Electrolysis/Meetings/2014-07-03

From MozillaWiki
Jump to: navigation, search


« previous week | index | next week »

jimm

  • I'm seeing a deadlock on Windows when trying to print a tab that is remote. (printing is covered by m2 bug 927188) Should a deadlock like this block m1?
    • yes, filed bug 1034321
  • (bug 942707) pdfjs currently under review. there's some discussion about whether the module should move to toolkit. reverting that part of the patch set will be pretty easy to so no biggie, just need to wait on consensus.
  • currently sorting out remaining issues with mstange's bug 1015721 so I can land bug 691601.
  • Interesting discussion in bug 948574 related to what restrictions we place on exposing privileged functionality via message manager in the child. (How locked down do we need to be?) Feedback welcome, I'm not familiar with the current policy.
    • bug 918680 - Tighten content processes sandbox by securing use of IPC
    • chrome should never trust data coming from content
    • data that is stored in some way should be gated on ui
    • billm: "I'm not sure we have a concrete policy besides 'be careful'. Reviewers should always be aware that the content process could be compromised."
    • filed bug 1034297 related to documenting rules for mm apis in the child for developers.

bjacob

  • Landed all of my compositor shutdown patches, project finished (finally).
    • I know it looks like i tried to fix all of the world at once, but it's not that! There practically isn't one of these 15-ish patches that I could remove and still be able to land the rest without getting backed out by sheriffs. Thus is the pain of already-in-mozilla-central intermittent/racing code.
    • Main bug was bug 774388, implementing a minimal not-intermittently-failing shutdown sequence for Compositor and ImageBridge IPDL protocols and for the Compositor thread; ended up with 11 patches there. All landed.
    • Side bug 1033358 fixed NS_IsMainThread to return correct results all the way from early initialization to end of shutdown - was previously lying when called outside of the lifetime of the nsThreadManager.
    • Side bug 1028383 used that to add crucial MOZ_ASSERT(NS_IsMainThread()) at places that were relying on that and were intermittently failing - would have saved me weeks if those asserts had been there since the beginning.
    • Side bug 1008254 was an intermittent (almost permanent) orange that turned out to be unrelated, but was looking a lot like it was related because of ensuing CompositorChild assertions, that I had to fix it before I knew that I could land without fixing it.

mconley

  • Example badges: http://people.mozilla.org/~mconley2/e10s-badges/
    • Should I just go ahead and make these?
      • Please add more badge ideas here:
        • e10s early adopter (M1 user)
        • e10s "adxopt an add-on" tester?
        • first ipdl bug
        • "Seems to understand IPC"
  • Bug 1002354 - [e10s] DOM Screen object returns bogus values in content process on Retina display
    • Coming to a conclusion, I think. I have a patch that does the job, just cleaning it up and making it build on all of the other platforms now.
    • Need to file a follow-up bug to merge PuppetScreenManager / PuppetScreen with nsScreenManagerProxy/ScreenProxy
  • Bug 1009628 - Need mozAfterRemotePaint event for remote iframes
    • De-bitrotted, and forked tabpanels for Firefox's async-y implementation
    • Clearing out ttaubert's concerns that he listed about my patch
  • Reviewing lots of billm patches for add-on shims

allison(:ally)

  • sick
  • spellchecker saga continues
  • got both r+s, while waiting for trees to re-open felipe came by with a drive by on the bandaid
  • fixed felipe's issue
  • landed
  • ......and bounced: didnt build on linux >.>
  • linux build issues fixed, but there are test failures
  • try pushes don't entirely agree
  • reftests : 2
    • reftests/spellcheck-superscript-2.html | image comparison (!=)
 skipped for b2g
    • reftests/spellcheck-contenteditable-attr.html image comparison (!=)
    • dont fail locally
  • mochitest-2

cpeterson

  • Should we retriage M1 bugs?
  • Elastic Search M1 bug count burndown is flat with ~60 open bugs: http://is.gd/8yF7rp
    • I count about 15 bugs here that are tracking or have patches
      • hmm, I think Elastic Search is recursively counting dependent bugs (which is of dubious value)
  • Estimated dev-hours left: 259 hours

mrbkap

  • Some reviews and random bugs
  • Login manager seems to have stuck
  • Some work on addons

billm

  • addon interposition being reviewed and updated based on feedback
  • sending Principals over MM
  • Reviewing dangerous uses of CPOWs?
  • Resume working on tests in ~1-2 weeks
  • Bob Owen almost has Windows sandbox ready
    • only issues are writing temp files and accessing audio devices?
    • font loading? APIs whitelisted on Windows, but not other platforms
    • working to enable sandbox with e10s (to log API warnings)

Alex

  • has patches up for review
  • starting summer intern project: make HTML parser work properly with e10s

Handyman

  • performance bug 994554 breakthrough
  • 1 Windows bug, 0 OS X bugs
  • Debugging blurry SVG
    • Bill recommends reftest analyzer.
    • For SVG questions, ask dholbert or jwatt in #layout
Retrieved from "https://wiki.mozilla.org/index.php?title=Electrolysis/Meetings/2014-07-03&oldid=1008208"