From marco.craveiro at gmail.com  Wed Mar  1 06:30:36 2017
From: marco.craveiro at gmail.com (Marco Craveiro)
Date: Wed Mar  1 06:30:51 2017
Subject: [odb-users] BLOBs and multi-database
In-Reply-To: <boris.20170227140051@codesynthesis.com>
References: <CAKRNd4yVUQSzn4PgHVdkiB+0mx0SOF5HxrHTeDJR1FP-=Vs5eg@mail.gmail.com>
	<boris.20170227140051@codesynthesis.com>
Message-ID: <CAKRNd4xDwoXtvv0_bCH_b-dG6i9wnrjx=Y1Jq03Aut3L58u3CA@mail.gmail.com>

Hi Boris,

<blob for multiple databases>
> You are probably looking for the database pragma prefix (introduction
> to Chapter 16, "Multi-Database Support"):
>
> #pragma db pgsql:type("BYTEA") oracle:type("BLOB")
> std::vector<char> data;

Yep that was exactly it, thanks very much.

Cheers
-- 
Marco Craveiro

MD, Domain Driven Consulting
about: http://about.me/marcocraveiro
blog: http://mcraveiro.blogspot.co.uk
twitter: https://twitter.com/MarcoCraveiro

That the Ideas are themselves manifestations (of the Idea-Idea) and
that the Idea-Idea is a-kind-of Manifestation-Idea?which is a-kind-of
itself, so that the system is completely self-describing? would have
been appreciated by Plato as an extremely practical joke [Plato]. --
Alan Key

From boris at codesynthesis.com  Wed Mar  1 11:12:30 2017
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Wed Mar  1 11:12:39 2017
Subject: [odb-users] Inverse and composite Id
In-Reply-To: <CAHCKNJiT6aa8JH0udvO=8nRR58y1u5bnBfyHCJkfeX7YDDCE6A@mail.gmail.com>
References: <CAHCKNJiT6aa8JH0udvO=8nRR58y1u5bnBfyHCJkfeX7YDDCE6A@mail.gmail.com>
Message-ID: <boris.20170301180901@codesynthesis.com>

Hi Sean,

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

> class A {
> 
> #pragma db value
>     struct Id {
>         int a;
>         int b;
> }
> 
> #pragma db id
>     Id m_id;
> 
> #pragma db value_not_null
>     std::shared_ptr<B> m_b;
> 
> };
> 
> 
> class B {
>     using id_t = std::uint32_t;
> 
> #pragma db id
>     id_t m_id;
> 
> #pragma db inverse(m_b)
>     std::vector<std::shared_ptr<A>> m_as;
> };
> 
> /<path>/odb_gen/A-odb.hxx:385:70: error: ?image_type? in ?class
> odb::access::composite_value_traits<B::Id, (odb::database_id)2u>? does not
> name a type

The error suggests that you have it the other way around (Id is in B, not
A, and A points to B). Assuming this is the case, my guess would be that
the header that contains A doesn't include the one containing B. See
Section 6.3, "Circular Relationships" in the manual for details on the
issue and possible solutions.

Boris

From boris at codesynthesis.com  Wed Mar  1 11:22:22 2017
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Wed Mar  1 11:22:30 2017
Subject: [odb-users] Using prefetch with Oracle
In-Reply-To: <CAKRNd4xW6SPCpD9tZ-Pbw236B8dK4D3yMfd3pzKZdNHBdRQNzw@mail.gmail.com>
References: <CAKRNd4x7K-WPp7C2xTtCfQ0rD4ou-4FYaxpEFeOU1vrHeWAbTA@mail.gmail.com>
	<boris.20170210180525@codesynthesis.com>
	<CAKRNd4z7h7A9Dj5UHhbofcJncCeskowGcFgsN-+hqwnF8AA62Q@mail.gmail.com>
	<20170213135652.GA22900@codesynthesis.com>
	<CAKRNd4yoBmOsTe6D_ZRm1yAYz-9g-YCk75hunfP0OnW1JzMjMA@mail.gmail.com>
	<CAKRNd4xW6SPCpD9tZ-Pbw236B8dK4D3yMfd3pzKZdNHBdRQNzw@mail.gmail.com>
Message-ID: <boris.20170301182048@codesynthesis.com>

Hi Marco,

Marco Craveiro <marco.craveiro@gmail.com> writes:

> On the plus side I am cataloguing the adventure under a series of blog
> posts - [1], [2] so far - which may (or may not) help other Oracle /
> ODB users.

Good stuff. I need to setup an Oracle 11 VM soon. Now I know which guide
I will use ;-).

Boris

From sean.clarke at sec-consulting.co.uk  Wed Mar  1 15:41:49 2017
From: sean.clarke at sec-consulting.co.uk (Sean Clarke)
Date: Wed Mar  1 15:42:01 2017
Subject: [odb-users] Inverse and composite Id
In-Reply-To: <boris.20170301180901@codesynthesis.com>
References: <CAHCKNJiT6aa8JH0udvO=8nRR58y1u5bnBfyHCJkfeX7YDDCE6A@mail.gmail.com>
	<boris.20170301180901@codesynthesis.com>
Message-ID: <CAHCKNJhUBGGZGT=NQPrmLnwcOuHBeyf8hrKEJ+dOUPsHyr8pBg@mail.gmail.com>

On 1 March 2017 at 16:12, Boris Kolpackov <boris@codesynthesis.com> wrote:

> The error suggests that you have it the other way around (Id is in B, not
> A, and A points to B). Assuming this is the case, my guess would be that
> the header that contains A doesn't include the one containing B. See
> Section 6.3, "Circular Relationships" in the manual for details on the
> issue and possible solutions.
>

?Hmm... OK, thanks Boris.

I have double checked (obviously because my classes are not really called A
or B)? and the case above is a reflection of what I am doing, but I did
report the error in the wrong file (sorry):

> /<path>/odb_gen/B-odb.hxx:385:70: error: ?image_type? in ?class
odb::access::composite_value_traits<A::Id, (odb::database_id)2u>? does not
name a type composite_value_traits< ::A::Id, id_pgsql >::image_type
value_value;

Thanks for the info on circular relationships, I have followed your
instructions (although I was achieving the same result before) and I still
get the same problem.

Will keep on playing and investigating....


Regards
Sean Clarke
From anhu.com at gmail.com  Thu Mar  2 03:53:46 2017
From: anhu.com at gmail.com (Hu An)
Date: Thu Mar  2 05:28:17 2017
Subject: [odb-users] Question about supported compilers (Visual Studio 2015)
Message-ID: <CADEhQo5EYOrtOxoiTtCcHrPF11yOG5kUE=pDOXCsG+hXnZ9o7w@mail.gmail.com>

Hi!
The page "Supported Platforms and C++ Compilers" (
http://www.codesynthesis.com/products/odb/platforms.xhtml) does not list
Visual Studio 2015. Can I still use ODB technology, if our solution is
entirely a VS 2015 solution or are there problems to expect?
Thanks for your answer,
A. Hu
From boris at codesynthesis.com  Thu Mar  2 08:38:55 2017
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Thu Mar  2 08:39:04 2017
Subject: [odb-users] Inverse and composite Id
In-Reply-To: <CAHCKNJhUBGGZGT=NQPrmLnwcOuHBeyf8hrKEJ+dOUPsHyr8pBg@mail.gmail.com>
References: <CAHCKNJiT6aa8JH0udvO=8nRR58y1u5bnBfyHCJkfeX7YDDCE6A@mail.gmail.com>
	<boris.20170301180901@codesynthesis.com>
	<CAHCKNJhUBGGZGT=NQPrmLnwcOuHBeyf8hrKEJ+dOUPsHyr8pBg@mail.gmail.com>
Message-ID: <boris.20170302153749@codesynthesis.com>

Hi Sean,

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

> Thanks for the info on circular relationships, I have followed your
> instructions (although I was achieving the same result before) and I still
> get the same problem.
> 
> Will keep on playing and investigating....

If you can send a small but complete test case (i.e., the two headers and
how you compiled them), I can take a look.

Boris

From boris at codesynthesis.com  Thu Mar  2 08:41:56 2017
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Thu Mar  2 08:42:05 2017
Subject: [odb-users] Question about supported compilers (Visual Studio
	2015)
In-Reply-To: <CADEhQo5EYOrtOxoiTtCcHrPF11yOG5kUE=pDOXCsG+hXnZ9o7w@mail.gmail.com>
References: <CADEhQo5EYOrtOxoiTtCcHrPF11yOG5kUE=pDOXCsG+hXnZ9o7w@mail.gmail.com>
Message-ID: <boris.20170302153909@codesynthesis.com>

Hi,

Hu An <anhu.com@gmail.com> writes:

> The page "Supported Platforms and C++ Compilers" does not list Visual
> Studio 2015. Can I still use ODB technology, if our solution is
> entirely a VS 2015 solution or are there problems to expect?

No, it should work. I believe quite a few people are using ODB with
VS 2015 by simply converting the solution for the previous version.

We will also add official support for VS 2015 (and 2017) in the
next release, which should be out "real soon now" (famous last
words, I know).

Boris

From sean.clarke at sec-consulting.co.uk  Thu Mar  2 17:28:52 2017
From: sean.clarke at sec-consulting.co.uk (Sean Clarke)
Date: Thu Mar  2 17:29:06 2017
Subject: [odb-users] Inverse and composite Id
In-Reply-To: <boris.20170301180901@codesynthesis.com>
References: <CAHCKNJiT6aa8JH0udvO=8nRR58y1u5bnBfyHCJkfeX7YDDCE6A@mail.gmail.com>
	<boris.20170301180901@codesynthesis.com>
Message-ID: <CAHCKNJgnB1x1k34ho_6_H8z+RgSk=HT+O8CFENszc52OzuzTnw@mail.gmail.com>

Still investigating.... it definitely is something to do with the:

#pragma db value
    struct Id {
        int a;
        int b;
}

if I comment that out and add:

using Id = std::uint32_t;

and amend the operator< appropriately it seems to work.



Regards
Sean Clarke

On 1 March 2017 at 16:12, Boris Kolpackov <boris@codesynthesis.com> wrote:

> Hi Sean,
>
> Sean Clarke <sean.clarke@sec-consulting.co.uk> writes:
>
> > class A {
> >
> > #pragma db value
> >     struct Id {
> >         int a;
> >         int b;
> > }
> >
> > #pragma db id
> >     Id m_id;
> >
> > #pragma db value_not_null
> >     std::shared_ptr<B> m_b;
> >
> > };
> >
> >
> > class B {
> >     using id_t = std::uint32_t;
> >
> > #pragma db id
> >     id_t m_id;
> >
> > #pragma db inverse(m_b)
> >     std::vector<std::shared_ptr<A>> m_as;
> > };
> >
> > /<path>/odb_gen/A-odb.hxx:385:70: error: ?image_type? in ?class
> > odb::access::composite_value_traits<B::Id, (odb::database_id)2u>? does
> not
> > name a type
>
> The error suggests that you have it the other way around (Id is in B, not
> A, and A points to B). Assuming this is the case, my guess would be that
> the header that contains A doesn't include the one containing B. See
> Section 6.3, "Circular Relationships" in the manual for details on the
> issue and possible solutions.
>
> Boris
>
From marco.craveiro at gmail.com  Fri Mar  3 03:22:53 2017
From: marco.craveiro at gmail.com (Marco Craveiro)
Date: Fri Mar  3 03:23:06 2017
Subject: [odb-users] Question about supported compilers (Visual Studio
	2015)
In-Reply-To: <boris.20170302153909@codesynthesis.com>
References: <CADEhQo5EYOrtOxoiTtCcHrPF11yOG5kUE=pDOXCsG+hXnZ9o7w@mail.gmail.com>
	<boris.20170302153909@codesynthesis.com>
Message-ID: <CAKRNd4yDhJ_WJx+ezsBWMM5eS06kVJACv-r-PN4v5LSP-a7YBA@mail.gmail.com>

Hi,

>> The page "Supported Platforms and C++ Compilers" does not list Visual
>> Studio 2015. Can I still use ODB technology, if our solution is
>> entirely a VS 2015 solution or are there problems to expect?
>
> No, it should work. I believe quite a few people are using ODB with
> VS 2015 by simply converting the solution for the previous version.
>
> We will also add official support for VS 2015 (and 2017) in the
> next release, which should be out "real soon now" (famous last
> words, I know).

For what it's worth, I used ODB against 2015 with Oracle and I did
exactly what Boris suggests there and it worked fine.

