GDA: Recent SX changes in trunk/GObject questions

[Derek Atkins]

Hi,

Quoting Phil Longstaff <plongstaff at rogers.com>:

> I've merged up-to-date trunk into the gda branch (though it's not 
> completely committed).  Scheduled transaction structure has changed.  
> In addition, account groups have been replaced by root accounts.

Yep, we know :)

> My plan is to have a book table with a record with (currently) two 
> fields.  The first is the root account guid and the second is the 
> template root account guid.  This will allow template accounts to be 
> added to the main account table and template transactions/splits to 
> be added to the main tx/split tables, since they can be traced up via 
> the account parent guids to the correct root.
>
> Comments?

You probably want more that just those entries in the book table.
Indeed, the tables should probably more closely mirror the actual
objects.

> Questions about use of GObjects in GC.  Is the plan to base backends 
> on GObjects as well?

No.  The Backends aren't objects, they're plugins.

>       Also, does it make sense to replace the whole begin edit/commit 
> edit architecture with an architecture where "object changed" events 
> are caught and acted upon by the GDA backend but not the file backend?

No, or at least, not for a LONG while.  I think we'll probably emit
changed events with commit_edit(), but I dont think the backend should
use that.

> Phil
>
> P.S. One side effect of the merge is that I am not seeing the crash I 
> used to see when saving a transaction via gda.  The crash was 
> happening because splits were being sorted, which was triggering more 
> db access, and eventually he whole mess got corrupted.

Bugs gets fixed.  Imagine that!  ;)

-derek

-- 
       Derek Atkins, SB '93 MIT EE, SM '95 MIT Media Laboratory
       Member, MIT Student Information Processing Board  (SIPB)
       URL: http://web.mit.edu/warlord/    PP-ASEL-IA     N1NWH
       warlord at MIT.EDU                        PGP key available