From odb at a-cunningham.com  Wed Aug  1 13:35:07 2018
From: odb at a-cunningham.com (Andrew Cunningham)
Date: Wed Aug  1 13:40:01 2018
Subject: [odb-users] SQlite 3.24.0
Message-ID: <CAGTJuHgT5+0tO2v7sNn+_pvm0EC9HDu34-Ewb9DDBicW_5vPtw@mail.gmail.com>

This performance enhancement might be interesting for ODB developers using
SQLite

UPDATE avoids unnecessary low-level disk writes when the contents of the
database file do not actually change. For example, "UPDATE t1 SET x=25
WHERE y=?" generates no extra disk I/O if the value in column x is already
25. Similarly, when doing UPDATE on records that span multiple pages, only
the subset of pages that actually change are written to disk. This is a
low-level performance optimization only and does not affect the behavior of
TRIGGERs or other higher level SQL structures.

https://www.sqlite.org/releaselog/3_24_0.html
From boris at codesynthesis.com  Thu Aug  2 06:10:51 2018
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Thu Aug  2 06:15:40 2018
Subject: [odb-users] SQlite 3.24.0
In-Reply-To: <CAGTJuHgT5+0tO2v7sNn+_pvm0EC9HDu34-Ewb9DDBicW_5vPtw@mail.gmail.com>
References: <CAGTJuHgT5+0tO2v7sNn+_pvm0EC9HDu34-Ewb9DDBicW_5vPtw@mail.gmail.com>
Message-ID: <boris.20180802120920@codesynthesis.com>

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

> This performance enhancement might be interesting for ODB developers using
> SQLite [...]

Nice, thanks for the heads up, Andrew!

From lists at tinloaf.de  Thu Aug  2 07:59:15 2018
From: lists at tinloaf.de (Lukas Barth)
Date: Thu Aug  2 08:04:15 2018
Subject: [odb-users] Using std::weak_ptr
Message-ID: <871d336d-24c6-f870-a0b5-0287983946ed@tinloaf.de>

[ This question was cross-posted to StackOverflow. If you want to see
the code in a better formatted way and / or grab SO reputation for your
answer, head to
https://stackoverflow.com/questions/51652818/c-odb-database-mapper-unable-to-use-stdweak-ptr-in-relationship
]

Hi,

I'm trying to model a one-to-many relationship with ODB. I'm basically
trying to recreate the example in
https://www.codesynthesis.com/products/odb/doc/manual.xhtml#6.2.2

I have to use std::weak_ptr for one side of the relationship, to avoid
circular ownership issues. However, my very simple example code does not
compile, as ODB doesn't seem to play nicely with std::weak_ptr.

In my example, every Bar has exactly one Foo, while every Foo has
multiple Bars. This is my code:

================ [Begin of main.hpp] ==================

#include <odb/core.hxx>
#include <string>
#include <memory>
#include <vector>

// Forward
class Foo;

#pragma db object
class Bar {
public:
    // A DBConfigKV has exactly *one* DBConfig
    #pragma db not_null
    std::shared_ptr<Foo> cfg;

private:
    #pragma db id auto
    unsigned long id_;
    friend class odb::access;
};

#pragma db object
class Foo {
public:
    // A Foo has multiple Bars
    // Using std::weak_ptr here instead of std::shared_ptr to avoid
    // circular ownership
    #pragma db value_not_null inverse(cfg)
    std::vector<std::weak_ptr<Bar>> entries;

private:
    #pragma db id auto
    unsigned long id_;
    friend class odb::access;
};

int main() {}

================ [End of main.hpp] ==================

I generate the database code with

> odb --std c++11 --database sqlite --generate-query --generate-schema \
> --at-once main.hpp

and compile with

> g++ --std=c++11 main.hpp main-odb.cxx

(I know that would crash at linking - I'm just trying to get it to compile.)

My compiler (GCC 7) tells me:

================ [Begin compiler output] ==================
main-odb.cxx: In static member function ?static void
odb::access::object_traits_impl<Foo,
(odb::database_id)1u>::entries_traits::init(odb::access::object_traits_impl<Foo,
(odb::database_id)1u>::entries_traits::value_type&, const
odb::access::object_traits_impl<Foo,
(odb::database_id)1u>::entries_traits::data_image_type&, odb::database*)?:
main-odb.cxx:794:43: error: no matching function for call to
?std::weak_ptr<Bar>::weak_ptr(odb::object_traits<Bar>::pointer_type)?
             obj_traits::object_type > (id));
                                           ^
In file included from /usr/include/c++/5/memory:82:0,
                 from main.hpp:3,
                 from main-odb.hxx:16,
                 from main-odb.cxx:7:
/usr/include/c++/5/bits/shared_ptr.h:492:2: note: candidate:
template<class _Tp1, class>
std::weak_ptr<_Tp>::weak_ptr(std::weak_ptr<_Tp1>&&)
  weak_ptr(weak_ptr<_Tp1>&& __r) noexcept
  ^
/usr/include/c++/5/bits/shared_ptr.h:492:2: note:   template argument
deduction/substitution failed:
main-odb.cxx:794:43: note:   mismatched types ?std::weak_ptr<_Tp>? and
?odb::object_traits<Bar>::pointer_type {aka Bar*}?
             obj_traits::object_type > (id));
================ [End compiler output] ==================