Cheers
-- 
Marco Craveiro

MD, Domain Driven Consulting
about: http://about.me/marcocraveiro
blog: http://mcraveiro.blogspot.co.uk
twitter: https://twitter.com/MarcoCraveiro

That the Ideas are themselves manifestations (of the Idea-Idea) and
that the Idea-Idea is a-kind-of Manifestation-Idea?which is a-kind-of
itself, so that the system is completely self-describing? would have
been appreciated by Plato as an extremely practical joke [Plato]. --
Alan Key

From c.sell at byterefinery.de  Sun Mar  5 16:37:42 2017
From: c.sell at byterefinery.de (c.sell@byterefinery.de)
Date: Sun Mar  5 16:37:51 2017
Subject: [odb-users] collection mapping error
Message-ID: <20170305213742.Horde.v_xEXGL6PFPkmOrc7zNsAdh@webmail.byterefinery.de>


Hello all,

I have a simple std::vector<QImage> collection mapping which is  
controlled by a section. I also have traits implementations for the  
QImage type. The database table generated from the mappings has the  
following columns:

object_id
index
value

with object_id referring back to the main object table.

Now, when I try to update the container, I get an exception stating  
that the object was already persistent. This happens during the first  
insert into the empty collection table, so it cannot be true. Further  
investigation shows that the error is in fact a constraint error,  
caused by the fact that no value is bound to the object_id column.  
Stepping to the ODB generated code shows that indeed values for  
'index' and 'value' are supplied, but not for object_id

What can I do?

regards,
Christian

From boris at codesynthesis.com  Mon Mar  6 03:44:05 2017
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Mar  6 03:44:14 2017
Subject: [odb-users] collection mapping error
In-Reply-To: <20170305213742.Horde.v_xEXGL6PFPkmOrc7zNsAdh@webmail.byterefinery.de>
References: <20170305213742.Horde.v_xEXGL6PFPkmOrc7zNsAdh@webmail.byterefinery.de>
Message-ID: <boris.20170306104246@codesynthesis.com>

Hi Christian,

c.sell@byterefinery.de <c.sell@byterefinery.de> writes:

> I have a simple std::vector<QImage> collection mapping which is controlled
> by a section.

We had a few bugs fixed in the section implementation. Does everything work
if you get rid of the section?

Boris

From c.sell at byterefinery.de  Mon Mar  6 04:01:18 2017
From: c.sell at byterefinery.de (c.sell@byterefinery.de)
Date: Mon Mar  6 04:01:27 2017
Subject: [odb-users] collection mapping error
In-Reply-To: <boris.20170306104246@codesynthesis.com>
References: <20170305213742.Horde.v_xEXGL6PFPkmOrc7zNsAdh@webmail.byterefinery.de>
	<boris.20170306104246@codesynthesis.com>
Message-ID: <20170306090118.Horde.oyO72rcLugysh9uAzZ2fNhp@webmail.byterefinery.de>

Hello Boris,

unfortunately no. I just removed the section and the issue is exactly  
the same. The only remaining unusual thing about the mapping I can  
detect is the fact that the collection elements are of a type that ODB  
doesn't directly support. Is there something I could have done wrong  
in my mapping traits implementation which could cause this?

thanks,
Christian


Zitat von Boris Kolpackov <boris@codesynthesis.com>:

> Hi Christian,
>
> c.sell@byterefinery.de <c.sell@byterefinery.de> writes:
>
>> I have a simple std::vector<QImage> collection mapping which is controlled
>> by a section.
>
> We had a few bugs fixed in the section implementation. Does everything work
> if you get rid of the section?
>
> Boris



From c.sell at byterefinery.de  Mon Mar  6 04:05:36 2017
From: c.sell at byterefinery.de (c.sell@byterefinery.de)
Date: Mon Mar  6 04:05:44 2017
Subject: [odb-users] collection mapping error
In-Reply-To: <boris.20170306104246@codesynthesis.com>
References: <20170305213742.Horde.v_xEXGL6PFPkmOrc7zNsAdh@webmail.byterefinery.de>
	<boris.20170306104246@codesynthesis.com>
Message-ID: <20170306090536.Horde.dJszFfqXunR7xI087S4GkTb@webmail.byterefinery.de>

below is the generated code that is executed immediately before the  
sql statement fails, and of which which I assume that is provides the  
parameter values. object_id is suspiciously missing:


   void access::object_traits_impl< ::rentalapp::Damage, id_sqlite  
 >::odb_pictures_traits::
   init (data_image_type& i,
         index_type* j,
         const value_type& v)
   {
     using namespace sqlite;

     statement_kind sk (statement_insert);
     ODB_POTENTIALLY_UNUSED (sk);

     bool grew (false);

     // index
     //
     if (j != 0)
     {
       bool is_null (false);
       sqlite::value_traits<
           index_type,
           sqlite::id_integer >::set_image (
         i.index_value,
         is_null,
         *j);
       i.index_null = is_null;
     }

     // value
     //
     {
       bool is_null (false);
       std::size_t cap (i.value_value.capacity ());
       sqlite::value_traits<
           value_type,
           sqlite::id_blob >::set_image (
         i.value_value,
         i.value_size,
         is_null,
         v);
       i.value_null = is_null;
       grew = grew || (cap != i.value_value.capacity ());
     }

     if (grew)
       i.version++;
   }


Zitat von Boris Kolpackov <boris@codesynthesis.com>:

> Hi Christian,
>
> c.sell@byterefinery.de <c.sell@byterefinery.de> writes:
>
>> I have a simple std::vector<QImage> collection mapping which is controlled
>> by a section.
>
> We had a few bugs fixed in the section implementation. Does everything work
> if you get rid of the section?
>
> Boris



From c.sell at byterefinery.de  Mon Mar  6 04:12:49 2017
From: c.sell at byterefinery.de (c.sell@byterefinery.de)
Date: Mon Mar  6 04:12:56 2017
Subject: [odb-users] collection mapping error
In-Reply-To: <boris.20170306104246@codesynthesis.com>
References: <20170305213742.Horde.v_xEXGL6PFPkmOrc7zNsAdh@webmail.byterefinery.de>
	<boris.20170306104246@codesynthesis.com>
Message-ID: <20170306091249.Horde.qFcvEIuC_nqXG6F6HuRljtr@webmail.byterefinery.de>

also, I am on 2.5.0-a11 using sqlite as backend. I had to go to 2.5  
due to another bug I ran into

btw, I will be migrating my app to windows soon, and it just dawns on  
me that I might have issues getting a suitable precompiled version  
(don't even remember where I got the 2.5.0-a11 from). As far as I can  
tell, the website has nothing. Any help?

thanks,
Christian


Zitat von Boris Kolpackov <boris@codesynthesis.com>:

> Hi Christian,
>
> c.sell@byterefinery.de <c.sell@byterefinery.de> writes:
>
>> I have a simple std::vector<QImage> collection mapping which is controlled
>> by a section.
>
> We had a few bugs fixed in the section implementation. Does everything work
> if you get rid of the section?
>
> Boris



From boris at codesynthesis.com  Mon Mar  6 05:33:47 2017
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Mar  6 05:33:56 2017
Subject: [odb-users] collection mapping error
In-Reply-To: <20170306091249.Horde.qFcvEIuC_nqXG6F6HuRljtr@webmail.byterefinery.de>
References: <20170305213742.Horde.v_xEXGL6PFPkmOrc7zNsAdh@webmail.byterefinery.de>
	<boris.20170306104246@codesynthesis.com>
	<20170306091249.Horde.qFcvEIuC_nqXG6F6HuRljtr@webmail.byterefinery.de>
Message-ID: <boris.20170306122530@codesynthesis.com>

Hi Christian,

1. Can you create a test case that reproduces this problem? Preferably
   with std::vector<char> used as a blob value.

2. Once we solve this, I can build you pre-release binary for Windows.

Boris

From c.sell at byterefinery.de  Mon Mar  6 07:57:34 2017
From: c.sell at byterefinery.de (c.sell@byterefinery.de)
Date: Mon Mar  6 07:57:43 2017
Subject: [odb-users] collection mapping error
In-Reply-To: <boris.20170306122530@codesynthesis.com>
References: <20170305213742.Horde.v_xEXGL6PFPkmOrc7zNsAdh@webmail.byterefinery.de>
	<boris.20170306104246@codesynthesis.com>
	<20170306091249.Horde.qFcvEIuC_nqXG6F6HuRljtr@webmail.byterefinery.de>
	<boris.20170306122530@codesynthesis.com>
Message-ID: <20170306125734.Horde.UoTXGi-GOSjxfSHX7c6-nWt@webmail.byterefinery.de>

Hello Boris,

I created a testcase using this mapping:

typedef std::vector<char> buffer;
#pragma db value(buffer) type("BLOB")

#pragma db object
struct Damage {
   #pragma db id auto
   long id_;

   std::string name;

   std::vector<buffer> pictures;
};


but it did not reproduce ther error. However, I have been able to  
track the issue down a little further. It appears that the binding  
code is not executed because in function insert(), the  
data_binding_test_version () check fails:

     if (sts.data_binding_test_version ())
     {
       const binding& id (sts.id_binding ());
       bind (sts.data_bind (), id.bind, id.count, di);
       sts.data_binding_update_version ();
     }


in my test case program, the if branch is entered once for the  
collection object. With my target app, it isn't ever

HTH to help me
Chris

From boris at codesynthesis.com  Mon Mar  6 08:26:28 2017
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Mar  6 08:26:37 2017
Subject: [odb-users] collection mapping error
In-Reply-To: <20170306125734.Horde.UoTXGi-GOSjxfSHX7c6-nWt@webmail.byterefinery.de>
References: <20170305213742.Horde.v_xEXGL6PFPkmOrc7zNsAdh@webmail.byterefinery.de>
	<boris.20170306104246@codesynthesis.com>
	<20170306091249.Horde.qFcvEIuC_nqXG6F6HuRljtr@webmail.byterefinery.de>
	<boris.20170306122530@codesynthesis.com>
	<20170306125734.Horde.UoTXGi-GOSjxfSHX7c6-nWt@webmail.byterefinery.de>
Message-ID: <boris.20170306152300@codesynthesis.com>

Hi Christian,

c.sell@byterefinery.de <c.sell@byterefinery.de> writes:

> It appears that the binding code is not executed because in function
> insert(), the data_binding_test_version () check fails:
> 
>     if (sts.data_binding_test_version ())
>     {
>       const binding& id (sts.id_binding ());
>       bind (sts.data_bind (), id.bind, id.count, di);
>       sts.data_binding_update_version ();
>     }
> 
> 
> in my test case program, the if branch is entered once for the collection
> object. With my target app, it isn't ever

Yeah, that's fishy. Smells like memory corruption. You can try two things:

1. Try to plug your QImage implementation into the test case. The idea is
   basically to try to "bring" the test case a step at a time closer to
   your application and see what triggers the change.

2. Try to run something like valgrind on your application (or the test
   case if you manage to make it fail).

Boris

From c.sell at byterefinery.de  Mon Mar  6 09:58:19 2017
From: c.sell at byterefinery.de (c.sell@byterefinery.de)
Date: Mon Mar  6 09:58:27 2017
Subject: [odb-users] collection mapping error
In-Reply-To: <boris.20170306152300@codesynthesis.com>
References: <20170305213742.Horde.v_xEXGL6PFPkmOrc7zNsAdh@webmail.byterefinery.de>
	<boris.20170306104246@codesynthesis.com>
	<20170306091249.Horde.qFcvEIuC_nqXG6F6HuRljtr@webmail.byterefinery.de>
	<boris.20170306122530@codesynthesis.com>
	<20170306125734.Horde.UoTXGi-GOSjxfSHX7c6-nWt@webmail.byterefinery.de>
	<boris.20170306152300@codesynthesis.com>
Message-ID: <20170306145819.Horde.q6qu9ZfmjHz0haIrGymowyB@webmail.byterefinery.de>

I don't think this is a memory overrun. All values are normal (0 or  
1), all pointers are valid

Zitat von Boris Kolpackov <boris@codesynthesis.com>:

> Hi Christian,
>
> c.sell@byterefinery.de <c.sell@byterefinery.de> writes:
>
>> It appears that the binding code is not executed because in function
>> insert(), the data_binding_test_version () check fails:
>>
>>     if (sts.data_binding_test_version ())
>>     {
>>       const binding& id (sts.id_binding ());
>>       bind (sts.data_bind (), id.bind, id.count, di);
>>       sts.data_binding_update_version ();
>>     }
>>
>>
>> in my test case program, the if branch is entered once for the collection
>> object. With my target app, it isn't ever
>
> Yeah, that's fishy. Smells like memory corruption. You can try two things:
>
> 1. Try to plug your QImage implementation into the test case. The idea is
>    basically to try to "bring" the test case a step at a time closer to
>    your application and see what triggers the change.
>
> 2. Try to run something like valgrind on your application (or the test
>    case if you manage to make it fail).
>
> Boris



