Hello, On Wed, Feb 19, 2003 at 12:01:30PM +0100, Daniel Rahn wrote: > > Provide for each statement different versions for each db and implement > > something like gettext for sql. Then you can define a db and the software > > automagically uses the correct sql syntax.
> > The other way would be to create data type definition schemes for > > each database. In the functions you would use no sql any more, but > > abstract commands that will be interpreted by some kind of sql parser, > > that would translate the abstract command into correct SQL for the > > specific database. uhm, I really dislike the parser idea, this would mean a lot of work, however once implemented usage would be fairly easy. I'l take a look at google later on to see if I can find something which would save us some work. What would be a really ugly hack but would be fairly easy to implement is the gettext thingy. We just need to take care that we might need several SQL statements for some rdbms in some places. We have to take care of that, so a functional approach might also do... > The system operation however should be based on a common standard, which > SQL92 is. Every database on the market today supports SQL92. not a good idea, in my company there a "big plans" (grin) for OTRS and I don't want to send several SELECT just because if would also work on some curious other databases I'll never use :-) -- Regards, Wiktor Wodecki
msg00082/pgp00000.pgp
Description: PGP signature
