-----Original Message----- From: Niko Kiirala [mailto:niko@...1267...] Sent: Thursday, October 08, 2009 10:25
Wed, 7 Oct 2009 22:36:34 +0200 J.B.C.Engelen@...1578... kirjoitti:
I have two requests:
- Are you using perceptualdiff? There are many 'new'
entries in the
list. This means that the resulting file is different from the reference when compared byte per byte. But I'd rather see
the result
of a visual comparison, using perceptualdiff [2]. perceptualdiff "runtests.py --compare-png=perceptualdiff"
Yes, I am. Without perceptualdiff, there were so many 'new' entries, that the results were mostly useless. It'd probably be good to have the rendering results available, at least for anything with 'new' as status. No need to blindly guess, what went wrong.
What does "new" mean when perceptualdiff is used? Does it mean that the result does not match the fail reference but also doesn't match the pass reference? I think it should just say fail in that case. (imho a fail reference is only useful for a developer that doesn't believe the pass reference is correct).