From c.sell at byterefinery.de  Mon Mar  6 10:19:43 2017
From: c.sell at byterefinery.de (c.sell@byterefinery.de)
Date: Mon Mar  6 10:19:51 2017
Subject: [odb-users] collection mapping error
In-Reply-To: <boris.20170306152300@codesynthesis.com>
References: <20170305213742.Horde.v_xEXGL6PFPkmOrc7zNsAdh@webmail.byterefinery.de>
	<boris.20170306104246@codesynthesis.com>
	<20170306091249.Horde.qFcvEIuC_nqXG6F6HuRljtr@webmail.byterefinery.de>
	<boris.20170306122530@codesynthesis.com>
	<20170306125734.Horde.UoTXGi-GOSjxfSHX7c6-nWt@webmail.byterefinery.de>
	<boris.20170306152300@codesynthesis.com>
Message-ID: <20170306151943.Horde.mdgtE3Y49cHVKvwsoLKjVCl@webmail.byterefinery.de>

Hi Boris,

it seems my assumptions were wrong. If I adapt the example to the  
exact same sequence of operations as the productive code (first a  
select, then an insert), it shows the exact same state at the  
respective points. bind() seems to be only executed once even across  
statements, and the data_binding_test_version() check always fails  
afterwards (in both my testcases, that is)

This leads me to the assumption that the issue lies not in binding,  
but rather in writing to the once bound memory area. I am back at  
point 0. Appreciate any guidance

Christian

Zitat von Boris Kolpackov <boris@codesynthesis.com>:

> Hi Christian,
>
> c.sell@byterefinery.de <c.sell@byterefinery.de> writes:
>
>> It appears that the binding code is not executed because in function
>> insert(), the data_binding_test_version () check fails:
>>
>>     if (sts.data_binding_test_version ())
>>     {
>>       const binding& id (sts.id_binding ());
>>       bind (sts.data_bind (), id.bind, id.count, di);
>>       sts.data_binding_update_version ();
>>     }
>>
>>
>> in my test case program, the if branch is entered once for the collection
>> object. With my target app, it isn't ever
>
> Yeah, that's fishy. Smells like memory corruption. You can try two things:
>
> 1. Try to plug your QImage implementation into the test case. The idea is
>    basically to try to "bring" the test case a step at a time closer to
>    your application and see what triggers the change.
>
> 2. Try to run something like valgrind on your application (or the test
>    case if you manage to make it fail).
>
> Boris



From c.sell at byterefinery.de  Mon Mar  6 11:15:09 2017
From: c.sell at byterefinery.de (c.sell@byterefinery.de)
Date: Mon Mar  6 11:15:17 2017
Subject: [odb-users] collection mapping error
In-Reply-To: <boris.20170306152300@codesynthesis.com>
References: <20170305213742.Horde.v_xEXGL6PFPkmOrc7zNsAdh@webmail.byterefinery.de>
	<boris.20170306104246@codesynthesis.com>
	<20170306091249.Horde.qFcvEIuC_nqXG6F6HuRljtr@webmail.byterefinery.de>
	<boris.20170306122530@codesynthesis.com>
	<20170306125734.Horde.UoTXGi-GOSjxfSHX7c6-nWt@webmail.byterefinery.de>
	<boris.20170306152300@codesynthesis.com>
Message-ID: <20170306161509.Horde.Z5pORhT1Ba7-iJEmoMT5vn0@webmail.byterefinery.de>

Hello Boris,

I have been able to boil down the error to what I consider the  
simplest form. It requires the collection to hold QImage and therefore  
depends on my custom profile. However, I seriously doubt that the  
traits code itself is the problem, as almost none of it ever gets  
executed.

The test program simply tries to persist one object which holds a  
vector<QImage>. The image itself is invalid, and therefore gets  
translated to a NULL marker. There is no data copying etc. performed.

However, the very first insert leads to said object_already_persisten  
error. if anything, this error is in error.

regards,
Christian
-------------- next part --------------
A non-text attachment was scrubbed...
Name: odbdemo.tar.gz
Type: application/x-compressed-tar
Size: 5468 bytes
Desc: not available
Url : http://codesynthesis.com/pipermail/odb-users/attachments/20170306/42eabc7e/odbdemo.tar-0001.bin
From c.sell at byterefinery.de  Mon Mar  6 12:10:36 2017
From: c.sell at byterefinery.de (c.sell@byterefinery.de)
Date: Mon Mar  6 12:10:45 2017
Subject: [odb-users] collection mapping error
In-Reply-To: <boris.20170306152300@codesynthesis.com>
References: <20170305213742.Horde.v_xEXGL6PFPkmOrc7zNsAdh@webmail.byterefinery.de>
	<boris.20170306104246@codesynthesis.com>
	<20170306091249.Horde.qFcvEIuC_nqXG6F6HuRljtr@webmail.byterefinery.de>
	<boris.20170306122530@codesynthesis.com>
	<20170306125734.Horde.UoTXGi-GOSjxfSHX7c6-nWt@webmail.byterefinery.de>
	<boris.20170306152300@codesynthesis.com>
Message-ID: <20170306171036.Horde.3PzqfOyuMCepX88md5ePl1X@webmail.byterefinery.de>

ok, found it. The issue lies in the line my traits implementation that says

     is_null = v.isNull();

meaning, the column ist marked NULL if QImage#isNull() returns true.  
Which, in my mind, makes a lot of sense. However, the generated DDL  
marks the column as NOT NULL, and the vector<char> standard traits  
implementation always at least writes an empty column.

Now someone explain to me why collection items can never be NULL.  
Anyway, while this is so, any traits implementation that generates  
NULL values is not suitable for value collections

regards,
Chris


Zitat von Boris Kolpackov <boris@codesynthesis.com>:

> Hi Christian,
>
> c.sell@byterefinery.de <c.sell@byterefinery.de> writes:
>
>> It appears that the binding code is not executed because in function
>> insert(), the data_binding_test_version () check fails:
>>
>>     if (sts.data_binding_test_version ())
>>     {
>>       const binding& id (sts.id_binding ());
>>       bind (sts.data_bind (), id.bind, id.count, di);
>>       sts.data_binding_update_version ();
>>     }
>>
>>
>> in my test case program, the if branch is entered once for the collection
>> object. With my target app, it isn't ever
>
> Yeah, that's fishy. Smells like memory corruption. You can try two things:
>
> 1. Try to plug your QImage implementation into the test case. The idea is
>    basically to try to "bring" the test case a step at a time closer to
>    your application and see what triggers the change.
>
> 2. Try to run something like valgrind on your application (or the test
>    case if you manage to make it fail).
>
> Boris



From boris at codesynthesis.com  Tue Mar  7 07:23:49 2017
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Tue Mar  7 07:24:00 2017
Subject: [odb-users] collection mapping error
In-Reply-To: <20170306171036.Horde.3PzqfOyuMCepX88md5ePl1X@webmail.byterefinery.de>
References: <20170305213742.Horde.v_xEXGL6PFPkmOrc7zNsAdh@webmail.byterefinery.de>
	<boris.20170306104246@codesynthesis.com>
	<20170306091249.Horde.qFcvEIuC_nqXG6F6HuRljtr@webmail.byterefinery.de>
	<boris.20170306122530@codesynthesis.com>
	<20170306125734.Horde.UoTXGi-GOSjxfSHX7c6-nWt@webmail.byterefinery.de>
	<boris.20170306152300@codesynthesis.com>
	<20170306171036.Horde.3PzqfOyuMCepX88md5ePl1X@webmail.byterefinery.de>
Message-ID: <boris.20170307142010@codesynthesis.com>

Hi Christian,

c.sell@byterefinery.de <c.sell@byterefinery.de> writes:

> However, the generated DDL marks the column as NOT NULL,

That's the default. You can override (i.e., allow NULLs) with:

#pragma db value_null


> and the vector<char> standard traits implementation always at least
> writes an empty column.

Right, since std::vector<> has no notion of a NULL value.


> Now someone explain to me why collection items can never be NULL. Anyway,
> while this is so, any traits implementation that generates NULL values is
> not suitable for value collections

None of this is correct/makes sense. But I hope the above explanations
will help (if not, please read the manual; we both could have saved a
lot of time if you did so from the beginning).

Boris

From c.sell at byterefinery.de  Tue Mar  7 08:26:04 2017
From: c.sell at byterefinery.de (c.sell@byterefinery.de)
Date: Tue Mar  7 08:26:13 2017
Subject: [odb-users] collection mapping error
In-Reply-To: <boris.20170307142010@codesynthesis.com>
References: <20170305213742.Horde.v_xEXGL6PFPkmOrc7zNsAdh@webmail.byterefinery.de>
	<boris.20170306104246@codesynthesis.com>
	<20170306091249.Horde.qFcvEIuC_nqXG6F6HuRljtr@webmail.byterefinery.de>
	<boris.20170306122530@codesynthesis.com>
	<20170306125734.Horde.UoTXGi-GOSjxfSHX7c6-nWt@webmail.byterefinery.de>
	<boris.20170306152300@codesynthesis.com>
	<20170306171036.Horde.3PzqfOyuMCepX88md5ePl1X@webmail.byterefinery.de>
	<boris.20170307142010@codesynthesis.com>
Message-ID: <20170307132604.Horde.8W9SmOxi71AvPAz7plskcrQ@webmail.byterefinery.de>

Hi Boris,

> Right, since std::vector<> has no notion of a NULL value.

It is your choice to represent an empty vector<char> by an empty byte  
sequence. It was my choice to represent an empty image by a NULL  
value. This is only dealing with how they are stored, the collection  
that holds them still sees the empty value

> None of this is correct/makes sense. But I hope the above explanations
> will help (if not, please read the manual; we both could have saved a
> lot of time if you did so from the beginning).

Sorry, but I dont see how reading the manual would have helped. I was  
faced with an error message stating "object already persistent", when  
in fact it was a constraint violation caused by a non-obvious (and  
undocumented) interdependence between the default data model (NOT NULL  
constraints on collection value columns) and the traits implementation  
(which should not set the NULL marker if used with the default data  
model).

I suggest an amendment to the documentation that points to the fact  
that the default data model for collections imposes NOT NULL  
constraints (I may have overread that). And maybe even some hint on  
how to deal with is_null in traits implementations (I dont think I  
overread that, I searched thoroughly for docs and only found a  
reference to sample code).

Anyway, not an issue, this is a free lunch. I just don't want to "not  
make sense".

May I still take you up on your offer regarding the pre-release build  
for windows?

regards,
Christian

From lloydkl.tech at gmail.com  Tue Mar  7 23:42:40 2017
From: lloydkl.tech at gmail.com (Lloyd)
Date: Tue Mar  7 23:42:52 2017
Subject: [odb-users] Compile error , Memory leak detection enabled
Message-ID: <CAA10ZTqZEM=Gd6WGMPSkYzqkT1jgphUWNbWxaVSXnHZLOaTLsg@mail.gmail.com>

Hi,

When I compile the application with Visual Studio CRT with memory leak
detection (https://msdn.microsoft.com/en-us/library/x98tx3cf.aspx) I get
the following error from ODB. Is it a problem associated with ODB?

Libraries\odb\libodb-2.4.0\odb/pointer-traits.hxx(156): error C2059: syntax
error: 'constant'
Libraries\odb\libodb-2.4.0\odb/pointer-traits.hxx(160): note: see reference
to class template instantiation 'odb::pointer_traits<T*>' being compiled

Thanks,
  Lloyd
From c.sell at byterefinery.de  Wed Mar  8 06:41:43 2017
From: c.sell at byterefinery.de (c.sell@byterefinery.de)
Date: Wed Mar  8 08:42:58 2017
Subject: [odb-users] collection mapping error
In-Reply-To: <20170307132604.Horde.8W9SmOxi71AvPAz7plskcrQ@webmail.byterefinery.de>
References: <20170305213742.Horde.v_xEXGL6PFPkmOrc7zNsAdh@webmail.byterefinery.de>
	<boris.20170306104246@codesynthesis.com>
	<20170306091249.Horde.qFcvEIuC_nqXG6F6HuRljtr@webmail.byterefinery.de>
	<boris.20170306122530@codesynthesis.com>
	<20170306125734.Horde.UoTXGi-GOSjxfSHX7c6-nWt@webmail.byterefinery.de>
	<boris.20170306152300@codesynthesis.com>
	<20170306171036.Horde.3PzqfOyuMCepX88md5ePl1X@webmail.byterefinery.de>
	<boris.20170307142010@codesynthesis.com>
	<20170307132604.Horde.8W9SmOxi71AvPAz7plskcrQ@webmail.byterefinery.de>
Message-ID: <20170308114143.Horde.h7545XASpxta-ccrkhUqJJG@webmail.byterefinery.de>

Hello,

so, I've done my reading, for the record. NULL value sematics are  
indeed documented. Assuming not-null semantics for values is a  
conceptual decision, which is also documented and circumventable  
(odb::nullable etc.)

the fact that the exception states "object already persistent" rather  
than "constraint violation" remains misleading.

the fact that the is_null parameter in the set_image function of  
value_traits implementations should be used with caution might deserve  
mentioning. All examples simply set it to false, and never say why.

