Re: [mantisbt-dev] Oracle DB install and support patch

[David H. <d...@hx...>]

On Thu, 2011-08-11 at 13:40 +0200, Damien Regad wrote:
> 1. Do we implement the DKuranov fixes to hopefully have Oracle working 
> without end-user tweaks in 1.2.x ? (I think it would be a good idea)

If those changes:
1) Have been tested against MySQL, PostgreSQL, etc and don't cause
breakages
2) Don't change the database schema in any way
3) Don't rewrite massive chunks of MantisBT in a way that could change
how MantisBT 1.2.x works for people (bugs or otherwise)

Then yes. Otherwise we need to look at taking those ideas and
reimplementing them in 'master', preferably using Paul's work on
implementing a new database layer to MantisBT (see the 'next' branch as
well as Paul's own 'db' branch on Github).

> 2. Is it then worth porting the changes to 1.3.x now ? (Depends whether 
> "next" replaces 1.3 or comes later, I guess)

It wouldn't help by placing these changes in 'master' at the moment. The
reason being that we'll just have to remove them prior to merging the
'next' branch (or whatever other attempt we put together for rewriting
MantisBT database support).

I think we need to focus on finishing Paul's database rewrite so that we
have a modern and robust starting point for supporting Oracle, MSSQL,
PostgreSQL, etc properly again.

I recall that Oracle imposes a number of restrictions such as maximum
field lengths... these are issues that are very hard and messy to fix
with our current database API. With a new database API though - it would
be a lot more straightforward to fix.

Regards,

David