790858 - Provide logging to allow developers to more easily diagnose perf issues with restyle/reflow

Status: RESOLVED WORKSFORME

(Core :: Layout, defect)

Description

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

See bug 780341 comment 33 and context

(In reply to Boris Zbarsky (:bz) from comment #31)
> > It might even make sense to share the same "performance logging" pref.
> 
> Sure.  It would be simple to add a static pref cache and just check that
> boolean when posting a restyle.  If set, dump out that we're asking for a
> restyle for an element with certain id/class/tag stuff, and what sort of
> restyle we're asking for ("self", "self and descendants", "self and
> descendants and all following siblings and their descendants" being the
> three kinds we have around)...  Would that be useful?

Sure.  The goal here would be for developers to see that their code is slow and then gather some information to bring to someone else, or locally fix if the bug is obvious.  In the example of this test, I was actually surprised that changing the transform triggered restyling --- it certainly makes sense, but that's not the first or even tenth thing that came to my mind.

Comment 1

[Boris Zbarsky [:bzbarsky]]

Note to self: http://mxr.mozilla.org/mozilla-central/source/layout/style/AnimationCommon.cpp#236 is an existing example of this stuff.

Comment 2

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

Important use case for content; takes some gecko expertise to do this currently.  Helps us help others help themselves.

Comment 3

[Gregory Pappas [:gregp]]

Firefox profiler does this