Performance/Snappy/2012-08-09
From MozillaWiki
< Performance | Snappy
Contents
- 1 Snappy Aug 9 Minutes
- 1.1 Actions
- 1.2 Incoming
- 1.3 Projects
- 1.3.1 Various frontend bugs (ttaubert)
- 1.3.2 Tab switch performance - jrmuizel
- 1.3.3 Better DOM event/task scheduling - nathan (telemetry)
- 1.3.4 Front-end - Gavin, Jared and Frank (Australis tab strip)
- 1.3.5 Fix cache - hurley, bbondy
- 1.3.6 Profiler - jrmuizel/BenWa/Ehsan (and more)
- 1.3.7 Peptest - mcote, ahal
- 1.3.8 GC pause reduction - billm, jcoppeard
- 1.3.9 Move DOM Storage off main thread - vlad
Snappy Aug 9 Minutes
Actions
No action items
Incoming
- We should look at the start-up regression in bug 778718
- We should block obviously crap add-ons ala bug 777266
- Asa is already looking into this
- Q3 goal for Frontend
- [action] Gavin will look to add a goal
- mozcamp Sept. 8-9, Snappy work week Sept 10-15 (taras to send agenda to gavin)
- bug 774949 - Do not validate JARs by default, provide async APIs to do so
Projects
Various frontend bugs (ttaubert)
- Results from the past week
- Landed bug 728294 - Analyze cycle collection logs after test runs to detect leaks
- Landed bug 779618 - [Page Thumbnails] make tabbrowser use the thumbnail service
- Ready to land bug 753448 - [New Tab Page] preload newtab pages in the background and swap them in when opening a new tab (will be landed pref'ed off because it regresses a couple of talos tests)
- Filed bug 779960 - Don't wait for the mouse pointer to rest before revealing the statuspanel when hovering a link (and attached a patch, still undecided if we want this)
- Todo this week
- bug 780123 - [New Tab Page] Use the hiddenWindow to preload a single newtab page that then serves multiple windows
Tab switch performance - jrmuizel
- Results from the past week
- Filed bug 781588
- Spent a bit of looking/thinking about one of the bottlenecks in the cairo gdi backend that is a good chunk of painting time (bug 781591)
Better DOM event/task scheduling - nathan (telemetry)
- Results from the past week
- Discussion about posted patches, looks like there's still some work to do. No official r+/-, though.
- Todo this week
- More work.
Front-end - Gavin, Jared and Frank (Australis tab strip)
- Todo this week
- Neil is going to step back from the few specific bugs he was looking into and investigate layout flushing during tab switches in general
Fix cache - hurley, bbondy
- Results from the past week
- No update, pending review for the past week: bug 777328 - Gather telemetry data for how much cache corruption reduction plan would help
- bug 731307 Patrick working on main thread IO related to proxies.
Profiler - jrmuizel/BenWa/Ehsan (and more)
- Results from the past week
- Progress on using the debug protocol for profiling
- Have a patch for adding the img url to profiles
- A bunch of Boot2Gecko work
Peptest - mcote, ahal
- Results from the past week
- various mozbase bugs that were causing bustage now fixed
- peptest now unhidden by default in tbpl for all platforms
- mozharness patch which allows peptest to use content heavy websites reviewed and awaiting landing
- Todo this week
- land mozharness patch (waiting on aki)
- land tests that make use of the content heavy sites
GC pause reduction - billm, jcoppeard
- Results from the past week
- bug 743112 - Incrementally release C++ held by dead JS. Usually fairly short (10ms or so?), but when tearing down pages it can take quite a while.
Move DOM Storage off main thread - vlad
- Results from the past week
- Tested out my changes: benefits from writing out DB changes asynchronously & prefetching data at page load
- Todo this week
- Test out trickier edge cases (max data stored in DB, flushing out data on shutdown, etc)