Speed up SunSpider by 2.2% by checks for isOutOfMemory() See attached patch.
Created attachment 16831 [details] fix
Actually now SunSpider is telling me it's 2.4% faster. :)
Comment on attachment 16831 [details] fix r=me
Comment on attachment 16831 [details] fix This version of the patch isn't quite right. For one thing, once memoryFull is set true, this patch will *never* set it back to false. That was not the old behavior. For another, we hit an assertion immediately in the debug build because operationInProgress is still true.
We *really* want this fix, but I ran out of time this morning trying to fix the two tiny problems with it and land it.
Created attachment 16841 [details] better fix (should fix both issues raised by darin)
Comment on attachment 16841 [details] better fix (should fix both issues raised by darin) r=me
r27001