There are three more candidates which I skipped. The important part: ODB
tries somewhere to create a std::weak_ptr<Bar> from a Bar *, which is
obviously not possible. It would have to create it from a
std::shared_ptr<Bar>. However, the ODB documentation explicitly says one
should (and in fact must) use std::weak_ptr in these cases.

What am I doing wrong?

Thanks a lot,

Lukas

From finjulhich at gmail.com  Thu Aug  2 15:45:20 2018
From: finjulhich at gmail.com (MM)
Date: Thu Aug  2 15:50:26 2018
Subject: [odb-users] multiple base classes
Message-ID: <CA+Jb2D1NXV8QOB2vZd1NihX5UYg_tRAKzm0fvEjTU0mj-BrSgw@mail.gmail.com>

Hello,

I have the following hierarchy:

class ELBase { ... };
class EL { ... };

class Derived : public ELBase, public EL { ... };

#pragma db object(ELBase) polymorphic table("eqlogb") transient definition
#pragma db object(EL) polymorphic table("eqlog") transient definition

#pragma db object(Derived) definition

Is this supported?
In fact EL is a implementation helper, and while I use ELBase pointers and
references, I don't access the objects via EL pointers and references.

Rds,
From boris at codesynthesis.com  Fri Aug  3 10:21:20 2018
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Fri Aug  3 10:26:11 2018
Subject: [odb-users] Using std::weak_ptr
In-Reply-To: <871d336d-24c6-f870-a0b5-0287983946ed@tinloaf.de>
References: <871d336d-24c6-f870-a0b5-0287983946ed@tinloaf.de>
Message-ID: <boris.20180803161801@codesynthesis.com>

Lukas Barth <lists@tinloaf.de> writes:

> main-odb.cxx:794:43: note:   mismatched types ?std::weak_ptr<_Tp>? and
> ?odb::object_traits<Bar>::pointer_type {aka Bar*}?
>              obj_traits::object_type > (id));
> 
> There are three more candidates which I skipped. The important part: ODB
> tries somewhere to create a std::weak_ptr<Bar> from a Bar *, which is
> obviously not possible. It would have to create it from a
> std::shared_ptr<Bar>. However, the ODB documentation explicitly says one
> should (and in fact must) use std::weak_ptr in these cases.
> 
> What am I doing wrong?

You need to make std::shared_ptr the object pointer for your objects.
See Section 3.3, "Object and View Pointers" for details.

Boris

From boris at codesynthesis.com  Fri Aug  3 10:27:41 2018
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Fri Aug  3 10:32:31 2018
Subject: [odb-users] multiple base classes
In-Reply-To: <CA+Jb2D1NXV8QOB2vZd1NihX5UYg_tRAKzm0fvEjTU0mj-BrSgw@mail.gmail.com>
References: <CA+Jb2D1NXV8QOB2vZd1NihX5UYg_tRAKzm0fvEjTU0mj-BrSgw@mail.gmail.com>
Message-ID: <boris.20180803162502@codesynthesis.com>

MM <finjulhich@gmail.com> writes:

> class ELBase { ... };
> class EL { ... };
> 
> class Derived : public ELBase, public EL { ... };
> 
> #pragma db object(ELBase) polymorphic table("eqlogb") transient definition
> #pragma db object(EL) polymorphic table("eqlog") transient definition
> 
> #pragma db object(Derived) definition
> 
> Is this supported?

Section 8.2, "Polymorphism Inheritance", fourth paragraph:

"Note that multiple polymorphism inheritance or mixing polymorphism 
and reuse inheritance is not supported."


> In fact EL is a implementation helper, and while I use ELBase pointers and
> references, I don't access the objects via EL pointers and references.

Then you cat try to make EL a transient base and persist its data members
as if they were in Derived (using virtual data member).