Apart from the above, the more I read into ODB, the more impressed I  
am. Beautiful software, very capable and quite well documented. I  
wrote a pretty complete O/R mapper for Smalltalk almost 20 years ago  
and know the Java bunch (Hibernate, JPA). So I feel I can make that  
statement (and some others)

regards,
Christian

Zitat von c.sell@byterefinery.de:

> Hi Boris,
>
>> Right, since std::vector<> has no notion of a NULL value.
>
> It is your choice to represent an empty vector<char> by an empty  
> byte sequence. It was my choice to represent an empty image by a  
> NULL value. This is only dealing with how they are stored, the  
> collection that holds them still sees the empty value
>
>> None of this is correct/makes sense. But I hope the above explanations
>> will help (if not, please read the manual; we both could have saved a
>> lot of time if you did so from the beginning).
>
> Sorry, but I dont see how reading the manual would have helped. I  
> was faced with an error message stating "object already persistent",  
> when in fact it was a constraint violation caused by a non-obvious  
> (and undocumented) interdependence between the default data model  
> (NOT NULL constraints on collection value columns) and the traits  
> implementation (which should not set the NULL marker if used with  
> the default data model).
>
> I suggest an amendment to the documentation that points to the fact  
> that the default data model for collections imposes NOT NULL  
> constraints (I may have overread that). And maybe even some hint on  
> how to deal with is_null in traits implementations (I dont think I  
> overread that, I searched thoroughly for docs and only found a  
> reference to sample code).
>
> Anyway, not an issue, this is a free lunch. I just don't want to  
> "not make sense".
>
> May I still take you up on your offer regarding the pre-release  
> build for windows?
>
> regards,
> Christian


From boris at codesynthesis.com  Wed Mar  8 08:50:23 2017
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Wed Mar  8 08:50:32 2017
Subject: [odb-users] collection mapping error
In-Reply-To: <20170308114143.Horde.h7545XASpxta-ccrkhUqJJG@webmail.byterefinery.de>
References: <20170305213742.Horde.v_xEXGL6PFPkmOrc7zNsAdh@webmail.byterefinery.de>
	<boris.20170306104246@codesynthesis.com>
	<20170306091249.Horde.qFcvEIuC_nqXG6F6HuRljtr@webmail.byterefinery.de>
	<boris.20170306122530@codesynthesis.com>
	<20170306125734.Horde.UoTXGi-GOSjxfSHX7c6-nWt@webmail.byterefinery.de>
	<boris.20170306152300@codesynthesis.com>
	<20170306171036.Horde.3PzqfOyuMCepX88md5ePl1X@webmail.byterefinery.de>
	<boris.20170307142010@codesynthesis.com>
	<20170307132604.Horde.8W9SmOxi71AvPAz7plskcrQ@webmail.byterefinery.de>
	<20170308114143.Horde.h7545XASpxta-ccrkhUqJJG@webmail.byterefinery.de>
Message-ID: <boris.20170308154520@codesynthesis.com>

Hi Christian,

c.sell@byterefinery.de <c.sell@byterefinery.de> writes:

> so, I've done my reading, for the record.

Thanks, I appreciate it.


> the fact that the exception states "object already persistent" rather than
> "constraint violation" remains misleading.

The reason for it is described in Subsection 18.5.4, "Constraint Violations"
of Section 18.5, "SQLite Limitations". Generally, it makes sense to read
through that whole section for every database that one is using.


> Apart from the above, the more I read into ODB, the more impressed I am.
> Beautiful software, very capable and quite well documented. I wrote a pretty
> complete O/R mapper for Smalltalk almost 20 years ago and know the Java
> bunch (Hibernate, JPA). So I feel I can make that statement (and some
> others)

Thanks, I appreciate the kind words.

BTW, a pre-release build for Windows is available here:

http://codesynthesis.com/~boris/tmp/odb/pre-release/a11/

Boris

From boris at codesynthesis.com  Wed Mar  8 08:57:35 2017
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Wed Mar  8 08:57:45 2017
Subject: [odb-users] Compile error , Memory leak detection enabled
In-Reply-To: <CAA10ZTqZEM=Gd6WGMPSkYzqkT1jgphUWNbWxaVSXnHZLOaTLsg@mail.gmail.com>
References: <CAA10ZTqZEM=Gd6WGMPSkYzqkT1jgphUWNbWxaVSXnHZLOaTLsg@mail.gmail.com>
Message-ID: <boris.20170308155307@codesynthesis.com>

Hi Lloyd,

Lloyd <lloydkl.tech@gmail.com> writes:

> When I compile the application with Visual Studio CRT with memory leak
> detection (https://msdn.microsoft.com/en-us/library/x98tx3cf.aspx) I get
> the following error from ODB. Is it a problem associated with ODB?
> 
> Libraries\odb\libodb-2.4.0\odb/pointer-traits.hxx(156): error C2059: syntax
> error: 'constant'
> Libraries\odb\libodb-2.4.0\odb/pointer-traits.hxx(160): note: see reference
> to class template instantiation 'odb::pointer_traits<T*>' being compiled

I think this is a problem associated with the "Visual Studio CRT with memory
leak detection". The line the error points to looks like this:

free (void* p)

There is no mentioning of name 'constant'. My suspicion is that this leak
detection build uses macro hackery to redefine some symbols (like 'free',
which is not really legal per the C++ standard). You may want to report
this to Microsoft (and/or use some more modern memory leak detector).

Boris

From odb at a-cunningham.com  Thu Mar  9 18:13:46 2017
From: odb at a-cunningham.com (Andrew Cunningham)
Date: Thu Mar  9 18:14:00 2017
Subject: [odb-users] erase of object in section
Message-ID: <CAGTJuHifhcrVtQtJ-Wd64P_3CVH=BfmmGnac55=4yTRkFT_g4Q@mail.gmail.com>

Hi Boris,

I have an object with members as below

#pragma db load(lazy) update(change)
    mutable odb::section                masssection_;
#pragma db section(masssection_)
    std::unique_ptr<CDBFloatMatrix>        massDistribMatrix_;


Before I erase the "containing" object do I need to load the section to
ensure the proper erasure  of the massDistribMatrix_ object?

I was assuming the answer is "no".

Andrew
From boris at codesynthesis.com  Fri Mar 10 05:16:52 2017
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Fri Mar 10 05:17:00 2017
Subject: [odb-users] erase of object in section
In-Reply-To: <CAGTJuHifhcrVtQtJ-Wd64P_3CVH=BfmmGnac55=4yTRkFT_g4Q@mail.gmail.com>
References: <CAGTJuHifhcrVtQtJ-Wd64P_3CVH=BfmmGnac55=4yTRkFT_g4Q@mail.gmail.com>
Message-ID: <boris.20170310120934@codesynthesis.com>

Hi Andrew,

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

> #pragma db load(lazy) update(change)
>     mutable odb::section                masssection_;
> #pragma db section(masssection_)
>     std::unique_ptr<CDBFloatMatrix>        massDistribMatrix_;
> 
> Before I erase the "containing" object do I need to load the section to
> ensure the proper erasure  of the massDistribMatrix_ object?

This "containing" terminology is fuzzy. You have a pointer in
massDistribMatrix_ and you have a "pointed-to" object. You can
"erase" the pointer (e.g., by setting it to NULL or to point to
some other object) or you can erase the pointed-to object, or,
perhaps, both. Note also that these two operations are completely
independent (normally; unless you use something like ON DELETE
CASCADE). So what exactly do you want to "erase"?

Boris

From balazs.bayer at gmail.com  Wed Mar 15 08:32:45 2017
From: balazs.bayer at gmail.com (=?UTF-8?Q?Bal=C3=A1zs_Bayer?=)
Date: Wed Mar 15 09:05:50 2017
Subject: [odb-users] Persisting class instantiated from a template
Message-ID: <CACjO_ym-zb5Jo2bv-KxBe=F7V4GfTrM9tufW7d8qYnscYXczEQ@mail.gmail.com>

Hi all,

I would like to persist a class instantiated from a template into an SQLite
database using ODB. Lets say I have the following class in a header called
*result.hpp*:

template <typename valuetype, int size>class Result {
  public:
    valuetype getValue(int i) const { return Values[i]; };
    int getId() const { return Id; };
    void setValue(valuetype _value, int i) { Values[i] = _value; }
    void setId(int _Id) { Id = _Id; }

  private:
    int Id;
    std::array<valuetype, size> Values;};


and I have another header file called *result-mapping.hpp* with the
following content:

#include "result.hpp"
typedef Result<float, 7> ResultFloatSeven;
#pragma db value(ResultFloatSeven) definition transient#pragma db
member(ResultFloatSeven::getId) virtual(int) access(getId) <-- line
11#pragma db member(ResultFloatSeven::getValue) virtual(float)
access(getValue)

When I try to generate the database support code with the command

odb --std c++11 -d sqlite --odb-prologue "#include \"result-mapping.hpp\""
--generate-query result.hpp

I get the following error:

result-mapping.hpp:11:57: error: unable to find suitable modifier function
'getId'

What is wrong here?

Thanks!
Bal?zs

Btw I laready posted this question on stackoverflow, but did not get any
feedback until now.

http://stackoverflow.com/questions/42684473/c-odb-persisting-class-instantiated-from-a-template
From boris at codesynthesis.com  Wed Mar 15 11:22:10 2017
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Wed Mar 15 11:22:19 2017
Subject: [odb-users] Persisting class instantiated from a template
In-Reply-To: <CACjO_ym-zb5Jo2bv-KxBe=F7V4GfTrM9tufW7d8qYnscYXczEQ@mail.gmail.com>
References: <CACjO_ym-zb5Jo2bv-KxBe=F7V4GfTrM9tufW7d8qYnscYXczEQ@mail.gmail.com>
Message-ID: <boris.20170315172043@codesynthesis.com>

Hi Bal?zs,

Bal?zs Bayer <balazs.bayer@gmail.com> writes:

> typedef Result<float, 7> ResultFloatSeven;
> #pragma db value(ResultFloatSeven) definition transient
> #pragma db member(ResultFloatSeven::getId) virtual(int) access(getId)
> #pragma db member(ResultFloatSeven::getValue) virtual(float) access(getValue)

You are using function names for virtual member names so ODB tries to
derive modifier names from these function names which is not what you
want. Try to change this to:

#pragma db member(ResultFloatSeven::Id_) virtual(int)
#pragma db member(ResultFloatSeven::Value_) virtual(float)

Also, I don't think you need to go through the virtual member route
in this case. ODB should be able to map everything automatically with
just:

typedef Result<float, 7> ResultFloatSeven;
#pragma db value(ResultFloatSeven) definition

Boris

From abv150ci at gmail.com  Wed Mar 15 14:36:31 2017
From: abv150ci at gmail.com (=?UTF-8?Q?Aar=C3=B3n_Bueno_Villares?=)
Date: Wed Mar 15 14:37:23 2017
Subject: [odb-users] Database schema migration
Message-ID: <CAMbYJhY0_h_RgyUQXjjBzg2fHzH0CCV6o2U2UMmy2nT0O16KEQ@mail.gmail.com>

Quote:

If your object model consists of a large number of header files and you
generate the database schema for each of them individually, then a
changelog will be created for each of your header files. [...]

The solution to both of these problems is to generate a combined database
schema file and a single changelog. [...]

