PostgreSQL development

[Derek Atkins]

Neil Williams <linux at codehelp.co.uk> writes:

> On Sunday 20 February 2005 4:51 am, Derek Atkins wrote:
>> "Matthew Vanecek" <mevanecek at yahoo.com> writes:
>> > Specifically, I'm working on GDA support (www.gnome-db.org), which
>> > encompasses Postgresql, mySQL, SQLite, and I believe Oracle and DB2.  So
>> > far, I've got the basics (Accounts, Transactions, Splits, KVP) loading
>> > and saving, with a fairly simple API that will make it easy to drop in
>> > other objects (e.g., I've got gncInvoice--but it won't work w/out
>> > Customers and Entries and other business stuff).  I need to put together
>> > a TODO, and hopefully in a month or so publish the code, either in
>> > gnucash or on sourceforge, so others can poke at it if they want.
>>
>> Personally I'd like to see it as a patch to the gnucash tree so we can
>> plug it in and start playing with it together.  It can still be a
>> gnc-module ;)
>
> Doesn't that require Scheme to be linked via the engine?

No, not at all.  All it requires is that the backend register itself
when it IS loaded, and QOF looks at the list of loaded backends.

>    Are we moving to the  load_library_backend() method?

Maybe, maybe not.

> in qofsession.c:
>
>   /* FIXME: this should use dlopen(), instead of guile/scheme,
>    *    to load the modules.  Right now, this requires the engine to
>    *    link to scheme, which is an obvious architecture flaw.
>    *    XXX this is fexed below, in the non-gnucash version. Cut
>    *    over at some point.
>
> I agree it's not worth changing gnc-backend-file-v2 to the new method, but I 
> thought the new backends were not going to be loaded as gnc-modules anymore?

Honestly I dont think it matters how the code is loaded into the
system.  Let's get the code into the source tree before we worry about
how the code actually gets loaded into the system, okay?  :)

In the end I DO want to have one, consistent method for loading
modules (be they QOFBackend modules, gnucash functionality modules, or
whatever).  Right now we have two methods, which is a PITA.  But
honestly I dont care at the moment how the code gets loaded.

-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