Updates:
Owner: [email protected]
Comment #3 on issue 29674 by [email protected]: ScriptValue GetString
API changed upstream, breaking lots of LayoutTests
http://code.google.com/p/chromium/issues/detail?id=29674
It turns out that this change does not break any layout tests, and that
the "hack" fix does work. I just did a run of
layout tests on Linux of r51801 (http://trac.webkit.org/changeset/51801)
with the patches from r51820 and r51821
(http://trac.webkit.org/changeset/51820 &
http://trac.webkit.org/changeset/51821), and the result is exactly the same
as for r51800 (http://trac.webkit.org/changeset/51800).
The V8 build was broken by r51801 (http://trac.webkit.org/changeset/51801).
The change r51800
(http://trac.webkit.org/changeset/51800) just before enabled SVG filters by
default and changed test expectations for a
large number of SVG tests.
As far as I can see the "hacked" change does the right thing. The reason
for passing the ScriptState (ExecState) to the
getString of a ScriptValue is that JSC added what is called ropes in r51671
(https://bugs.webkit.org/show_bug.cgi?
id=32136). Ropes are trees of concattenated strings (like ConsString in V8)
and they can cause allocation to happen
when accessing the string as it is generated in its flat form at that
point. The ScriptState (ExecState) passed is used
to signal an out of memory exception if that happens when trying to access
the string. For the V8 bindings we don't
need to pass an explicit environment in cases where an out of memory can
happen.
On the Windows webkit.org canary there are 22 failing tests after
the "hack" went in (r51821). Just before the change
which broke compilation of the V8 bindings there where 20 failing tests
(r51800). These 20 are included in the 22 so
two additional failures appeared while the build was broken.
The build just before the change which broke compilation is 17177
(http://build.chromium.org/buildbot/waterfall.fyi/builders/Webkit%20(webkit.org)/builds/17177)
webkit_tests
366 fixable
(72 skipped)
failed 20
big-sized-filter-2.svg
big-sized-filter.svg
feGaussianBlur.svg
feMerge.svg
feOffset.svg
filterRes.svg
shadow-on-filter.svg
subRegion-in-userSpace.svg
empty-merge.svg
feDisplacementMap-01.svg
js-svg-constructors.svg
svg-features.html
visibility-override-filter.svg
operatorAttribute.html
feDisplacementMap.svg
feMerge-wrong-input.svg
feTile.svg
shadow-on-rect-with-filter.svg
subRegion-one-effect.svg
subRegion-two-effects.svg
The build just after the "hack" is 17207
(http://build.chromium.org/buildbot/waterfall.fyi/builders/Webkit%20(webkit.org)/builds/17207)
webkit_tests
369 fixable
(72 skipped)
failed 22
* block-selection-gap-in-composited-layer.html
big-sized-filter-2.svg
big-sized-filter.svg
feGaussianBlur.svg
feMerge.svg
feOffset.svg
filterRes.svg
shadow-on-filter.svg
subRegion-in-userSpace.svg
* block-selection-gap-stale-cache.html
empty-merge.svg
feDisplacementMap-01.svg
js-svg-constructors.svg
svg-features.html
visibility-override-filter.svg
operatorAttribute.html
feDisplacementMap.svg
feMerge-wrong-input.svg
feTile.svg
shadow-on-rect-with-filter.svg
subRegion-one-effect.svg
subRegion-two-effects.svg
unexpected flaky 3
about-blank-iframe-in-main-frame.html
data-url-script-in-iframe.html
detaching-frame-2.html
On the Linux webkit.org canary these are the builds in question
http://build.chromium.org/buildbot/waterfall.fyi/builders/Webkit%20Linux%20(webkit.org)/builds/12709
http://build.chromium.org/buildbot/waterfall.fyi/builders/Webkit%20Linux%20(webkit.org)/builds/12742
A few changes later there was a huge jump in failures on Linux from 21 to
478.
http://build.chromium.org/buildbot/waterfall.fyi/builders/Webkit%20Linux%20(webkit.org)/builds/12744
http://build.chromium.org/buildbot/waterfall.fyi/builders/Webkit%20Linux%20(webkit.org)/builds/12745
which seemed to be related to changes in
WebKit/WebKit/chromium/src/WebViewImpl.*.
On the Mac build webkit.org canary it is
http://build.chromium.org/buildbot/waterfall.fyi/builders/Webkit%20Mac%20(webkit.org)/builds/11026
http://build.chromium.org/buildbot/waterfall.fyi/builders/Webkit%20Mac%20(webkit.org)/builds/11056
On the Mac builder r51800 did not get its own build but was combined with
r51801.
--
You received this message because you are listed in the owner
or CC fields of this issue, or because you starred this issue.
You may adjust your issue notification preferences at:
http://code.google.com/hosting/settings
--
Automated mail from issue updates at http://crbug.com/
Subscription options: http://groups.google.com/group/chromium-bugs
