We should convert the remaining HTMLCollections to being cached on their respective base nodes.
Created attachment 120847 [details] Patch
Comment on attachment 120847 [details] Patch Clearing flags on attachment: 120847 Committed r103883: <http://trac.webkit.org/changeset/103883>
All reviewed patches have been landed. Closing bug.
Comment on attachment 120847 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=120847&action=review > LayoutTests/ChangeLog:8 > + - Updated gc-9.html to document the new lifetime behavior of HTMLCollections. This change is not in the patch, and hasn't been landed.
(In reply to comment #4) > (From update of attachment 120847 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=120847&action=review > > > LayoutTests/ChangeLog:8 > > + - Updated gc-9.html to document the new lifetime behavior of HTMLCollections. > > This change is not in the patch, and hasn't been landed. Oops, I landed the missing gc-9.html update in <http://trac.webkit.org/changeset/103886> and forgot to note that here.
It seems like this patch improved the dom perf by about 10%: http://build.chromium.org/f/chromium/perf/mac-release-10.6/dom_perf/report.html?history=500&rev=116397 window is http://trac.webkit.org/log/?action=stop_on_copy&mode=stop_on_copy&rev=103897&stop_rev=103871&limit=100&verbose=on