Re: [gnome-db] patch to replace gda_connection_escape_string() with gda_connection_value_to_sql_string()

[Bas Driessen <bas driessen xobas com>]

On Sat, 2005-03-19 at 14:10 +0800, Alan Knowles wrote:
> Committed, but I buggered up the commit message. how did you get the
> head of Changlog to be used as the message?
>
> Regards
> Alan
>
> On Thu, 2005-03-17 at 16:23 +0100, Rodrigo Moya wrote:
> > On Wed, 2005-03-16 at 21:04 +0800, Alan Knowles wrote:
> > > I've finally found time to get back on this - 
> > > http://devel.akbkhome.com/svn/index.php/libgda/final.to_sql.patch.txt
> > > 
> > > Is tested against mysql, and builds against postgres. I'm not quite sure
> > > where the release/commit status is, as this breaks the API, for anyone
> > > who happened to use the old escape_string method..
> > > 
> > > let me know if it should be commited (I may have to dig up my old commit
> > > keys)
> > > 
> > yes, please commit, looks good indeed. Only commit to CVS HEAD, there we
> > can break the API.

Alan,

There is a problem with this patch in "providers/mysql/gda-mysql-recordset.c" There you call "gda_mysql_provider_value_to_sql_string" in functions "gda_mysql_recordset_append_row" and "gda_mysql_recordset_remove_row". My applications (using MySQL, PostgreSQL is still OK) that call those functions abort now with :

symbol lookup error: /opt/builds/gnome-db/cvs/live/head/libgda/lib/libgda-1.3/providers/libgda-mysql.so: undefined symbol: gda_mysql_provider_value_to_sql_string

So I made a small (ugly) hack so that it works again. From those 2 functions I call the old (restored) functions again. I have no time at the moment to properly investigate. Perhaps you can review your work and update it so it works in the record set functions as well?

Thanks,

Bas.