On Friday 2017-12-15 16:00 -0600, Andrew McCreight wrote: > For me, a blocker to only having WPT is leak checking, both in terms of > XPCOM leak checking and LeakSanitizer. (The latter is probably going to > automatically work if you run them in ASan, but it would be good to check.) > I know there's a bug for XPCOM leak checking open already.
There are also some other pieces that our existing test harnesses (reftest and mochitest) do but the WPT harnesses don't, such as assertion checking (in a way that allows us to mark existing NS_ASSERTIONs as known failures, and get reports of unexpected changes in either direction, without losing the rest of the capability of the test). These make me somewhat hesitant to port even existing reftests that can be ported to use WPT (see bug 1263568 for previous discussion of these issues), particularly in areas where assertions have been useful at catching problems. For example, I'd be hesitant to add new reftests for css-multicol directly to a wpt directory rather than to a directory that is run by the layout/tools/reftest harness since assertion checking is particularly important for multicol tests. There are also a bunch of other features of the reftest harness documented in layout/tools/reftest/README.txt that aren't part of WPT that have been important for testing various things related to painting, layers, async pan/zoom, scrolling, printing, etc., that are things that simply can't be done in WPT. -David -- π L. David Baron http://dbaron.org/ π π’ Mozilla https://www.mozilla.org/ π Before I built a wall I'd ask to know What I was walling in or walling out, And to whom I was like to give offense. - Robert Frost, Mending Wall (1914)
signature.asc
Description: PGP signature
_______________________________________________ dev-platform mailing list dev-platform@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-platform
