It’s more efficient for the DOM binding if DOM getter/setter functions use const AtomicString& for their results and arguments. Existing code uses String for results and const String& for arguments. The former results in unnecessary reference counting churn and re-lookup in the atomic string hash table. The latter can result in unnecessary memory allocation when the string in question is already present in the atomic string hash table.
Created attachment 29820 [details] patch with some work on this and other minor style issues
Created attachment 31632 [details] patch for part 1
Comment on attachment 31632 [details] patch for part 1 Clear review flag since this is landed and there is more work to do. (Sorry, bdash.)
I'll take care of the V8 bindings.
Created attachment 31652 [details] patch for part 2
Comment on attachment 31652 [details] patch for part 2 Clear review flag since this is landed and there is more work to do.
Created attachment 32722 [details] patch that does this for eight classes
Can you hold off landing this for a bit -- let me catch up w/V8 bindings.
Comment on attachment 32722 [details] patch that does this for eight classes LGTM :)
V8 side landed as http://trac.webkit.org/changeset/45874, so we're good to go.
Comment on attachment 32722 [details] patch that does this for eight classes Landed as http://trac.webkit.org/changeset/45893
Created attachment 34947 [details] patch that does this for four more classes
Comment on attachment 34947 [details] patch that does this for four more classes Landed as http://trac.webkit.org/changeset/47352
Created attachment 50715 [details] More of this type of thing
I don’t think we need this bug any more. We’ve done lots more of this type of work over the years.