To generate the database schema, we perform a separate invocation of the
ODB compiler. This time, however, we instruct it to only generate the
schema (--generate-schema-only) and produce it combined (--at-once) for all
the files in our object model

However, ODB has a --changelog *file *option which allows to specify your
own changelog filename. Wouldn't be easier to just specify the same change
log file on each invocation of the ODB compiler over the different class
files? It removes an extra command and it's very easy to add a new ODB
option in a typical makefile (I use a ODB_FLAGS variable to my makefile).
From balazs.bayer at gmail.com  Wed Mar 15 13:29:02 2017
From: balazs.bayer at gmail.com (=?UTF-8?Q?Bal=C3=A1zs_Bayer?=)
Date: Thu Mar 16 04:41:30 2017
Subject: [odb-users] Persisting class instantiated from a template
In-Reply-To: <boris.20170315172043@codesynthesis.com>
References: <CACjO_ym-zb5Jo2bv-KxBe=F7V4GfTrM9tufW7d8qYnscYXczEQ@mail.gmail.com>
	<boris.20170315172043@codesynthesis.com>
Message-ID: <CACjO_ym+kqhVdHb+mKnuekGFgT0TyZ2dJsu7vWpn8XBianF0cg@mail.gmail.com>

When I tried with this one:

#pragma db value(ResultFloatSeven) definition transient
#pragma db member(ResultFloatSeven::Id_) virtual(int)
#pragma db member(ResultFloatSeven::Value_) virtual(float)

I got the following errors:

result-mapping.hpp:12:12: error: virtual data member declaration 'Id'
conflicts with a previous declaration
result.hpp:41:7: info: 'Id' was previously declared here
result-mapping.hpp:12:12: error: virtual data member declaration 'Id'
conflicts with a previous declaration
result.hpp:41:7: info: 'Id' was previously declared here

and with this one:

#pragma db value(ResultFloatSeven) definition

the followings:

result.hpp:42:31: error: data member 'Values' is private and no suitable
accessor function could be automatically found
result.hpp:42:31: info: consider making class 'odb::access' a friend of
class 'ResultFloatSeven'
result.hpp:42:31: info: or use '#pragma db get' to explicitly specify the
accessor function or expression
result.hpp:42:31: error: data member 'Values' is private and no suitable
accessor function could be automatically found
result.hpp:42:31: info: consider making class 'odb::access' a friend of
class 'ResultFloatSeven'
result.hpp:42:31: info: or use '#pragma db get' to explicitly specify the
accessor function or expression
From balazs.bayer at gmail.com  Wed Mar 15 13:38:56 2017
From: balazs.bayer at gmail.com (=?UTF-8?Q?Bal=C3=A1zs_Bayer?=)
Date: Thu Mar 16 04:41:30 2017
Subject: [odb-users] Persisting class instantiated from a template
In-Reply-To: <CACjO_ym+kqhVdHb+mKnuekGFgT0TyZ2dJsu7vWpn8XBianF0cg@mail.gmail.com>
References: <CACjO_ym-zb5Jo2bv-KxBe=F7V4GfTrM9tufW7d8qYnscYXczEQ@mail.gmail.com>
	<boris.20170315172043@codesynthesis.com>
	<CACjO_ym+kqhVdHb+mKnuekGFgT0TyZ2dJsu7vWpn8XBianF0cg@mail.gmail.com>
Message-ID: <CACjO_ykjGxHNmsyPZEOZc9Lr1rKi4rZKxa=Pj7svkNGqYC1eqg@mail.gmail.com>

Sorry, the errors for

#pragma db value(ResultFloatSeven) definition transient
#pragma db member(ResultFloatSeven::Id_) virtual(int)
#pragma db member(ResultFloatSeven::Value_) virtual(float)

are:

result.hpp:42:31: error: data member 'Values' is private and no suitable
accessor function could be automatically found
result.hpp:42:31: info: consider making class 'odb::access' a friend of
class 'ResultFloatSeven'
result.hpp:42:31: info: or use '#pragma db get' to explicitly specify the
accessor function or expression
result.hpp:42:31: error: data member 'Values' is private and no suitable
accessor function could be automatically found
result.hpp:42:31: info: consider making class 'odb::access' a friend of
class 'ResultFloatSeven'
result.hpp:42:31: info: or use '#pragma db get' to explicitly specify the
accessor function or expression
From boris at codesynthesis.com  Thu Mar 16 12:25:00 2017
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Thu Mar 16 12:25:10 2017
Subject: [odb-users] Database schema migration
In-Reply-To: <CAMbYJhY0_h_RgyUQXjjBzg2fHzH0CCV6o2U2UMmy2nT0O16KEQ@mail.gmail.com>
References: <CAMbYJhY0_h_RgyUQXjjBzg2fHzH0CCV6o2U2UMmy2nT0O16KEQ@mail.gmail.com>
Message-ID: <boris.20170316182050@codesynthesis.com>

Hi,

Aar?n Bueno Villares <abv150ci@gmail.com> writes:

> Wouldn't be easier to just specify the same change log file on each
> invocation of the ODB compiler over the different class files?

Yeah, sure. It would also be pretty broken.

When you regenerate the changelog, ODB makes a diff between database
models. But now the difference can mean two things: something changed
or this is a database model for a different subset.

Boris

From boris at codesynthesis.com  Thu Mar 16 12:37:37 2017
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Thu Mar 16 12:37:46 2017
Subject: [odb-users] Persisting class instantiated from a template
In-Reply-To: <CACjO_ykjGxHNmsyPZEOZc9Lr1rKi4rZKxa=Pj7svkNGqYC1eqg@mail.gmail.com>
References: <CACjO_ym-zb5Jo2bv-KxBe=F7V4GfTrM9tufW7d8qYnscYXczEQ@mail.gmail.com>
	<boris.20170315172043@codesynthesis.com>
	<CACjO_ym+kqhVdHb+mKnuekGFgT0TyZ2dJsu7vWpn8XBianF0cg@mail.gmail.com>
	<CACjO_ykjGxHNmsyPZEOZc9Lr1rKi4rZKxa=Pj7svkNGqYC1eqg@mail.gmail.com>
Message-ID: <boris.20170316183337@codesynthesis.com>

Hi Bal?zs,

Ok, I tried to actually compile your example with this contents of
the mapping file:

typedef Result<float, 7> ResultFloatSeven;

#pragma db value(ResultFloatSeven) definition transient
#pragma db member(ResultFloatSeven::Id_) virtual(int)
#pragma db member(ResultFloatSeven::Value_) virtual(float)

The error I get is this:

mapping.hpp:7:12: error: no suitable accessor function could be automatically found for virtual data member 'Value_'
mapping.hpp:7:12: info: use '#pragma db get' to explicitly specify the accessor function or expression

Which makes sense, after looking at what you have:

valuetype getValue(int i) const { return Values[i]; }
void setValue(valuetype _value, int i) { Values[i] = _value; }

These are not suitable accessors/modifiers (ODB cannot magically know
what to pass for i). You need something like this:

valuetype getValue() const;
void setValue(valuetype)

Boris

From balazs.bayer at gmail.com  Thu Mar 16 13:17:27 2017
From: balazs.bayer at gmail.com (=?UTF-8?Q?Bal=C3=A1zs_Bayer?=)
Date: Thu Mar 16 13:24:10 2017
Subject: [odb-users] Persisting class instantiated from a template
In-Reply-To: <boris.20170316183337@codesynthesis.com>
References: <CACjO_ym-zb5Jo2bv-KxBe=F7V4GfTrM9tufW7d8qYnscYXczEQ@mail.gmail.com>
	<boris.20170315172043@codesynthesis.com>
	<CACjO_ym+kqhVdHb+mKnuekGFgT0TyZ2dJsu7vWpn8XBianF0cg@mail.gmail.com>
	<CACjO_ykjGxHNmsyPZEOZc9Lr1rKi4rZKxa=Pj7svkNGqYC1eqg@mail.gmail.com>
	<boris.20170316183337@codesynthesis.com>
Message-ID: <CACjO_ykyZjTMgMY02MgM0MFO3=m209Khtg1TGND2B2KenqBX4A@mail.gmail.com>

Hi Boris,

the thing is, that as Value is a std::array, and although in these code
snippets its size is only 7, I am expecting to have ~1000 long arrays. And
if I do this:

std::array<float, 1000> const &getValue() const { return Value; };
void setValue(const std::array<float, 1000> &NewValue) { Value =
std::move(NewValue); };

which is somehow analogue to

valuetype getValue() const;
void setValue(valuetype)

then I guess the problem is, that when I want to change only one specific
member of the Value array, I will have to construct another huge array, and
pass it to setValue, which does not seem to be efficient to me. Do you have
any suggestions, how to overcome this situation?

Many Thanks,
Balazs
From boris at codesynthesis.com  Fri Mar 17 11:24:17 2017
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Fri Mar 17 11:24:26 2017
Subject: [odb-users] Persisting class instantiated from a template
In-Reply-To: <CACjO_ykyZjTMgMY02MgM0MFO3=m209Khtg1TGND2B2KenqBX4A@mail.gmail.com>
References: <CACjO_ym-zb5Jo2bv-KxBe=F7V4GfTrM9tufW7d8qYnscYXczEQ@mail.gmail.com>
	<boris.20170315172043@codesynthesis.com>
	<CACjO_ym+kqhVdHb+mKnuekGFgT0TyZ2dJsu7vWpn8XBianF0cg@mail.gmail.com>
	<CACjO_ykjGxHNmsyPZEOZc9Lr1rKi4rZKxa=Pj7svkNGqYC1eqg@mail.gmail.com>
	<boris.20170316183337@codesynthesis.com>
	<CACjO_ykyZjTMgMY02MgM0MFO3=m209Khtg1TGND2B2KenqBX4A@mail.gmail.com>
Message-ID: <boris.20170317172206@codesynthesis.com>

Hi Bal?zs,

Bal?zs Bayer <balazs.bayer@gmail.com> writes:

> std::array<float, 1000> const &getValue() const { return Value; };
> void setValue(const std::array<float, 1000> &NewValue) { Value =
> std::move(NewValue); };
> 
> then I guess the problem is, that when I want to change only one specific
> member of the Value array, I will have to construct another huge array, and
> pass it to setValue, which does not seem to be efficient to me. Do you have
> any suggestions, how to overcome this situation?

You can do by-reference access and modification:

const std::array<float, 1000>& getValue() const   { return Value; };
std::array<float, 1000>&       setValue()         { return Value; };

Boris

From obermann.lukas at gmail.com  Fri Mar 17 14:09:54 2017
From: obermann.lukas at gmail.com (Lukas Obermann)
Date: Fri Mar 17 14:10:08 2017
Subject: [odb-users] Subselects in View
Message-ID: <CC447CBF-19F1-4C94-8BFB-A543CF002DD7@gmail.com>

Hi there,

I know that it is not recommended to do complex queries with odb, but I wonder if it is possible to get a simple subselect working in a view or somewhere else.

Like this simplified query:
SELECT table1.name, (SELECT AVG(score) FROM table2 WHERE table2.t_id = table1.t_id) as score FROM table1

Thanks!
Lukas

From balazs.bayer at gmail.com  Fri Mar 17 13:07:39 2017
From: balazs.bayer at gmail.com (=?UTF-8?Q?Bal=C3=A1zs_Bayer?=)
Date: Mon Mar 20 09:42:04 2017
Subject: [odb-users] Persisting class instantiated from a template
In-Reply-To: <boris.20170317172206@codesynthesis.com>
References: <CACjO_ym-zb5Jo2bv-KxBe=F7V4GfTrM9tufW7d8qYnscYXczEQ@mail.gmail.com>
	<boris.20170315172043@codesynthesis.com>
	<CACjO_ym+kqhVdHb+mKnuekGFgT0TyZ2dJsu7vWpn8XBianF0cg@mail.gmail.com>
	<CACjO_ykjGxHNmsyPZEOZc9Lr1rKi4rZKxa=Pj7svkNGqYC1eqg@mail.gmail.com>
	<boris.20170316183337@codesynthesis.com>
	<CACjO_ykyZjTMgMY02MgM0MFO3=m209Khtg1TGND2B2KenqBX4A@mail.gmail.com>
	<boris.20170317172206@codesynthesis.com>
Message-ID: <CACjO_ynmyTzg+-uxn06PFskbyzipWFg3pbKqMuEKVtMYkpyRFQ@mail.gmail.com>

Hi Boris,

now its fine. Sorry for my poor c++ knowledge, and thanks very much for
your help.

Best regards,
Bal?zs
From balazs.bayer at gmail.com  Sat Mar 18 06:33:45 2017
From: balazs.bayer at gmail.com (=?UTF-8?Q?Bal=C3=A1zs_Bayer?=)
Date: Mon Mar 20 09:42:04 2017
Subject: [odb-users] Persisting class instantiated from a template
In-Reply-To: <CACjO_ynmyTzg+-uxn06PFskbyzipWFg3pbKqMuEKVtMYkpyRFQ@mail.gmail.com>
References: <CACjO_ym-zb5Jo2bv-KxBe=F7V4GfTrM9tufW7d8qYnscYXczEQ@mail.gmail.com>
	<boris.20170315172043@codesynthesis.com>
	<CACjO_ym+kqhVdHb+mKnuekGFgT0TyZ2dJsu7vWpn8XBianF0cg@mail.gmail.com>
	<CACjO_ykjGxHNmsyPZEOZc9Lr1rKi4rZKxa=Pj7svkNGqYC1eqg@mail.gmail.com>
	<boris.20170316183337@codesynthesis.com>
	<CACjO_ykyZjTMgMY02MgM0MFO3=m209Khtg1TGND2B2KenqBX4A@mail.gmail.com>
	<boris.20170317172206@codesynthesis.com>
	<CACjO_ynmyTzg+-uxn06PFskbyzipWFg3pbKqMuEKVtMYkpyRFQ@mail.gmail.com>
Message-ID: <CACjO_ykhC7cF6t0Lo5SvUUhoODkni0R0c4VKaEmvFOHaMw-9ow@mail.gmail.com>

actually one more thing:

with this:

int Id = 1;
std::array<float, 7> Values = {{0, 1, 2, 3, 4, 5, 6}};
ResultFloatSeven S(Values, Id);

{
    std::unique_ptr<odb::database> db(new odb::sqlite::database(
        "./test.db", SQLITE_OPEN_READWRITE | SQLITE_OPEN_CREATE));

    odb::transaction t(db->begin());
    db->persist(S);
    t.commit();
}

Now I get the following compilation error for the line "db->persist(S);":

