Created attachment 67087 [details]
(In reply to comment #2)
IMHO String::fromUTF8 avoids the TextEncoding overhead?
I also want to replace most of the toString calls with a new toAtomicString method. QualifiedName (used for elements and attributes) takes AtomicString as arguments. Creating AtomicString instead of String can avoid many memcpy operations IMHO.
If we implement a AtomicString::fromUTF8 which works directly with the utf8 data for hash and compare we can avoid many convertUTF8ToUTF16 operations too.
(In reply to comment #3)
> IMHO String::fromUTF8 avoids the TextEncoding overhead?
OK. Next time you can just say so in the code or change log, and save a round of questioning ;-)
Comment on attachment 67087 [details]
Clearing flags on attachment: 67087
Committed r67214: <http://trac.webkit.org/changeset/67214>
All reviewed patches have been landed. Closing bug.
Created attachment 74475 [details]
Missed the second toString in the first patch.
The commit-queue encountered the following flaky tests while processing attachment 74475 [details]:
Please file bugs against the tests. These tests were authored by email@example.com and firstname.lastname@example.org. The commit-queue is continuing to process your patch.
Comment on attachment 74475 [details]
Clearing flags on attachment: 74475
Committed r72476: <http://trac.webkit.org/changeset/72476>