mysqlcppapi-main

[Mysqlcppapi-main] a patch - where to send?

[Radoslaw G. <rga...@ay...>]

Hello,

 I am the newbe with using the sourceforge so sorry for my lack of knowledg=
e.

 I have been using the mysqlcppapi library successfully for some time, so a=
t=20
first I would like to thank all maintainers and developers for the good wor=
k=20
they have done.

 I did some slight changes to the package and I would like to ask where she=
ll=20
I send the patch? Is the sourceforge project's <patches> link the right pla=
ce=20
to do it?

 The changes are quite minor - I just got rid of few warnings during the=20
compilation (defined but not used methods' parameters) and corrected=20
"mysql_shutdown" invocation to match the new MySQL interface (4.1.10) - but=
=20
without preprocessor directives to check the MySQL version (sorry).=20
=20
 I would like to ask, if it was possible, for the more complex exceptions=20
scheme - for now I have to check the exception's description to check the=20
reason (e.g. all "artificial deadlocks" of MySQL which, as they suggest,=20
should cause transactions retrial).

 Thanks in advance for any suggestions and help about my question and the=20
changes I made as well.
=2D-=20

    Mit freundlichen Gr=FC=DFen
    Best regards
   =20
    //	Radoslaw Garbacz
    //  aycan Digitalsysteme GmbH
    //  Innere Aumuehlstrasse 5
    //  97076 Wuerzburg
    //  Germany
    //  phone: +49 931.270 40 90
    //  fax:   +49 931.270 40 91
    //  mailto:rga...@ay...
    //  http://www.aycan.de

Re: [Mysqlcppapi-main] a patch - where to send?

[Jonathan W. <co...@co...>]

On Tue, Mar 08, 2005 at 01:48:04PM +0100, Radoslaw Garbacz wrote:

> Hello,

Hi,

>  I am the newbe with using the sourceforge so sorry for my lack of knowledge.
> 
>  I have been using the mysqlcppapi library successfully for some time, so at 
> first I would like to thank all maintainers and developers for the good work 
> they have done.

Thanks for the kind words (I am the current maintainer but can't take
credit for much of the library myself).

>  I did some slight changes to the package and I would like to ask where shell 
> I send the patch? Is the sourceforge project's <patches> link the right place 
> to do it?

Yes, or sending it to this list might get a quicker response.

>  The changes are quite minor - I just got rid of few warnings during the 
> compilation (defined but not used methods' parameters) and corrected 
> "mysql_shutdown" invocation to match the new MySQL interface (4.1.10) - but 
> without preprocessor directives to check the MySQL version (sorry). 

Please check the latest version from CVS, at least the mysql_shutdown
change is already done.  I *badly* need to make a new release incorporating
those changes.

>  I would like to ask, if it was possible, for the more complex exceptions 
> scheme - for now I have to check the exception's description to check the 
> reason (e.g. all "artificial deadlocks" of MySQL which, as they suggest, 
> should cause transactions retrial).

That's a good idea, feel free to post suggestions to this list.

jon

-- 
"You will find that the state is the kind of organization which,
 though it does big things badly, does small things badly too."
	- John Kenneth Galbraith

Re: [Mysqlcppapi-main] a patch - where to send?

[Radoslaw G. <rga...@ay...>]

Hi,

 Thank you for the quick response.

On Tuesday 08 March 2005 14:19, you wrote:
> Yes, or sending it to this list might get a quicker response.

 Considering the fact that "mysql_shutdown" is already done there is almost=
=20
nothing to change, just the file: mysqlcppapi/ColData.h
@@ -221,7 +221,7 @@
=20
=20
 template <class T_Str> template<class Type>=20
=2DType ColData_Generic<T_Str>::conv(Type dummy) const
+Type ColData_Generic<T_Str>::conv(Type /*dummy*/) const
 {
   std::string strbuf(buf);
   strip_all_blanks(strbuf);

> >  I would like to ask, if it was possible, for the more complex exceptio=
ns
> > scheme - for now I have to check the exception's description to check t=
he
> > reason (e.g. all "artificial deadlocks" of MySQL which, as they suggest,
> > should cause transactions retrial).
>
> That's a good idea, feel free to post suggestions to this list.

 The failure reasons I have to check by comparing strings:
 1.	mysqlcppapi::ex_BadQuery.what() =3D=3D "Duplicate entry"
	the_proposal: ex_DuplicateEntry

 2.	mysqlcppapi::ex_BadQuery.what() =3D=3D "mysql_real_connect() failed"
	the_proposal:ex_ConnectionFailure

 3.	mysqlcppapi::ex_BadQuery.what() =3D=3D ".*try restarting transaction.*"
	The suggestion to restart the transaction can happen in various situations=
, =20
so I check the existence of this phrase inside the exception description.
	the_proposal:	ex_InternalTimeout | ex_TransRestart
					e.g. ex_InternalDeadLock: public ex_InternalTimeout

 I also added files such as: AUTHOR, COPYING, etc. to the set of files bein=
g=20
installed - it is prepared for SuSE like directories scheme (packages'=20
documentation is placed in "/usr/share/doc/packages/<package>") - to make i=
t=20
in a traditional way ("/usr/share/package") just the line:
pkgdatadir              =3D $(datadir)/doc/packages/@PACKAGE@
has to be removed from the main Makefile.am

 I have the SPEC to build RPMs so if you were interested you can use it=20
(please check whether there is no LGPL license violation).

