Closed Bug 723640 Opened 8 years ago Closed 6 years ago
Monkey: implement same Reg Exp native optimization as JM+TI
There are comments in the MRegExp as to how you can reuse the source RegExp object without cloning it in certain situations (where you understand the uses to be exactly the no-escape-permitting natives).
Do we still need this?
(In reply to Till Schneidereit [:till] from comment #1) > Do we still need this? Yes, definitely! I just noticed that this should speedup some Peacekeeper jQuery tests quite a bit.
This is more or less a port of your JM no-clone implementation to Ion. This should help quite a bit on the Peacekeeper domJQueryBasicFilters test. JM had another path to inline the object allocation in some cases, but I didn't do this yet because of RegExpShared refcounting complexity. Would it make sense to make RegExpShared a GC thing? That would make it a lot easier to inline pointers to it.
Attachment #818928 - Flags: review?(bhackett1024)
Attachment #818928 - Flags: review?(bhackett1024) → review+
Status: ASSIGNED → RESOLVED
Closed: 6 years ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla27
You need to log in before you can comment on or make changes to this bug.