In file included from main.cpp:9:
In file included from /usr/include/odb/sqlite/database.hxx:16:
In file included from /usr/include/odb/database.hxx:23:
/usr/include/odb/traits.hxx:178:5: error: implicit instantiation of
undefined template 'odb::access::object_traits_impl<result::Result<float,
7>, 5>'
    access::object_traits_impl<T, DB>,
    ^
/usr/include/odb/database.txx:38:5: note: in instantiation of template
class 'odb::object_traits_impl<result::Result<float, 7>, 5>' requested here
    object_traits::persist (*this, obj);
    ^
/usr/include/odb/database.ixx:160:12: note: in instantiation of function
template specialization 'odb::database::persist_<result::Result<float, 7>,
5>' requested here
    return persist_<T, id_common> (obj);
           ^
main.cpp:27:9: note: in instantiation of function template specialization
'odb::database::persist<result::Result<float, 7> >' requested here
    db->persist(S);
        ^
/usr/include/odb/forward.hxx:99:11: note: template is declared here
    class object_traits_impl;
          ^
In file included from main.cpp:9:
In file included from /usr/include/odb/sqlite/database.hxx:16:
In file included from /usr/include/odb/database.hxx:632:
/usr/include/odb/database.txx:38:18: error: incomplete definition of type
'odb::object_traits_impl<result::Result<float, 7>, 5>'
    object_traits::persist (*this, obj);
    ~~~~~~~~~~~~~^~
/usr/include/odb/database.ixx:160:12: note: in instantiation of function
template specialization 'odb::database::persist_<result::Result<float, 7>,
5>' requested here
    return persist_<T, id_common> (obj);
           ^
main.cpp:27:9: note: in instantiation of function template specialization
'odb::database::persist<result::Result<float, 7> >' requested here
    db->persist(S);
        ^
In file included from main.cpp:9:
In file included from /usr/include/odb/sqlite/database.hxx:16:
In file included from /usr/include/odb/database.hxx:632:
/usr/include/odb/database.txx:40:29: error: no member named
'reference_cache_traits' in 'odb::object_traits_impl<result::Result<float,
7>, 5>'
    typename object_traits::reference_cache_traits::position_type p (

Any idea?

Thanks,
Balazs
From boris at codesynthesis.com  Mon Mar 20 09:45:52 2017
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Mar 20 09:46:00 2017
Subject: [odb-users] Persisting class instantiated from a template
In-Reply-To: <CACjO_ykhC7cF6t0Lo5SvUUhoODkni0R0c4VKaEmvFOHaMw-9ow@mail.gmail.com>
References: <CACjO_ym-zb5Jo2bv-KxBe=F7V4GfTrM9tufW7d8qYnscYXczEQ@mail.gmail.com>
	<boris.20170315172043@codesynthesis.com>
	<CACjO_ym+kqhVdHb+mKnuekGFgT0TyZ2dJsu7vWpn8XBianF0cg@mail.gmail.com>
	<CACjO_ykjGxHNmsyPZEOZc9Lr1rKi4rZKxa=Pj7svkNGqYC1eqg@mail.gmail.com>
	<boris.20170316183337@codesynthesis.com>
	<CACjO_ykyZjTMgMY02MgM0MFO3=m209Khtg1TGND2B2KenqBX4A@mail.gmail.com>
	<boris.20170317172206@codesynthesis.com>
	<CACjO_ynmyTzg+-uxn06PFskbyzipWFg3pbKqMuEKVtMYkpyRFQ@mail.gmail.com>
	<CACjO_ykhC7cF6t0Lo5SvUUhoODkni0R0c4VKaEmvFOHaMw-9ow@mail.gmail.com>
Message-ID: <boris.20170320154436@codesynthesis.com>

Hi Bal?zs,

Bal?zs Bayer <balazs.bayer@gmail.com> writes:

> Now I get the following compilation error for the line "db->persist(S);":

Did you #include all the relevant *-odb.hxx files?

If that doesn't help, please send a minimal but complete test case that
reproduces the problem.

Boris

From boris at codesynthesis.com  Mon Mar 20 12:21:02 2017
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Mon Mar 20 12:21:11 2017
Subject: [odb-users] Subselects in View
In-Reply-To: <CC447CBF-19F1-4C94-8BFB-A543CF002DD7@gmail.com>
References: <CC447CBF-19F1-4C94-8BFB-A543CF002DD7@gmail.com>
Message-ID: <boris.20170320181701@codesynthesis.com>

Hi Lukas,

Lukas Obermann <obermann.lukas@gmail.com> writes:

> I know that it is not recommended to do complex queries with odb, [...]

Not recommended by whom?

You can handle pretty much any query with an ODB native view. It may not
be as convenient as for simpler queries, but it is definitely possible
and not "not recommended".


> SELECT table1.name, (SELECT AVG(score) FROM table2 WHERE
>   table2.t_id = table1.t_id) as score FROM table1

You can handle it pretty easily with a native view. You can even get fancy
and do something like this:

#include <string>

#pragma db object
struct table1
{
  #pragma db id
  int t_id;

  std::string name;
};

#pragma db object
struct table2
{
  #pragma db id
  int t_id;

  int score;
};

#pragma db view object(table1) object(table2)                            \
  query("SELECT " + table1::name + ", "                                  \
        "("                                                              \
        " SELECT AVG(" + table2::score + ")"                             \
        " FROM " + (object_traits_impl<table2, id_common>::table_name) + \
        " WHERE " + (table2::t_id == table1::t_id) +                     \
        ") AS score"                                                     \
        "FROM " + (object_traits_impl<table1, id_common>::table_name))
struct view
{
  std::string name;
  int avg_score;
};

You can read moe about table_name here:

[1] http://www.codesynthesis.com/pipermail/odb-users/2016-April/003209.html

Boris

From balazs.bayer at gmail.com  Mon Mar 20 15:33:30 2017
From: balazs.bayer at gmail.com (=?UTF-8?Q?Bal=C3=A1zs_Bayer?=)
Date: Tue Mar 21 04:38:57 2017
Subject: [odb-users] Persisting class instantiated from a template
In-Reply-To: <boris.20170320154436@codesynthesis.com>
References: <CACjO_ym-zb5Jo2bv-KxBe=F7V4GfTrM9tufW7d8qYnscYXczEQ@mail.gmail.com>
	<boris.20170315172043@codesynthesis.com>
	<CACjO_ym+kqhVdHb+mKnuekGFgT0TyZ2dJsu7vWpn8XBianF0cg@mail.gmail.com>
	<CACjO_ykjGxHNmsyPZEOZc9Lr1rKi4rZKxa=Pj7svkNGqYC1eqg@mail.gmail.com>
	<boris.20170316183337@codesynthesis.com>
	<CACjO_ykyZjTMgMY02MgM0MFO3=m209Khtg1TGND2B2KenqBX4A@mail.gmail.com>
	<boris.20170317172206@codesynthesis.com>
	<CACjO_ynmyTzg+-uxn06PFskbyzipWFg3pbKqMuEKVtMYkpyRFQ@mail.gmail.com>
	<CACjO_ykhC7cF6t0Lo5SvUUhoODkni0R0c4VKaEmvFOHaMw-9ow@mail.gmail.com>
	<boris.20170320154436@codesynthesis.com>
Message-ID: <CACjO_y=+BnWnyRMPc3bM5jirjN3U-MQhjfN_pPq=1fFsVyOXWw@mail.gmail.com>

Hi Boris,

you can get the source files via these links:

main.cpp <http://pastebin.com/9AcrSf0B>
result.hpp <http://pastebin.com/FU2vxWBS>
result-mapping.hpp <http://pastebin.com/sPBKuEaC>

I generate the support code  and compile the whole propgram with these
commands:

odb --std c++11 -d sqlite -s --odb-epilogue '#include "result-mapping.hpp"'
--generate-query result.hpp
odb --std c++11 -d sqlite -s --generate-query result-mapping.hpp
clang++ -std=c++11 -c -o result-odb.o result-odb.cxx
clang++ -std=c++11 -c -o result-mapping-odb.o result-mapping-odb.cxx
clang++ -std=c++11 -c -o main.o main.cpp <= at this point I get the below
error message
clang++ -o fut main.o result-odb.o result-mapping-odb.o -lodb-sqlite -lodb

The error message is:

In file included from main.cpp:8:
In file included from ./result-odb.hxx:18:
In file included from ./result-mapping-odb.hxx:23:
/usr/include/odb/traits.hxx:178:5: error: implicit instantiation of
undefined template 'odb::access::object_traits_impl<result::Result<float,
7>, 5>'
    access::object_traits_impl<T, DB>,
    ^
/usr/include/odb/database.txx:38:5: note: in instantiation of template
class 'odb::object_traits_impl<result::Result<float, 7>, 5>' requested here
    object_traits::persist (*this, obj);
    ^
/usr/include/odb/database.ixx:160:12: note: in instantiation of function
template specialization 'odb::database::persist_<result::Result<float, 7>,
5>' requested here
    return persist_<T, id_common> (obj);
           ^
main.cpp:25:9: note: in instantiation of function template specialization
'odb::database::persist<result::Result<float, 7> >' requested here
    db->persist(S);
        ^
/usr/include/odb/forward.hxx:99:11: note: template is declared here
    class object_traits_impl;
          ^
In file included from main.cpp:11:
In file included from /usr/include/odb/sqlite/database.hxx:16:
In file included from /usr/include/odb/database.hxx:632:
/usr/include/odb/database.txx:38:18: error: incomplete definition of type
'odb::object_traits_impl<result::Result<float, 7>, 5>'
    object_traits::persist (*this, obj);
    ~~~~~~~~~~~~~^~
/usr/include/odb/database.ixx:160:12: note: in instantiation of function
template specialization 'odb::database::persist_<result::Result<float, 7>,
5>' requested here
    return persist_<T, id_common> (obj);
           ^
main.cpp:25:9: note: in instantiation of function template specialization
'odb::database::persist<result::Result<float, 7> >' requested here
    db->persist(S);
        ^
In file included from main.cpp:11:
In file included from /usr/include/odb/sqlite/database.hxx:16:
In file included from /usr/include/odb/database.hxx:632:
/usr/include/odb/database.txx:40:29: error: no member named
'reference_cache_traits' in 'odb::object_traits_impl<result::Result<float,
7>, 5>'
    typename object_traits::reference_cache_traits::position_type p (
             ~~~~~~~~~~~~~~~^
3 errors generated.

Thanks,
Balazs
From abv150ci at gmail.com  Wed Mar 22 09:05:55 2017
From: abv150ci at gmail.com (=?UTF-8?Q?Aar=C3=B3n_Bueno_Villares?=)
Date: Wed Mar 22 09:06:48 2017
Subject: [odb-users] Porting ODB to MXE
Message-ID: <CAMbYJhYKKrTRJG0N=5W51nGBd-QLvH1es+NAt4reu28s=F9kyQ@mail.gmail.com>

I was trying to compile ODB for my MXE (mingw) installation for
cross-compilation (from Ubuntu to Windows), but I had issues with
libodb-mysql, because it needs libmysqlclient (I'm only compiling for
windows the static versions of the libraries), but the ./configure script
tries to find libmysqlclient_r.a and a soft-link doesn't work.

To don't repeat myself, here I have left a link with a more detailed
explanation:

https://github.com/mxe/mxe/issues/1731

Best regards,
Aar?n.
From pustovalovdmit at gmail.com  Wed Mar 22 09:53:10 2017
From: pustovalovdmit at gmail.com (=?UTF-8?B?0J/Rg9GB0YLQvtCy0LDQu9C+0LIg0JTQvNC40YLRgNC40Lk=?=)
Date: Wed Mar 22 09:53:33 2017
Subject: [odb-users] odb (2.3.0) with MySQL sql-mode NO_AUTO_VALUE_ON_ZERO
Message-ID: <CAFSDMpKQKTgx-zLPJ6rSCXsC=+AqhN=hx9eWcei=4=Hfi97maA@mail.gmail.com>

Hello!

I use MySQl with enabled sql-mode NO_AUTO_VALUE_ON_ZERO, which means that
it's not possible to use 0 in INSERT statements for PK column to indicate
MySQL that it should automatically assign value for PK.

I have persistent object of following definition:

#pragma db objectstruct person{
    #pragma db id auto
    uint64_t id_ = 0;

    std::string first_name_;
    std::string last_name_;};

And table:

CREATE TABLE `person` (`id` BIGINT UNSIGNED NOT NULL
AUTO_INCREMENT,`first_name` VARCHAR(50) NOT NULL,`last_name`
VARCHAR(50) NOT NULL,
 PRIMARY KEY(id)) ENGINE=InnoDB AUTO_INCREMENT=1 DEFAULT CHARSET=utf8
collate=utf8_unicode_ci;

I assume that 'id auto' specificator should generate following sql in
mapper classes:

INSERT INTO person (first_name, last_name) VALUES ("John", "Dohn");

or

INSERT INTO person (id, first_name, last_name) VALUES (NULL, "John", "Dohn");

but real sql is:

INSERT INTO person (id, first_name, last_name) VALUES (0, "John", "Dohn");

which leads to inserting row with PK value equal to zero and leads to
duplicate PK errors, instead of inserting row with autoincremented id.

