|
From: Ask S. G. <cl...@sl...> - 2001-04-16 15:33:56
|
On Sun, 15 Apr 2001 20:01:49 -0500, Stephen Clouse said: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > Yee-haw, as they say down south. install-slashsite now survives without a > single error (using the slashcode theme, anyway). > > The placeholder idea I described last week is in here as well, and works as > expected. All the Slash::DB::MySQL code that hasn't been overridden for Oracle > continues to work, which makes the transition quite a bit easier. Of course, > cleaning up the MySQL code with the newly-extended functions is still in the > works. I need to actually fire things up with a MySQL server and make sure > nothing has broken horribly. I might post a patch with just the > Slash::DB::Utility changes and see if I can get some QA on it. > > The most annoying problem at the moment is dealing with the column names that > were altered due to Oracle reserved word conflicts. I'm very tempted to try > hacking this in Slash::DB::Utility (letting it surreptitiously alter the field > names passed to the SQL generator functions), because I'm having to override > entire subroutines (that would work otherwise) to change three letters. Is it > at all possible to synchronize the column names between the different ports? Or > are we too far along at this point? The changes were fairly minor (this ripped > >from the notes at the top of my schema create script): > > # - *(uid) => user_id > # - *(comment) => comment_text > # - *(mode) => comment_mode > # - comments(date) => comment_date > # - pollquestions(date) => poll_date > # - sessions(session) => session_id > > OTOH, there's no guarantee that the next database deserving of a port won't have > conflicts as well, so it's probably a moot point. I'll try to find some way of > munging the field names transparently in the sql* functions. > > This week I'll actually fire up Apache and see what falls apart -- most likely a > lot :) Even so, I'm feeling rather triumphant at the moment. You couldn't > finish installing on Oracle a week ago :) Steven, Congratulations! That's one heck of an accomplishment! My thoughts on the field names: aside from the table creation scripts (these must be made with the proper name for your backend), it shouldn't be too hard to have the Backend Specific module take key names (some methods of the Slash::DB objects refer to key names and these must be converted as well!) and field names and maps them to something appropriate to the backend. It's best to leave these on a per module basis, since not all modules will need this. Maybe we can have some facility in Slash::Utility that can handle this if necessary. I'll look into this sometime this week and see if the team can't figure out a clean way to handle this. I would appreciate a preview copy of your patch, though, so I can see what's been done so far (and fit what fixes I may need to make into the framework of your patch). Thanks a lot for the hard work! - Cliff |