Rainer Müller wrote:
David Evans wrote:
With regard to QA, a major problem with MacPorts is the lack of a
separate port tree for ports under test. With something as far reaching
as gnome or kde, when a major revision occurs, it would be better to
work out all the kinks for the suite of ports in such a testing environment
without inflicting it on the MacPorts user community at large. Then,
the next step would be to establish a testing protocol and recruiting
a group of testers who are willing to work with less than perfect ports.
I still think we do not have the resources to support more than one
tree. With so many unmaintained ports, who would decide if updates
should go into stable or not?
Rainer
Yes, I agree that is already a problem. For instance, there is code in
MacPorts trunk that would be
useful in a release but it has been deferred because "it hasn't been
tested enough" but there is no
plan in place to test or any criteria to say when something is ready for
release. Thus there is no
way to advance the situation other than by ad hoc decision.
Even if we don't have a lot of testers yet, if we create a role for
them, maybe they will come. At any
rate, because individual contributors may not have hardware or other
resources to test ports in all configurations
it makes sense to make testing a group activity and come up with some
basic guidelines as to what constitutes
good enough to turn loose on the user community.
Ryan's idea to informally make test branches for specific collections of
ports while they are being upgraded
and tested as a whole is a reasonable alternative to a formal test tree
and is a whole lot better QA than just
committing the ports with minimal testing as see if anyone complains
(and they do!).
So how much testing is enough (open topic for discussion).
Dave
_______________________________________________
macports-users mailing list
[email protected]
http://lists.macosforge.org/mailman/listinfo.cgi/macports-users
