Remember that we're not intending to break Oracle support at our end.
If we recode into SQL92 when practical at our end, it would hopefully add a little motivation to use SQL92 constructs at the aD end. Since our goal is to avoid breaking modules, and to separate out PG-dependent stuff, aD at least will have the option of integrating these changes.
Ben and I were thinking we (all those working on OpenACS, not just Ben and I) could help aD out by writing a code base that's more SQL92 compliant. There's potentially more to think about than just PG/Oracle support ...
Let me point out that Bryan made the point when he met with Ben and I last month that they often write Oracle-isms simply because they weren't aware that an SQL92-compliant equivalent existed *and* was supported by Oracle. Much as he stated above.
So I'd like to take the opportunity to encourage movement in the SQL92 direction.
