I think the key thing with RCs was touched on by James - we need to be
complete bastards as to what's allowed in after RC1 otherwise every RC is
really RC1, however human nature and available time means RCN (where N>1)
gets less testing than RC1.
Can we set karma levels on the RC branch such that only a very restricted
set of people can commit, and anyone who wants to submit a fix has to argue
and explain to get it in, and anyone who wants to commit a feature can be
sent away with a flea in their ear.
Also for Windows testing it would help if someone who understands the test
system posts a step by step hand holding list of things to do to make it
work on Windows - it will then get used much more.
Also on the test suite, it always worries me that it tests the executable
rather than via the web server/browser and so must miss out on testing loads
of functionality as well as testing in the wrong environment. Would it be
possible to modify the tests so that optionally they can be run via a
browser (even if it means clicking on a load of 'next' buttons). Test output
could still be spooled to a single file so that there is a central pool of
results which can be reported back to the list.
Cheers
--
Phil Driscoll
Dial Solutions
+44 (0)113 294 5112
http://www.dialsolutions.com
http://www.dtonline.org
--
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]
