At 09:39 AM 5/2/2001 -0400, Adam Trachtenberg wrote:
>On Wed, 2 May 2001, Andi Gutmans wrote:
>
> > The COM problem would have been found IMO if we had released a bigger RC.
>
>I think the COM problem would have been found if somebody ran the test
>suite immediately before releasing 4.0.5 final. I think modifying the RC
>process to ensure that the last thing we do before releasing is to make
>sure the release passes the validation suite is a simpler step than
>opening the QA process to thousands of less qualified testers.
The COM isn't in the test suite because it only runs on Windows. In any
case, the test suite doesn't catch most problems. It is often tiny bugs
which only happen under very certain setups.
>I know the QAT does its best to build and validate the RCs on as many
>platforms and configurations as possible. But, always adding in new
>features, to go along with bug fixes, makes it harder to gather all the
>testers together to review each RC. If we had less RCs or the RCs modified
>fewer files, it'd be easier to ensure like mistakes like COM don't slip
>through the cracks.
>
>We always have minor buglets in the RCs. It's much easier for the QAT to
>locate them internally than to wade through all the excess bug reports
>that will come in. Mozilla makes their daily builds easily available,
>which is great. OTOH, one of their biggest QA problems is getting people
>to handle the huge amount of bug reports -- many of which are dups,
>missing info, or bogus.
Well we can experiment with RC1 of 4.0.6 and see what happens.
I think we can only gain from doing this. If there are bug reports that
come in which aren't relevant they would come in after the 4.0.6 release
anyway.
Andi
--
PHP Development Mailing List <http://www.php.net/>
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
To contact the list administrators, e-mail: [EMAIL PROTECTED]
