warlord at MIT.EDU
Mon Aug 17 10:37:01 EDT 2009
Vladimir Bashkirtsev <vladimir at bashkirtsev.com> writes:
> Derek Atkins wrote:
>> Vladimir Bashkirtsev <vladimir at bashkirtsev.com> writes:
>>>> However, changing the architecture of GnuCash to be a pure DB app would
>>>> entail rewriting MOST of the engine. I wouldn't recommend going that
>>>> route in the short term.
>>> Well... Rewriting most of engine is definitely not something I plan. :)
>>> So I should take on board your idea to go with audit log. It should
>>> not be too hard to implement. Then use autoincrement in DB and have
>>> GnuCash to check it at regular intervals and before any operation
>>> which requires access to the data. If there's new entries in the log
>>> then just replay them to get updated. Something tells me that ability
>>> to store log records and ability to replay them back already is part
>>> of the engine.
>>> Have I missed anything? If not then I am quite excited and... (read below)
>> Well, right now we do not have an audit log, nor do we have a way
>> to replay the audit log. That would have to be developed.
>> Note that it might also be part of an "Undo" feature, which would
>> also be nice to have.
> As very quick solution for semi-multiuser solution is to have only one
> copy of GnuCash to have write access and rest just with read access
> re-reading DB each time update timestamp is changed.
UGGH... And how do you decide which copy of GnuCash is the master
(read-write) vs slave (read-only)? And what if that copy goes away?
And how do the various GnuCash copies talk to each other?
Seriously, either do multi-user right or don't do it.. Quick hacks
like these have their own sets of issues and require just as much
programming as doing it right the first time.
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
More information about the gnucash-devel