Is there some way to force odb compliter to generate proper SQL-code
because I don't want to disable NO_AUTO_VALUE_ON_ZERO?
From boris at codesynthesis.com  Wed Mar 22 12:39:17 2017
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Wed Mar 22 12:39:26 2017
Subject: [odb-users] Persisting class instantiated from a template
In-Reply-To: <CACjO_y=+BnWnyRMPc3bM5jirjN3U-MQhjfN_pPq=1fFsVyOXWw@mail.gmail.com>
References: <boris.20170315172043@codesynthesis.com>
	<CACjO_ym+kqhVdHb+mKnuekGFgT0TyZ2dJsu7vWpn8XBianF0cg@mail.gmail.com>
	<CACjO_ykjGxHNmsyPZEOZc9Lr1rKi4rZKxa=Pj7svkNGqYC1eqg@mail.gmail.com>
	<boris.20170316183337@codesynthesis.com>
	<CACjO_ykyZjTMgMY02MgM0MFO3=m209Khtg1TGND2B2KenqBX4A@mail.gmail.com>
	<boris.20170317172206@codesynthesis.com>
	<CACjO_ynmyTzg+-uxn06PFskbyzipWFg3pbKqMuEKVtMYkpyRFQ@mail.gmail.com>
	<CACjO_ykhC7cF6t0Lo5SvUUhoODkni0R0c4VKaEmvFOHaMw-9ow@mail.gmail.com>
	<boris.20170320154436@codesynthesis.com>
	<CACjO_y=+BnWnyRMPc3bM5jirjN3U-MQhjfN_pPq=1fFsVyOXWw@mail.gmail.com>
Message-ID: <boris.20170322183734@codesynthesis.com>

Hi Bal?zs,

> #pragma db value(ResultFloatSeven) definition
>
> [...]
>
> ResultFloatSeven S(Values, Id);
>
> [...]
>
> db->persist(S);

You cannot persist a value directly, only as part of an object. Please
read the documentation.

Boris

From boris at codesynthesis.com  Wed Mar 22 12:44:39 2017
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Wed Mar 22 12:44:48 2017
Subject: [odb-users] Porting ODB to MXE
In-Reply-To: <CAMbYJhYKKrTRJG0N=5W51nGBd-QLvH1es+NAt4reu28s=F9kyQ@mail.gmail.com>
References: <CAMbYJhYKKrTRJG0N=5W51nGBd-QLvH1es+NAt4reu28s=F9kyQ@mail.gmail.com>
Message-ID: <boris.20170322184343@codesynthesis.com>

Hi Aar?n,

Aar?n Bueno Villares <abv150ci@gmail.com> writes:

> I was trying to compile ODB for my MXE (mingw) installation for
> cross-compilation (from Ubuntu to Windows), but I had issues with
> libodb-mysql, because it needs libmysqlclient (I'm only compiling for
> windows the static versions of the libraries), but the ./configure script
> tries to find libmysqlclient_r.a and a soft-link doesn't work.

Perhaps this would be of help, especially the last section:

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

Boris

From boris at codesynthesis.com  Wed Mar 22 12:58:51 2017
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Wed Mar 22 12:59:00 2017
Subject: [odb-users] odb (2.3.0) with MySQL sql-mode NO_AUTO_VALUE_ON_ZERO
In-Reply-To: <CAFSDMpKQKTgx-zLPJ6rSCXsC=+AqhN=hx9eWcei=4=Hfi97maA@mail.gmail.com>
References: <CAFSDMpKQKTgx-zLPJ6rSCXsC=+AqhN=hx9eWcei=4=Hfi97maA@mail.gmail.com>
Message-ID: <boris.20170322185628@codesynthesis.com>

Hi,

?????????? ??????? <pustovalovdmit@gmail.com> writes:

> I use MySQl with enabled sql-mode NO_AUTO_VALUE_ON_ZERO, which means that
> it's not possible to use 0 in INSERT statements for PK column to indicate
> MySQL that it should automatically assign value for PK.
> 
> Is there some way to force odb compliter to generate proper SQL-code
> because I don't want to disable NO_AUTO_VALUE_ON_ZERO?

Not at the moment. But we had a similar issue with SQLite which was
addressed by allowing NULL values for auto ids. Here is the commit:

http://scm.codesynthesis.com/?p=odb/odb.git;a=commit;h=4fd2c107242fe9f8e6ba7ded2a789f4de2ccd040

Perhaps we could do the same for MySQL?

Boris

From abv150ci at gmail.com  Wed Mar 22 16:58:10 2017
From: abv150ci at gmail.com (=?UTF-8?Q?Aar=C3=B3n_Bueno_Villares?=)
Date: Wed Mar 22 16:59:03 2017
Subject: [odb-users] Porting ODB to MXE
In-Reply-To: <boris.20170322184343@codesynthesis.com>
References: <CAMbYJhYKKrTRJG0N=5W51nGBd-QLvH1es+NAt4reu28s=F9kyQ@mail.gmail.com>
	<boris.20170322184343@codesynthesis.com>
Message-ID: <CAMbYJhaoJ0HUQDPzz8aj96-7PRg_LydZkgi-9dsRkjhQ0wvJ4w@mail.gmail.com>

That didn't help to me.

I finally solved it editing the configure script manually and compiling the
conftest.cpp samples manually to see what happened, adding the set -o
xtrace option to the configure script to track the steps and a lot of stuff
that I'm trying to forget as fast as possible.

I'm not sure what is the minimial amount of work required to fix it, but I
solve it with the following steps if I don't forget anything (sorry for my
English):

https://github.com/mxe/mxe/issues/1731

On 22 March 2017 at 17:44, Boris Kolpackov <boris@codesynthesis.com> wrote:

> Hi Aar?n,
>
> Aar?n Bueno Villares <abv150ci@gmail.com> writes:
>
> > I was trying to compile ODB for my MXE (mingw) installation for
> > cross-compilation (from Ubuntu to Windows), but I had issues with
> > libodb-mysql, because it needs libmysqlclient (I'm only compiling for
> > windows the static versions of the libraries), but the ./configure script
> > tries to find libmysqlclient_r.a and a soft-link doesn't work.
>
> Perhaps this would be of help, especially the last section:
>
> http://www.codesynthesis.com/products/odb/doc/install-mingw.xhtml
>
> Boris
>
From abv150ci at gmail.com  Wed Mar 22 17:54:18 2017
From: abv150ci at gmail.com (=?UTF-8?Q?Aar=C3=B3n_Bueno_Villares?=)
Date: Wed Mar 22 17:55:10 2017
Subject: [odb-users] Porting ODB to MXE
In-Reply-To: <CAMbYJhaoJ0HUQDPzz8aj96-7PRg_LydZkgi-9dsRkjhQ0wvJ4w@mail.gmail.com>
References: <CAMbYJhYKKrTRJG0N=5W51nGBd-QLvH1es+NAt4reu28s=F9kyQ@mail.gmail.com>
	<boris.20170322184343@codesynthesis.com>
	<CAMbYJhaoJ0HUQDPzz8aj96-7PRg_LydZkgi-9dsRkjhQ0wvJ4w@mail.gmail.com>
Message-ID: <CAMbYJhYWvzVa6RcX3Fc_d2saXcYpGOMu4oU84WhJ4XQrvu2z_Q@mail.gmail.com>

I get linking errors when compiling my app.

On 22 March 2017 at 21:58, Aar?n Bueno Villares <abv150ci@gmail.com> wrote:

> That didn't help to me.
>
> I finally solved it editing the configure script manually and compiling
> the conftest.cpp samples manually to see what happened, adding the set -o
> xtrace option to the configure script to track the steps and a lot of stuff
> that I'm trying to forget as fast as possible.
>
> I'm not sure what is the minimial amount of work required to fix it, but I
> solve it with the following steps if I don't forget anything (sorry for my
> English):
>
> https://github.com/mxe/mxe/issues/1731
>
> On 22 March 2017 at 17:44, Boris Kolpackov <boris@codesynthesis.com>
> wrote:
>
>> Hi Aar?n,
>>
>> Aar?n Bueno Villares <abv150ci@gmail.com> writes:
>>
>> > I was trying to compile ODB for my MXE (mingw) installation for
>> > cross-compilation (from Ubuntu to Windows), but I had issues with
>> > libodb-mysql, because it needs libmysqlclient (I'm only compiling for
>> > windows the static versions of the libraries), but the ./configure
>> script
>> > tries to find libmysqlclient_r.a and a soft-link doesn't work.
>>
>> Perhaps this would be of help, especially the last section:
>>
>> http://www.codesynthesis.com/products/odb/doc/install-mingw.xhtml
>>
>> Boris
>>
>
>
From obermann.lukas at gmail.com  Thu Mar 23 04:25:40 2017
From: obermann.lukas at gmail.com (Lukas Obermann)
Date: Thu Mar 23 04:25:55 2017
Subject: [odb-users] Subselects in View
In-Reply-To: <boris.20170320181701@codesynthesis.com>
References: <CC447CBF-19F1-4C94-8BFB-A543CF002DD7@gmail.com>
	<boris.20170320181701@codesynthesis.com>
Message-ID: <05A064D9-465F-4082-AA37-BCBA026229D3@gmail.com>

Hi Boris,

thank you for the help!

I think I mixed that up a bit. With ?complex? I meant building queries on-the-fly without the use of views, that use joins ans subselects.

I have two more questions regarding the native view.
How to handle if the relation from table2 to table2 is done via a lazy_shared_ptr? And, can I also reference it when the relation is done via a vector, thus having a separate table in between them.

Thanks,
Lukas



> Am 20.03.2017 um 17:21 schrieb Boris Kolpackov <boris@codesynthesis.com>:
> 
> Hi Lukas,
> 
> Lukas Obermann <obermann.lukas@gmail.com> writes:
> 
>> I know that it is not recommended to do complex queries with odb, [...]
> 
> Not recommended by whom?
> 
> You can handle pretty much any query with an ODB native view. It may not
> be as convenient as for simpler queries, but it is definitely possible
> and not "not recommended".
> 
> 
>> SELECT table1.name, (SELECT AVG(score) FROM table2 WHERE
>>  table2.t_id = table1.t_id) as score FROM table1
> 
> You can handle it pretty easily with a native view. You can even get fancy
> and do something like this:
> 
> #include <string>
> 
> #pragma db object
> struct table1
> {
>  #pragma db id
>  int t_id;
> 
>  std::string name;
> };
> 
> #pragma db object
> struct table2
> {
>  #pragma db id
>  int t_id;
> 
>  int score;
> };
> 
> #pragma db view object(table1) object(table2)                            \
>  query("SELECT " + table1::name + ", "                                  \
>        "("                                                              \
>        " SELECT AVG(" + table2::score + ")"                             \
>        " FROM " + (object_traits_impl<table2, id_common>::table_name) + \
>        " WHERE " + (table2::t_id == table1::t_id) +                     \
>        ") AS score"                                                     \
>        "FROM " + (object_traits_impl<table1, id_common>::table_name))
> struct view
> {
>  std::string name;
>  int avg_score;
> };
> 
> You can read moe about table_name here:
> 
> [1] http://www.codesynthesis.com/pipermail/odb-users/2016-April/003209.html
> 
> Boris


From pustovalovdmit at gmail.com  Thu Mar 23 11:05:11 2017
From: pustovalovdmit at gmail.com (=?UTF-8?B?0J/Rg9GB0YLQvtCy0LDQu9C+0LIg0JTQvNC40YLRgNC40Lk=?=)
Date: Thu Mar 23 11:05:35 2017
Subject: [odb-users] odb (2.3.0) with MySQL sql-mode NO_AUTO_VALUE_ON_ZERO
In-Reply-To: <boris.20170322185628@codesynthesis.com>
References: <CAFSDMpKQKTgx-zLPJ6rSCXsC=+AqhN=hx9eWcei=4=Hfi97maA@mail.gmail.com>
	<boris.20170322185628@codesynthesis.com>
Message-ID: <CAFSDMpK4avOreG__09SYY7viBjP7LazCmJ7yKb5GgvGHUD2L0A@mail.gmail.com>

Thx, Boris. This approach works.

