Firefox and Opera accepts string characters as in: http://www.w3.org/TR/CSS2/grammar.html string1 \"([^\n\r\f\\"]|\\{nl}|{escape})*\" string2 \'([^\n\r\f\\']|\\{nl}|{escape})*\' Basically everything except newline and starting quote.
Created attachment 112654 [details] patch
Comment on attachment 112654 [details] patch View in context: https://bugs.webkit.org/attachment.cgi?id=112654&action=review > Source/WebCore/css/tokenizer.flex:16 > +string1 \"([^\n\r\f\\"]|\\{nl}|{nonascii}|{escape})*\" > +string2 \'([^\n\r\f\\']|\\{nl}|{nonascii}|{escape})*\' You shouldn’t need {nonascii} any more since you are using a negated character class. Did you research where the set of characters in the current tokenizer came from?
> Did you research where the set of characters in the current tokenizer came from? http://trac.webkit.org/changeset/3695 (9 years ago by Hyatt, reviewed by darin/gramps)
(In reply to comment #3) > > Did you research where the set of characters in the current tokenizer came from? > > http://trac.webkit.org/changeset/3695 (9 years ago by Hyatt, reviewed by darin/gramps) Hyatt checked it in, but he didn’t write it. The giveaway is all the “konq” prefixed things. This presumably was a check-in of code developed in the KDE repository as part of KHTML, so we’d have to look there.
> Hyatt checked it in, but he didn’t write it. The giveaway is all the “konq” prefixed things. This presumably was a check-in of code developed in the KDE repository as part of KHTML, so we’d have to look there. I don't think it is possible now. https://projects.kde.org/projects/kde/kdelibs/repository/revisions/master/changes/khtml/css/tokenizer.flex The oldest entry has already contained this rule. And I couldn't find an SVN repository, probably dropped.
Created attachment 113291 [details] patch2
Comment on attachment 113291 [details] patch2 Clearing flags on attachment: 113291 Committed r99086: <http://trac.webkit.org/changeset/99086>
All reviewed patches have been landed. Closing bug.
This should have fixed some CSS 2.1 test suite tests, I presume? Did it? See bug 47141 for an incomplete list of 2.1 issues.