From finjulhich at gmail.com  Fri Aug  3 14:21:42 2018
From: finjulhich at gmail.com (MM)
Date: Fri Aug  3 14:26:52 2018
Subject: [odb-users] internal compiler error:
	cult:compiler::context::no_entry
Message-ID: <CA+Jb2D1av3BJgCgBtk0Ch7MZXE10WKrQzqWVuHxurhzXGuwp0w@mail.gmail.com>

For the purposes of preparing a small self-sufficient test case to diagnose
this issue, with 2.5.0-b3 taken and compiled from
https://codesynthesis.com/~boris/tmp/odb/pre-release/, I have this question.

Currently, I call odb compiler like so:

odb -d sqlite --sqlite-override-null --std c++14 --profile boost
--omit-drop --generate-query --generate-schema-only --schema-format sql
--at-once -I/src -I/usr/include --changelog-dir /src/odb --options-file
/src/vendors/odb/boost-multiarray.options --input-name mydbname
/src/file_1.hpp ... /src/file_n.hpp

terminate called after throwing an instance of
'cutl::compiler::context::no_entry'

/src/file_M.hpp:20:12: internal compiler error: Aborted
 #pragma db member(prtfelt::strat) virtual(blitzq::nstring) \

This is just to generate the schema, not c++ is generated.

Does the odb compiler use each header file argument separately? or does it
concatenate all the headers to produce the SQL.

This can help me understand how to focus on the smallest test case to
report.

Thanks,
From boris at codesynthesis.com  Mon Aug  6 08:13:14 2018
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Aug  6 08:18:17 2018
Subject: [odb-users] internal compiler error:
	cult:compiler::context::no_entry
In-Reply-To: <CA+Jb2D1av3BJgCgBtk0Ch7MZXE10WKrQzqWVuHxurhzXGuwp0w@mail.gmail.com>
References: <CA+Jb2D1av3BJgCgBtk0Ch7MZXE10WKrQzqWVuHxurhzXGuwp0w@mail.gmail.com>
Message-ID: <boris.20180806141139@codesynthesis.com>

MM <finjulhich@gmail.com> writes:

> odb -d sqlite --sqlite-override-null --std c++14 --profile boost
> --omit-drop --generate-query --generate-schema-only --schema-format sql
> --at-once -I/src -I/usr/include --changelog-dir /src/odb --options-file
> /src/vendors/odb/boost-multiarray.options --input-name mydbname
> /src/file_1.hpp ... /src/file_n.hpp
> 
> Does the odb compiler use each header file argument separately? or does it
> concatenate all the headers to produce the SQL.

Since you have specified the --at-once option, they are effectively
concatenated in the order specified on the command line.

From adanesh at noornet.net  Sat Aug 25 03:49:41 2018
From: adanesh at noornet.net (=?utf-8?B?2LnZhNuMINiv2KfZhti0?=)
Date: Sat Aug 25 03:56:14 2018
Subject: [odb-users] Is C++17 supported?
Message-ID: <b5dbc107c9f64024aa40545caeb85a03@noornet.net>

Hi Boris!
According to this link https://www.codesynthesis.com/pipermail/odb-users/2018-March/003973.html, c++17 must be supported; But:

C:\odb-2.5.0-b.3-i686-windows\bin>odb --std c++17
invalid value 'c++17' for option '--std'

And when I include ?optional? in a header file, it generates an error:
fatal error: optional: No such file or directory

Best regards,
Ali
From boris at codesynthesis.com  Mon Aug 27 09:13:43 2018
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Aug 27 09:19:53 2018
Subject: [odb-users] Is C++17 supported?
In-Reply-To: <b5dbc107c9f64024aa40545caeb85a03@noornet.net>
References: <b5dbc107c9f64024aa40545caeb85a03@noornet.net>
Message-ID: <boris.20180827151035@codesynthesis.com>

adanesh@noornet.net writes:

> C:\odb-2.5.0-b.3-i686-windows\bin>odb --std c++17
> invalid value 'c++17' for option '--std'

2.5.0-b.3 is too old. Try building the latest snapshot using these
instructions:

https://codesynthesis.com/products/odb/doc/install-build2.xhtml

This will give you both --std c++17 and a recent enough GCC
(for the ODB compiler) to be able to use it.

From thomas.baudier at creatis.insa-lyon.fr  Tue Aug 28 10:44:07 2018
From: thomas.baudier at creatis.insa-lyon.fr (Thomas BAUDIER)
Date: Tue Aug 28 10:50:29 2018
Subject: [odb-users] No gcc-plugin support
Message-ID: <9a173c5ebec6c332774cd0747b112e8f@pop3.creatis.insa-lyon.fr>

Hi,

I wanted to install odb using build2 
(https://codesynthesis.com/products/odb/doc/install-build2.xhtml#linux). 
I use OpenSuse LEAP 15 but the gcc-plugin-devel is not available for 
this platform. How can install latest odb version without gcc-plugin 
support?

