75iO8+OoavU3t3HisTA@mail.gmail.com>
	<boris.20150925160052@codesynthesis.com>
Message-ID: <CADsB1iDq6sdNzvcCsKFMqrJzprCRf39=LzRLTB3oRPDarm2s5g@mail.gmail.com>

On 25 September 2015 at 15:04, Boris Kolpackov <boris@codesynthesis.com>
wrote:

> Hi,
>
> MM <finjulhich@gmail.com> writes:
>
> > What sort of work around is possible if I cannot change my global
> > container?
>
> The same workaround that you've been using for a while now: virtual
> members. With virtual members, the actual data can be anywhere as
> long as you can provide suitable accessors/modifiers. So if you
> create an object with a virtual data member that always returns
> your global container, and you only persist a single instance of
> this object, then that should do the trick.
>
> Boris
>

Thank you. I have implemented successfully this approach for 1 global
container.
Another global container is bigger in number of elements as it contains
history as well.
As the query statement I use loads the single instance of the wrapper
object, which then populates the global container while loading.
Is it possible in this case to use a view to filter on what is loaded for
the global container itself?

Rds,
From boris at codesynthesis.com  Wed Sep 30 10:21:16 2015
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Wed Sep 30 10:21:11 2015
Subject: [odb-users] persist non-member containers
In-Reply-To: <CADsB1iDq6sdNzvcCsKFMqrJzprCRf39=LzRLTB3oRPDarm2s5g@mail.gmail.com>
References: <CADsB1iDhggSZJAsAZGT3=Amj+x2Aikp75iO8+OoavU3t3HisTA@mail.gmail.com>
	<boris.20150925160052@codesynthesis.com>
	<CADsB1iDq6sdNzvcCsKFMqrJzprCRf39=LzRLTB3oRPDarm2s5g@mail.gmail.com>
Message-ID: <boris.20150930161843@codesynthesis.com>

Hi,

MM <finjulhich@gmail.com> writes:

> Is it possible in this case to use a view to filter on what is loaded for
> the global container itself?

Not currently, though there are plans to add support for JOIN'ing
containers onto views. 

In the meantime, perhaps you could make the elements of the container
into full-fledged objects. Then you could control what gets loaded with
a simple query.

Boris

From christian at gsvitec.com  Wed Sep 30 12:44:37 2015
From: christian at gsvitec.com (Christian Sell)
Date: Wed Sep 30 12:44:40 2015
Subject: [odb-users] odb persistence in a library
In-Reply-To: <boris.20150929163052@codesynthesis.com>
References: <1793361386.9709959.1443432199513.JavaMail.open-xchange@patina.store>
	<boris.20150929163052@codesynthesis.com>
Message-ID: <359492180.430846.1443631477487.JavaMail.open-xchange@ptangptang.store>

Hello,
 
still working on the "persistence in library" subject. Here's a problem I
encountered during my first test run:
 
we have an abstract, polymorphic superclass that is mapped inside the library.
Now we want to allow the user of the library to extend the object model by
adding new subclasses, which should also be made persistent AND be accessible
through the same polymorphic relationship. Graphically:
 
Library
    Class1 polymorphic
       id
       name
 
     Class2 extends Class1
 
    Class3
       id
      vector<Class1> poly_container
 
LibraryUser
     Class4 extends Class1
 
 
this immediately fails during odb compilation of the LibraryUser files, because
odb complains that Class4 does not designate an object id. The reason is obvious
- odb doesn't know that Class1 has been mapped polymorphically elsewhere. Can
this situation be solved somehow?

thanks,
Christian