[rt-devel] Oracle install procedure for RT-3.0.2pre1

[Brook Schofield]

All,
	at least Jesse won't be having nightmares over potentially supporting 
another platform. The Oracle support - while largely complete from an 
infrastructure point of view - isn't useful to an enduser - not without 
a fair bit of spit and polish. Hopefully these files will form the 
basis for Oracle support within RT.

TODO:
	* Track down why creation of the System User (RT_System) attempts to 
insert NULL data into the GroupMembers table
SimpleQuery: INSERT INTO GroupMembers (GroupId, MemberId, id) VALUES 
(?, ?, ?)
NULL, 1, 1
	  but not for any other user...
	* Change the GroupMembers table schema to support NOT NULL for the 
GroupId column
	* Implement a dropdb() option for the install procedure
	* Make any other necessary changes to get 'make initialize-database' 
working
	* Test, test and test again

To accomplish this you'll need to:

* apply the patches from

[fsck.com #2431] Patch to lib/RT/Principal_Overlay.pm
	http://rt3.fsck.com/Ticket/Display.html?id=2431

[fsck.com #2432] Oracle support files for RT-3.0.2pre1
	http://rt3.fsck.com/Ticket/Display.html?id=2432

* update SearchBuilder to DBIx::SearchBuilder-0.81_04
	http://www.fsck.com/pub/rt/devel/DBIx-SearchBuilder-0.81_04.tar.gz

...and probe through the code to workout why this one weird oddity is 
happening.

Hopefully I haven't left out anything major. I'm going to holiday for a 
while and will be pretty much uncontactable. I expect you all to work 
on this quietly and hand it in your completed work to me on Tuesday May 
6th AEST (or Monday for those in - GMT ;-)

Oracle install procedure:

./configure --with-db-type=Oracle --with-db-dba=RT 
--with-db-database=RTPRD.domain.com --with-db-rt-user=RT 
--with-db-rt-pass=RT

which now detects whether a Database instance/user exists and therefore 
only needs to do the schema population. Your DBA will need to create 
you a RT user to perform the above.

NB: Terminology used in Oracle is vastly different to the terms used in 
MySQL and PostgreSQL.

Lots of holes - but hopefully by bouncing the problem around a few 
people we can come up with an optimal solution.

-Brook