GDA: problem when modifying a transaction split

[Derek Atkins]

Phil Longstaff <plongstaff at rogers.com> writes:

>> The Splits, however, ARE marked dirty...
>
> Yes, but the only commit received by the backend is for the transaction.

Yes, because Splits aren't first-class objects.  You can't have a
Split without a Transaction....  So historically you commit the Txn
and it automatically handles all the splits, too.

> I have special-cased this for now so that if a clean transaction is
> commited, the splits are checked and any dirty ones are committed.
> However, see my other e-mail re a deleted split.  When I receive the
> commit request, deleted splits are not children of the transaction any
> longer, so I have no way of knowing they should be deleted from the db.
> I could change my special-case so that all transaction commits occur
> whether the transaction is clean and dirty since a transaction commit
> involves deleting all splits for the tx in the db and then saving them
> again (+ all slots...).  I've been trying to optimize db access as much
> as possible, but this just expands it.
>
> An alternative is change how the register commits changes but this gets
> into completely new areas of the code.

As I said, it has nothing to do with the register, but the engine.

> Phil

-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