Bug 85155 – REGRESSION(r115388): ~50% regression in DOM/Template, ~10% regression in Dromaeo/dom-traverse

• Status: RESOLVED
• Resolution: FIXED
• Component: JavaScriptCore

[Antti Koivisto]

http://trac.webkit.org/changeset/115388 seems to have had caused improbably large changes in the result of several http://webkit-perf.appspot.com/ tests

~50% regression in DOM/Template. This is visible on Lion and Qt bots (which all use JSC):

Lion: http://webkit-perf.appspot.com/graph.html#tests=[[110,2001,32196]]&sel=1335478018713.1335,1335511830796.091,0,40&displayrange=7&datatype=running
Qt: http://webkit-perf.appspot.com/graph.html#tests=[[110,2001,957069]]&sel=1335436428985.3481,1335524776685.9792,0,40&displayrange=7&datatype=running

~10% regression in Dromaeo/dom-traverse on Lion bot only:

http://webkit-perf.appspot.com/graph.html#tests=[[40020,2001,32196]]&sel=1335453053856.2793,1335524108065.345,0,3000&displayrange=7&datatype=running

On the other hand this appears to be a ~50% progression in Bindings/event-target-wrap from the same change on Lion bot only:

http://webkit-perf.appspot.com/graph.html#tests=[[8062,2001,32196]]&sel=1335464925822.3901,1335519506679.2725,0,1500&displayrange=7&datatype=running

[Radar WebKit Bug Importer]

<rdar://problem/11362533>

[Oliver Hunt]

Rolled the offending patch out:
Committed r115861: <http://trac.webkit.org/changeset/115861>

[Antti Koivisto]

Looks like we are back where we were:

http://webkit-perf.appspot.com/graph.html#tests=[[110,2001,32196]]&sel=1335972870145.3215,1336031197585.046,0,40&displayrange=7&datatype=running