Platform/2013-05-28

« previous week | index | next week »

Actions

Notices/Schedule

• FF21 post-mortem held today - https://etherpad.mozilla.org/21-0-Post-Mortem
• FF22b3 going to build today/tomorrow, released before the end of the week
  • After 6/4, no more speculative fixes. After 6/11, only new critical regressions will see fixes

Key Issues

Future of the platform meeting

As posted to dev-platform (What should be next for the platform meeting?):

The general feeling about the platform meeting seems to be:

1. it has some useful content/purpose that we want to maintain
2. it has no particular focus and is not optimized for anyone

For 1, the uses that I've heard include (please feel free to add to this list): - opportunity to find owners for stability, orange factor, other bugs - opportunity to present ideas (like moving to Git) to a chunk of engineering - like to know what other teams are up to

Other useful content that has been suggested includes: - content for hackers - engineering paper cuts

For 2, it seems clear that this meeting is broken in its current form. I have heard the following suggestions to maintain the benefits in 1 without the platform meeting: a. set up a regular newsletter or other form to push team status to engineering (easier said than done) b. set up an engineering managers meeting to handle cross team issues that require escalation like finding bug owners (requires buy in from a good number of managers)

Has the platform meeting run it's course? Is it time to retire this staple of the weekly engineering meeting schedule?

Threads

• dev-platform
  • Code Review Session (sjohnson)
• dev-webapi
  • Doc Sprint around WebAPIs on May 31st (David Bruant)
• mobile-firefox-dev
  • Using Gecko as a Widget (GeckoView) (mfinkle)
  • Feedback on Prompt.jsm (wesj)

Products/Projects

Firefox OS

Firefox Desktop

Firefox Mobile

• Async prompt service bug 870062
• Refactor to eliminate GeckoApp.mAppContext for bug 867517, Gecko-based WebView

Stability

OrangeFactor

• Past week's OrangeFactor: 5.74 (Previous Week: 6.88).
• Thanks to philor for disabling failure-prone Android tests that were causing frequent suite timeouts.
• Windows iX Slave Top Oranges
  • Win7 media test failures discussed last week still unassigned - bug 872748, bug 872760, bug 872788, bug 872996.
  • #1 top orange (bug 858377 - Intermittent test_393498.js | false == true) currently unowned. WinXP iX slaves primarily.
  • #4 top orange (bug 774689 - Intermittent 632781-verybig.html | image comparison (==), max difference: 255, number of differing pixels: 14400 ) currently unowned. WinXP iX slaves primarily.
• #7 top orange (bug 875194 - Intermittent Android test_video_crossorigin.html | uncaught exception - NS_ERROR_NOT_AVAILABLE: Component is not available at http://mochi.test:8888/tests/content/canvas/test/crossorigin/test_video_crossorigin.html:33) currently unowned.
• 21 intermittent failures fixed in the last week - List - Thanks!.

Performance

Metro

Games

Mobile Web Compat

Critsmash

MemShrink

• Brian Hackett delayed the analysis of JS scripts until they are Baseline-compiled, which reduces how much memory is used by scripts that are only run a few times. Among other things, this reduces startup memory by 14mb (10%), entirely eliminating a spike in memory usage we saw in the first 30 seconds after startup on AreWeSlimYet.
• Joe Drew and Andreas Gal landed some nice fixes to reduce image memory usage on B2G, which continues to be a concern.
• Read more about these and many other memory improvements in the MemShrink report for weeks 97-98.

Roundtable

• (armenzg) new win7, winXP test machines live in production
• tbpl database update Thursday - no downtime, but just fyi.
• hg upgrade coming 01june - brief downtime, more details in bug 741353