??, 22 ???. 2017 ?. ? 19:58, Boris Kolpackov <boris@codesynthesis.com>:

> Hi,
>
> ?????????? ??????? <pustovalovdmit@gmail.com> writes:
>
> > I use MySQl with enabled sql-mode NO_AUTO_VALUE_ON_ZERO, which means that
> > it's not possible to use 0 in INSERT statements for PK column to indicate
> > MySQL that it should automatically assign value for PK.
> >
> > Is there some way to force odb compliter to generate proper SQL-code
> > because I don't want to disable NO_AUTO_VALUE_ON_ZERO?
>
> Not at the moment. But we had a similar issue with SQLite which was
> addressed by allowing NULL values for auto ids. Here is the commit:
>
>
> http://scm.codesynthesis.com/?p=odb/odb.git;a=commit;h=4fd2c107242fe9f8e6ba7ded2a789f4de2ccd040
>
> Perhaps we could do the same for MySQL?
>
> Boris
>
From Wolfgang.Neuner at swarovski.com  Fri Mar 24 02:19:35 2017
From: Wolfgang.Neuner at swarovski.com (Neuner Wolfgang)
Date: Fri Mar 24 02:20:04 2017
Subject: [odb-users] 'lazy_weak_ptr' is not a member of 'odb'
Message-ID: <AM4PR01MB18739EFFDAEB87BDE995BD0CFE3E0@AM4PR01MB1873.eurprd01.prod.exchangelabs.com>

i try to use the lazy pointers as described in 6.4.

But i get the following odb compile error:

clib.hxx:40:14: error: 'lazy_weak_ptr' is not a member of 'odb'
  std::vector<odb::lazy_weak_ptr<COdbClibRight> > m_ClibRights;
              ^
clib.hxx:40:14: error: 'lazy_weak_ptr' is not a member of 'odb'
clib.hxx:40:46: error: template argument 1 is invalid
  std::vector<odb::lazy_weak_ptr<COdbClibRight> > m_ClibRights;
                                              ^
clib.hxx:40:46: error: template argument 2 is invalid
clib.hxx:40:48: error: expected unqualified-id before '>' token
  std::vector<odb::lazy_weak_ptr<COdbClibRight> > m_ClibRights;
                                                ^
clib.hxx:52:7: error: 'lazy_shared_ptr' in namespace 'odb' does not name a template type
  odb::lazy_shared_ptr<COdbAuthMandant> m_AuthMandant;


I included the lazy-ptr header, but the odb compiler cannot find the definition of the lazy pointers.
Where is the problem?




My odb command line:
odb.exe  -I C:\Data\Entwicklung\APIs\libodb-2.4.0\ --database mysql --generate-query --generate-schema clib.hxx




My clib.hxx:
#include <odb/lazy-ptr.hxx>

class COdbClibRight;

#pragma db object table("auth_mandant")
class COdbAuthMandant
{
public:
                COdbAuthMandant() {};

                int GetId() const { return m_iId; };

                const std::string& GetName() const { return m_sName; };
                void SetName(std::string& name){ m_sName = name; };
                const std::string& GetAddress() const { return m_sAddress; };
                void SetAddress(std::string& address) { m_sAddress = address; };



private:
                #pragma db member id auto column("id") get(GetId)
                int m_iId;

                #pragma db member column("name") get(GetName) set(SetName) type("VARCHAR(50)")
                std::string m_sName;

                #pragma db member column("address") get(GetAddress) set(SetAddress) type("VARCHAR(250)")
                std::string m_sAddress;

                #pragma db member column("contact") get(GetContact) set(SetContact) type("VARCHAR(250)")
                std::string m_sContact;

                #pragma db member column("factory") get(GetFactory) set(SetFactory) type("VARCHAR(50)")
                std::string m_sFactory;

                #pragma db value_not_null inverse(m_AuthMandant)
                std::vector<odb::lazy_weak_ptr<COdbClibRight> > m_ClibRights;
};


#pragma db object table("clib_right")
class COdbClibRight
{
public:
                COdbClibRight() {};

private:
                #pragma db not_null
                odb::lazy_shared_ptr<COdbAuthMandant> m_AuthMandant;
};
From boris at codesynthesis.com  Fri Mar 24 05:58:01 2017
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Fri Mar 24 05:58:09 2017
Subject: [odb-users] 'lazy_weak_ptr' is not a member of 'odb'
In-Reply-To: <AM4PR01MB18739EFFDAEB87BDE995BD0CFE3E0@AM4PR01MB1873.eurprd01.prod.exchangelabs.com>
References: <AM4PR01MB18739EFFDAEB87BDE995BD0CFE3E0@AM4PR01MB1873.eurprd01.prod.exchangelabs.com>
Message-ID: <boris.20170324115527@codesynthesis.com>

Hi Wolfgang,

Neuner Wolfgang <Wolfgang.Neuner@swarovski.com> writes:

> clib.hxx:40:14: error: 'lazy_weak_ptr' is not a member of 'odb'
>   std::vector<odb::lazy_weak_ptr<COdbClibRight> > m_ClibRights;
>
>
> odb.exe  -I C:\Data\Entwicklung\APIs\libodb-2.4.0\ --database mysql
> --generate-query --generate-schema clib.hxx

odb::lazy_weak_ptr corresponds to std::lazy_ptr which is only
available since C++11. So you need to run the ODB compiler in
the C++11 mode (--std c++11).

Boris

From marco.craveiro at gmail.com  Sat Mar 25 15:43:24 2017
From: marco.craveiro at gmail.com (Marco Craveiro)
Date: Sat Mar 25 15:43:38 2017
Subject: [odb-users] Using prefetch with Oracle
In-Reply-To: <boris.20170301182048@codesynthesis.com>
References: <CAKRNd4x7K-WPp7C2xTtCfQ0rD4ou-4FYaxpEFeOU1vrHeWAbTA@mail.gmail.com>
	<boris.20170210180525@codesynthesis.com>
	<CAKRNd4z7h7A9Dj5UHhbofcJncCeskowGcFgsN-+hqwnF8AA62Q@mail.gmail.com>
	<20170213135652.GA22900@codesynthesis.com>
	<CAKRNd4yoBmOsTe6D_ZRm1yAYz-9g-YCk75hunfP0OnW1JzMjMA@mail.gmail.com>
	<CAKRNd4xW6SPCpD9tZ-Pbw236B8dK4D3yMfd3pzKZdNHBdRQNzw@mail.gmail.com>
	<boris.20170301182048@codesynthesis.com>
Message-ID: <CAKRNd4x8n-3HgLvCytbzFO-2gBkxoO-DY6HnurY_zBZ1zu5Nvg@mail.gmail.com>

Hi Boris,


Ok at long last I've finished my series of posts ([1], [2]); I can now
reproduce the problems I was having in production locally, so I can
finally start working on the ODB patch.

> Good stuff. I need to setup an Oracle 11 VM soon. Now I know which guide
> I will use ;-).

Actually, if you just want to build a Linux Oracle 11 VM, you don't
need to read all of that stuff :-) my recommendations:

1. Use a RPM based distribution such as CentOS or Fedora. This will
greatly simplify your life (the articles explain how to install the
RPMs on Debian).
2. Install just the Oracle Express package as per part I [3]. You need
that one package - I was confused and downloaded lots of other
packages, but actually you don't need to do all of that.
3. Run the oracle env script to ensure you have all of the directories
in the path, and set your includes and libs accordingly. It will also
place SQL Plus on the path.

That's pretty much it. Any problems let me know as it's still fresh in
my mind, so I may be able to help.

Cheers
-- 
Marco Craveiro

MD, Domain Driven Consulting
about: http://about.me/marcocraveiro
blog: http://mcraveiro.blogspot.co.uk
twitter: https://twitter.com/MarcoCraveiro

That the Ideas are themselves manifestations (of the Idea-Idea) and
that the Idea-Idea is a-kind-of Manifestation-Idea?which is a-kind-of
itself, so that the system is completely self-describing? would have
been appreciated by Plato as an extremely practical joke [Plato]. --
Alan Key

[1] http://mcraveiro.blogspot.co.uk/2017/03/nerd-food-northwind-or-using-dogen-with.html
[2] http://mcraveiro.blogspot.co.uk/2017/03/nerd-food-northwind-or-using-dogen-with_25.html
[3] http://mcraveiro.blogspot.co.uk/2017/02/nerd-food-northwind-or-using-dogen-with.html

From balazs.bayer at gmail.com  Fri Mar 24 18:27:22 2017
From: balazs.bayer at gmail.com (=?UTF-8?Q?Bal=C3=A1zs_Bayer?=)
Date: Mon Mar 27 09:19:07 2017
Subject: [odb-users] Persisting class instantiated from a template
In-Reply-To: <boris.20170322183734@codesynthesis.com>
References: <boris.20170315172043@codesynthesis.com>
	<CACjO_ym+kqhVdHb+mKnuekGFgT0TyZ2dJsu7vWpn8XBianF0cg@mail.gmail.com>
	<CACjO_ykjGxHNmsyPZEOZc9Lr1rKi4rZKxa=Pj7svkNGqYC1eqg@mail.gmail.com>
	<boris.20170316183337@codesynthesis.com>
	<CACjO_ykyZjTMgMY02MgM0MFO3=m209Khtg1TGND2B2KenqBX4A@mail.gmail.com>
	<boris.20170317172206@codesynthesis.com>
	<CACjO_ynmyTzg+-uxn06PFskbyzipWFg3pbKqMuEKVtMYkpyRFQ@mail.gmail.com>
	<CACjO_ykhC7cF6t0Lo5SvUUhoODkni0R0c4VKaEmvFOHaMw-9ow@mail.gmail.com>
	<boris.20170320154436@codesynthesis.com>
	<CACjO_y=+BnWnyRMPc3bM5jirjN3U-MQhjfN_pPq=1fFsVyOXWw@mail.gmail.com>
	<boris.20170322183734@codesynthesis.com>
Message-ID: <CACjO_y=j-s_3kzJB8RhfSMu+2KPEzHz_3wQAP7LrnoFzNPLQvw@mail.gmail.com>

Hi Boris,

with

#pragma db object(ResultFloatSeven) definition
#pragma db member(ResultFloatSeven::Id) id

it is ok now.

Many thanks for your help, and best regards,
Bal?zs



On 22 March 2017 at 17:39, Boris Kolpackov <boris@codesynthesis.com> wrote:

> Hi Bal?zs,
>
> > #pragma db value(ResultFloatSeven) definition
> >
> > [...]
> >
> > ResultFloatSeven S(Values, Id);
> >
> > [...]
> >
> > db->persist(S);
>
> You cannot persist a value directly, only as part of an object. Please
> read the documentation.
>
> Boris
>
From fanz at linux.com  Mon Mar 27 15:24:04 2017
From: fanz at linux.com (Fan Zhang)
Date: Mon Mar 27 15:24:17 2017
Subject: [odb-users] best practice to access query result across transaction
Message-ID: <A0DA4C48-6AF8-4125-A60A-5CDF9C36FC3A@linux.com>

Hi, guys,

First post here. I learned from here (http://www.codesynthesis.com/pipermail/odb-users/2015-July/002682.html <http://www.codesynthesis.com/pipermail/odb-users/2015-July/002682.html>) that the query result can not be accessed after the transaction is ended. I?m therefore copying stuff to a vector, which can be expensive. I guess odb must have some built-in support for this. Could you shed some light?

Thanks,

Fan
From boris at codesynthesis.com  Tue Mar 28 10:12:29 2017
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Tue Mar 28 10:12:38 2017
Subject: [odb-users] best practice to access query result across
	transaction
In-Reply-To: <A0DA4C48-6AF8-4125-A60A-5CDF9C36FC3A@linux.com>
References: <A0DA4C48-6AF8-4125-A60A-5CDF9C36FC3A@linux.com>
Message-ID: <boris.20170328160942@codesynthesis.com>

Hi Fan,

Fan Zhang <fanz@linux.com> writes:

> [...] that the query result can not be accessed after the transaction
> is ended. I?m therefore copying stuff to a vector, which can be expensive.
> I guess odb must have some built-in support for this.

The reason you cannot access the query result after the transaction is
because most databases stream the data as you iterate over it. Given
this understanding, I wonder what kind of "magic" you expect ODB to
provide here?

Boris

From fanz at linux.com  Tue Mar 28 10:35:27 2017
From: fanz at linux.com (Fan Zhang)
Date: Tue Mar 28 10:35:39 2017
Subject: [odb-users] best practice to access query result across
	transaction
In-Reply-To: <boris.20170328160942@codesynthesis.com>
References: <A0DA4C48-6AF8-4125-A60A-5CDF9C36FC3A@linux.com>
	<boris.20170328160942@codesynthesis.com>
Message-ID: <04BF5FA6-5795-4D58-9F4F-C2DE0471A831@linux.com>

I see. Just want to double check. Then I?ll just copy it.

Fan

> On Mar 28, 2017, at 10:12 AM, Boris Kolpackov <boris@codesynthesis.com> wrote:
> 
> Hi Fan,
> 
> Fan Zhang <fanz@linux.com> writes:
> 
>> [...] that the query result can not be accessed after the transaction
>> is ended. I?m therefore copying stuff to a vector, which can be expensive.
>> I guess odb must have some built-in support for this.
> 
> The reason you cannot access the query result after the transaction is
> because most databases stream the data as you iterate over it. Given
> this understanding, I wonder what kind of "magic" you expect ODB to
> provide here?
> 
> Boris