Now WTF::StringImpl::create(const char*, unsigned) is: ALWAYS_INLINE static PassRefPtr<StringImpl> create(const char* s, unsigned length) { return create(s, length); }; which is an infinite recursion.
I think it meant to be create(reinterpret_cast<const LChar*>(s), length).
Created attachment 113016 [details] patch
Comment on attachment 113016 [details] patch View in context: https://bugs.webkit.org/attachment.cgi?id=113016&action=review > Source/JavaScriptCore/ChangeLog:5 > + It'll be nice if you explained what was wrong with the existing code and what you fixed.
In fact, I suggest you rename the bug title to something like "StringImpl::create(const char*, unsigned) calls itself"
Created attachment 113019 [details] patch v2
Comment on attachment 113019 [details] patch v2 View in context: https://bugs.webkit.org/attachment.cgi?id=113019&action=review > Source/JavaScriptCore/ChangeLog:9 > + Reviewed by Ryosuke Niwa. Ugh... this line is supposed to appear ABOVE the description.
The changes look good to me.
Is this dead code? Otherwise, it should be possible to make a regression test.
(In reply to comment #8) > Is this dead code? Otherwise, it should be possible to make a regression test. I believe this change is needed in an effort to support 8-bit string optimization.
Comment on attachment 113019 [details] patch v2 Clearing flags on attachment: 113019 Committed r98903: <http://trac.webkit.org/changeset/98903>
All reviewed patches have been landed. Closing bug.