Attached:=20
=2D the patch against the version 1.9.3
=2D the spec file

=2D-=20

    Mit freundlichen Gr=FC=DFen
    Best regards
   =20
    //	Radoslaw Garbacz
    //  aycan Digitalsysteme GmbH
    //  Innere Aumuehlstrasse 5
    //  97076 Wuerzburg
    //  Germany
    //  phone: +49 931.270 40 90
    //  fax:   +49 931.270 40 91
    //  mailto:rga...@ay...
    //  http://www.aycan.de

Re: [Mysqlcppapi-main] a patch - where to send?

[Jonathan W. <co...@co...>]

On Thu, Mar 10, 2005 at 10:34:48AM +0100, Radoslaw Garbacz wrote:

> Hi,
> 
>  Thank you for the quick response.

Hi Radoslaw,

> On Tuesday 08 March 2005 14:19, you wrote:
> > Yes, or sending it to this list might get a quicker response.
> 
>  Considering the fact that "mysql_shutdown" is already done there is almost 
> nothing to change, just the file: mysqlcppapi/ColData.h
> @@ -221,7 +221,7 @@
>  
>  
>  template <class T_Str> template<class Type> 
> -Type ColData_Generic<T_Str>::conv(Type dummy) const
> +Type ColData_Generic<T_Str>::conv(Type /*dummy*/) const
>  {
>    std::string strbuf(buf);
>    strip_all_blanks(strbuf);

That change is also in CVS already.  I'll try to make a release this weekend.

> > >  I would like to ask, if it was possible, for the more complex exceptions
> > > scheme - for now I have to check the exception's description to check the
> > > reason (e.g. all "artificial deadlocks" of MySQL which, as they suggest,
> > > should cause transactions retrial).
> >
> > That's a good idea, feel free to post suggestions to this list.
> 
>  The failure reasons I have to check by comparing strings:
>  1.	mysqlcppapi::ex_BadQuery.what() == "Duplicate entry"
> 	the_proposal: ex_DuplicateEntry
> 
>  2.	mysqlcppapi::ex_BadQuery.what() == "mysql_real_connect() failed"
> 	the_proposal:ex_ConnectionFailure
> 
>  3.	mysqlcppapi::ex_BadQuery.what() == ".*try restarting transaction.*"
> 	The suggestion to restart the transaction can happen in various situations,  
> so I check the existence of this phrase inside the exception description.
> 	the_proposal:	ex_InternalTimeout | ex_TransRestart
> 					e.g. ex_InternalDeadLock: public ex_InternalTimeout

Another, simpler alternative would be to allow you to check an error code.
That would mean you don't need to match strings but we don't have to define
a whole hierarchy of classes.  I'll have a think about it when I get a
chance at the weekend.

>  I also added files such as: AUTHOR, COPYING, etc. to the set of files being 

btw I plan to rename COPYING to COPYING.LIB since the file is a copy of
the LGPL.

> installed - it is prepared for SuSE like directories scheme (packages' 
> documentation is placed in "/usr/share/doc/packages/<package>") - to make it 
> in a traditional way ("/usr/share/package") just the line:
> pkgdatadir              = $(datadir)/doc/packages/@PACKAGE@
> has to be removed from the main Makefile.am

OK, I might ask you more about this when I prepare a release.

>  I have the SPEC to build RPMs so if you were interested you can use it 
> (please check whether there is no LGPL license violation).

Thanks, there are a couple of things I'd change.  The library is LGPL
licensed, but the spec file says GPL.  I would also change the
description to make it clear mysqlcppapi is an incompatible fork, not
just a modified version of mysql++.

Thanks again for your suggestions, I'll be in touch again soon

jon

-- 
"Fanaticism consists in redoubling your efforts when you have forgotten 
 your aims."
	- George Santayana

Re: [Mysqlcppapi-main] a patch - where to send?

[Radoslaw G. <rga...@ay...>]

On Thursday 10 March 2005 11:10, Jonathan Wakely wrote:
> That change is also in CVS already.  I'll try to make a release this
> weekend.

 I see that everything has already been done :-)
	 - sorry for not checking the CVS.

> Another, simpler alternative would be to allow you to check an error code.
> That would mean you don't need to match strings but we don't have to defi=
ne
> a whole hierarchy of classes.  I'll have a think about it when I get a
> chance at the weekend.

 Yes, this is the best solution, and it would be enough for me.

 Thank you for your help.
=2D-=20

    Mit freundlichen Gr=FC=DFen
    Best regards
   =20
    //	Radoslaw Garbacz
    //  aycan Digitalsysteme GmbH
    //  Innere Aumuehlstrasse 5
    //  97076 Wuerzburg
    //  Germany
    //  phone: +49 931.270 40 90
    //  fax:   +49 931.270 40 91
    //  mailto:rga...@ay...
    //  http://www.aycan.de

Re: [Mysqlcppapi-main] a patch - where to send?

[Jonathan W. <co...@co...>]

On Thu, Mar 10, 2005 at 11:55:07AM +0100, Radoslaw Garbacz wrote:

> On Thursday 10 March 2005 11:10, Jonathan Wakely wrote:
> > That change is also in CVS already.  I'll try to make a release this
> > weekend.
> 
>  I see that everything has already been done :-)
> 	 - sorry for not checking the CVS.

No, it's my fault for not releasing the improvements properly.

> > Another, simpler alternative would be to allow you to check an error code.
> > That would mean you don't need to match strings but we don't have to define
> > a whole hierarchy of classes.  I'll have a think about it when I get a
> > chance at the weekend.
> 
>  Yes, this is the best solution, and it would be enough for me.

OK, I'll do something along these lines.

jon