Is Bugzilla on Windows/IIS a Bad Idea?

[Max Kanat-Alexander]

On Tue, 2005-08-16 at 17:37 -0400, Habers, Mark wrote:
> I was throwing a duplicate key error at that line...and checksetup
> stopped there. That's all I can say with confidence until I try and
> repeat the install on a local machine somewhere with my old db again.
> I'll do it again this week if I can. 

	Yeah, I'd appreciate that, so that I could see the exact error.

> I seem to recall getting past the
> problem by doing that portion manually after issuing a repair table
> command... [snip]

	If REPAIR TABLE was in fact the command you used to fix it, then the
fault was not in checksetup but in the fact that your MySQL tables were
corrupted. There is certainly nothing that checksetup could have done
about that, really. :-)

> I am more familiar with SQL Server than MySQL... sorry if this is a dumb
> question, but is it just a limitation of MySQL that prevents a huge
> script like checksetup from continuing on when it runs into a
> non-showstopper kind of problem?

	If it was in fact a REPAIR TABLE that you had to issue to fix the
problem, then the problem is definitely a showstopper. There is very
little that any script or program can do with a corrupt table.

> Would it be technically possible to port this over to something like
> SQL Server or Oracle?

	Oracle: https://bugzilla.mozilla.org/show_bug.cgi?id=189947
	MS-SQL: https://bugzilla.mozilla.org/show_bug.cgi?id=285122

	If your perl skills are good, you're free to attempt to tackle or help
tackle either one of those. Instructions exist in the comments.

	Otherwise, you're also free to hire me/somebody or otherwise induce
me/somebody to get Bugzilla running on either of those database systems.
MS-SQL will probably be particularly hard, as it is based on a
very-non-ANSI-standard database. Oracle is one I'd like to do, but it's
a bit difficult to get a testing installation working on a remote server
that I have only SSH access to (since Oracle, strangely enough for an
enterprise product, requires a GUI to install in a normal fashion).

> The frustration
> factor increases big time when 1 line out of several thousand has an
> issue of some sort and the script just dies instead of executing the
> rest...

	checksetup in general is dependent upon everything working in order.
Nearly always it does. In fact, we have quite intense regression-testing
of checksetup to make sure that it can upgrade any unmodified database
from any version to our daily CVS version.

> In fact, if you guys would like some Windows specific QA, I can probably
> smk test the install of 'almost ready for production' builds for you
> from time to time. I'm probably a good guinea pig after using BZ on
> Windows way before I probably should have attempted it.

	That would be wonderful. We have QA efforts right before we release,
and you can contact lpsolit -at- gmail or join the #qa-bugzilla channel
on irc.mozilla.org to find out how you can help. :-)

	-Max
-- 
http://www.everythingsolved.com/
Competent, Friendly Bugzilla Services. And Everything Else, too.