- Update on current leakage numbers.
- Update from peterv.
- Status of pending CC code in Bugzilla.
- Status update from sicking regarding review of peterv's patch.
- Update from dbaron on debugging tools.
- Update from jst on xpconnect mysteries.
- Discuss next steps.
- Tinderbox didn't show that much of an improvement; however, our builds are showing improvements.
- There are conflicting numbers.
- SeaMonkey shows a drop, peterv's local shows a drop, but the Firefox build doesn't show improvements.
Jonas: I will take the XBL patch and see if that has an impact.
dbaron: Worried about a situation with the current code base where he sees his CPU pegged for several seconds before responsiveness returns. It was suggested that we start recording CC times.
Status of Pending CC Code in Bugzilla
- Add cycle collection to RDF datasources
- Make XPConnect traverse more JS edges
- Support cycle collection of refcounted non-xpcom objects (aka, the XBL patch)
- Make cycle collection suspect all native wrapper roots
- cycle collector shutdown dump based on known should-have-been-freed objects
- Johnny will add CC timing printfs:
- Cycle collection graph call times.
- JS GC times.
- dbaron is going to land a patch related to debugging (bug 372766 above).
- Sicking is going to drive the XML patch in as well as the patch from peterv (adding more content objects).
- jst will continue with xpconnect.
- Graydon will land bug 368869.
peterv suggestion: We should try to get to a point were we can start and stop without leaking.