Re: [gnome-db] V3 vs V4

[Phil Longstaff <plongstaff rogers com>]

Vivien Malerba wrote:
> On Sat, May 31, 2008 at 8:21 PM, Phil Longstaff <plongstaff rogers com> wrote:
>> For the past few weeks, I've ported my code from V3 to V4.  My findings:
>>
>> 1) the sqlite provider for V4 seems to be in pretty good shape.  I've
>> found and fixed a number of memory leaks, but on the whole, I can use
>> it.  I have an operation to create an index which crashes, but I don't
>> know if the problem is mine or in libgda, and need to chase it down.
>>
>> 2) the mysql provider is missing a lot.  when I look for try to update
>> the meta_store, all of the mysql routines report that they are not
>> implemented.  in addition, when I inquire whether a table exists or not,
>> it returns that it does not, so when I try to create it, it fails
>> because it already exists.  In addition, as I reported in a previous
>> posting, I need a mechanism to return the correct string dbms type based
>> upon string length.
> 
> This provider still needs some work...
> 
>> 3) I haven't tried the postgresql provider yet.
>>
> 
> This provider is as stable and correct as the Sqlite one.
> 
>> I need to make a decision soon whether to stick with V4 or revert back
>> to V3, or possibly switch to a different db backend provider.  What is
>> the future of GDA-3 vs GDA-4?  What is the expected timing for GDA-4?
>> Will there by any more GDA-3 releases while V4 is being worked on?  I
>> understand that this is a volunteer project, but can you say anything
>> about what I, a volunteer on another project, can expect?
> 
> I don't expect to make any more release of the V3. The V4 is now
> almost API stable (I just want to introduce an api for the two phase
> commits). I'll make some other releases before it is actually stable,
> though.
> I envision a stable V4 release of Libgda around the end of the summer
> (just to be n the safe side). That stable release will have database
> providers for at least SQLite, MySQL and PostgreSQL (and maybe Oracle
> if I have the time).
> 
> Also I would like to propose libgda (and maybe libgnomedb) as part of
> the Gnome platform for the next Gnome release (it will be for the
> 2.26).
> 
> Of course I'll be glad to help you with any problem you'll encounter
> with libgda.
> 
> So you can expect some support from me and a stable library by the end
> of the summer. If there is a feature you'd need before makin a stable
> release, please tell me.

Thanks Vivien.

I don't know if this is proper list etiquette or not, but I cross posted
this to the gnucash development list as well.  When I brought up the
GDA-3 vs GDA-4 issue there, there was some discussion as to how to
proceed, along with some unhappiness expressed about the state of libgda.

I've now found another way of handling the issue which led me to log a
bug against V3 (527923) which you had indicated would not be fixed for
V3.  I don't need the two-phase commit.

The feedback I got on the gnucash development list was that GDA-4 was
the better bet, and it seems as though of the big 3 (sqlite, mysql and
postgresql), sqlite and postgresql are used ahead of mysql (i.e. I know
people who want sqlite and postgresql, but a quick scan of my e-mail
didn't show anyone specifically wanting mysql).  However, I am more
familiar with mysql than postgresql and therefore would use mysql rather
than postgresql (but would also possibly go with sqlite).

OK.  GDA-4 it is.  I will do some tests with the postgresql backend
before I try to commit any gda-backend changes for GDA-4.

Phil