Devan Goodwin wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > On Fri, 30 Jan 2009 15:26:36 +0100 > Jan Pazdziora <[email protected]> wrote: > > > On Fri, Jan 30, 2009 at 09:17:39AM -0500, Jeff Ortel wrote: > > > > > > Whether this is done via 1) compatibility layer or 2) migrating > > > the queries to ANSI SQL or both is another question. The advantage > > > of doing it in the compatibility layer is (maybe) lower risk when > > > running on Oracle. But, since we know there are many queries that > > > must be rewritten (for example: queries with the oracle (+) syntax > > > for outer joins) anyway, why not just do the right thing now and > > > get it over with while we're committed to spending the $$ and time > > > to do this? If satellite is going to become a multi-database > > > application, shouldn't the application code be as database agnostic > > > as possible? > > > > Agreed with the intention. But why is this part of the PostgreSQL > > effort. Shouldn't this be completely independent goal, with its own > > requirements and plan and test plan? > > I don't see why it should be separated in the first place but if it > were, it should probably be done before the PostgreSQL effort > (otherwise there'd be some wasted effort involved) implying that we (a) > need to put that on hold and (b) find people to do it. We've already got > people with a strong background in both databases and are already > prepping for a big test/QA impact, I say hit it while we're in there.
In addition, there is more risk of bugs in using a compatibility layer than in rewriting the queries to be ANSI-compliant. We have already tripped over a number of Orafce bugs, while odds are Oracle and Postgres implement ANSI just fine. -- Bruce Momjian <[email protected]> http://momjian.us EnterpriseDB http://enterprisedb.com + If your life is a hard drive, Christ can be your backup. + _______________________________________________ Spacewalk-devel mailing list [email protected] https://www.redhat.com/mailman/listinfo/spacewalk-devel