Thank you for your help
Thomas

From boris at codesynthesis.com  Tue Aug 28 14:36:22 2018
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Tue Aug 28 14:42:35 2018
Subject: [odb-users] No gcc-plugin support
In-Reply-To: <9a173c5ebec6c332774cd0747b112e8f@pop3.creatis.insa-lyon.fr>
References: <9a173c5ebec6c332774cd0747b112e8f@pop3.creatis.insa-lyon.fr>
Message-ID: <boris.20180828203139@codesynthesis.com>

Thomas BAUDIER <thomas.baudier@creatis.insa-lyon.fr> writes:

> I wanted to install odb using build2
> (https://codesynthesis.com/products/odb/doc/install-build2.xhtml#linux). I
> use OpenSuse LEAP 15 but the gcc-plugin-devel is not available for this
> platform.

Yes, last time I've heard SUSE was disabling support for plugins in
their build of GCC.


> How can install latest odb version without gcc-plugin support?

You can't. The ODB compiler relies on GCC plugin support for parsing
C++ code. Going forward, your options seem to be (1) to check if maybe
SUSE (or someone else) provides a build of GCC with plugins enabled or
(2) build GCC from source yourself (it's actually not that hard).

Boris

From finjulhich at gmail.com  Wed Aug 29 15:56:40 2018
From: finjulhich at gmail.com (MM)
Date: Wed Aug 29 16:03:13 2018
Subject: [odb-users] internal compiler error:
	cult:compiler::context::no_entry
Message-ID: <CA+Jb2D2KcgnkERc+sw0OaaCjHJ3_x9tp5gbGw_cM6zcukorwfQ@mail.gmail.com>

On Mon, 6 Aug 2018 at 13:13, Boris Kolpackov <boris@codesynthesis.com>
wrote:

> MM <finjulhich@gmail.com> writes:
>
> > odb -d sqlite --sqlite-override-null --std c++14 --profile boost
> > --omit-drop --generate-query --generate-schema-only --schema-format sql
> > --at-once -I/src -I/usr/include --changelog-dir /src/odb --options-file
> > /src/vendors/odb/boost-multiarray.options --input-name mydbname
> > /src/file_1.hpp ... /src/file_n.hpp
> >
> > Does the odb compiler use each header file argument separately? or does
> it
> > concatenate all the headers to produce the SQL.
>
> Since you have specified the --at-once option, they are effectively
> concatenated in the order specified on the command line.


Still to try to reduce to a reportable example, 1 more question:

In this resulting concatenated stream of c++ code, what happens to #include
directives? They are processed as usual?

MM
From boris at codesynthesis.com  Thu Aug 30 10:16:56 2018
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Thu Aug 30 10:23:15 2018
Subject: [odb-users] internal compiler error:
	cult:compiler::context::no_entry
In-Reply-To: <CA+Jb2D2KcgnkERc+sw0OaaCjHJ3_x9tp5gbGw_cM6zcukorwfQ@mail.gmail.com>
References: <CA+Jb2D2KcgnkERc+sw0OaaCjHJ3_x9tp5gbGw_cM6zcukorwfQ@mail.gmail.com>
Message-ID: <boris.20180830161241@codesynthesis.com>

MM <finjulhich@gmail.com> writes:

> On Mon, 6 Aug 2018 at 13:13, Boris Kolpackov <boris@codesynthesis.com> wrote:
> 
> > MM <finjulhich@gmail.com> writes:
> >
> > > odb -d sqlite --sqlite-override-null --std c++14 --profile boost
> > > --omit-drop --generate-query --generate-schema-only --schema-format sql
> > > --at-once -I/src -I/usr/include --changelog-dir /src/odb --options-file
> > > /src/vendors/odb/boost-multiarray.options --input-name mydbname
> > > /src/file_1.hpp ... /src/file_n.hpp
> > >
> > > Does the odb compiler use each header file argument separately? or does
> > > it concatenate all the headers to produce the SQL.
> >
> > Since you have specified the --at-once option, they are effectively
> > concatenated in the order specified on the command line.
> 
> Still to try to reduce to a reportable example, 1 more question:
> 
> In this resulting concatenated stream of c++ code, what happens to #include
> directives? They are processed as usual?

Yes, the files are just concatenated together and compiled as a single
translation unit. You can see the relevant code in odb/odb.cxx starting
at line 801:

https://git.codesynthesis.com/cgit/odb/odb/tree/odb/odb.cxx