>>>>> "Mark" == Mark Wielaard <[EMAIL PROTECTED]> writes:

Mark> Yes that would be very nice. This is really one of the issues that hold
Mark> up the creation of the release branch. I want to make sure that we don't
Mark> have (bad) mauve regressions. But I didn't have a full comparison
Mark> between a 0.91 run and a current CVS run. I do manual runs now and I am
Mark> going through them slowly (as you can see from these emails). But it is
Mark> a bit hard to keep up with you people committing 10 till 20 patches a
Mark> day :)

Andrew Hughes set this up on builder but when I look now I see that
the baseline files are empty.  It sort of doesn't matter since you
would have to re-run Mauve anyway, given that the test names have all
changed.

If you had a build of 0.91 + the corresponding jamvm you could make a
Mauve baseline and run a comparison quite easily.  Ideally we would do
this after each release and simply reset the baseline file on
builder.  Then we could see regression reports versus the previous
release automatically.

Tom

Reply via email to