Electrolysis/Meetings/2014-08-21

« previous week | index | next week »

cpeterson

• Enabling e10s for Firefox's tenth-anniversary on November 9?
  • Gavin is reviewing M2 front-end bugs to see where Fx team can help.
    • Felipe will try to fix some bugs this week.
    • Dave Townsend will be joining Gavin's team and helping with their e10s efforts.
  • mccr8 fixing e10s leak checking
  • smaug investigating drag-and-drop
  • johns resuming work on windowed plugins (bug 669200)
  • Most add-on compat bugs have been deferred to post-M2. We may want to retriage some add-ons bugs to be fixed sooner, e.g.:
    • bug 1042195 - 1Password add-on
• I broke down the M2 milestone into sub-milestones based on bug priority on the e10s roadmap wiki:
  • M2.1, M2.2, M2.3, and M2.x
  • https://wiki.mozilla.org/Electrolysis/Roadmap

felipe

• started working on Social, but will work on scrollbar and thumbnails first

mconley

• M2.2 Bug 961362 - [e10s] HTML5 video's fullscreen feature zooms video to fill tab, not screen
  • Landed and stuck! \o/
• M2.x Bug 1050447 - Address bar lost focus after opening a new tab using Electrolysis
  • Bounced over the weekend. This was super frustrating, but finally got a fix landed. Thanks to jimm for extra testing.
• M2.1 Bug 899347 - Make click-to-play work in e10s
  • Picked this up from mbrubeck.
  • De-bitrotted, and trying to get CTP tests to pass.
• M2.1 Bug 973001 - getUserMedia doesn't work in e10s window (no permission prompt appears)
  • This looks reasonably straight-forward, I've mostly de-bitrotted billm's work, but this is on hold while I lean into bug 973001.
• mconley, bjacob and jimm are a little worried about tests...
  • We are going to reach a point where we have to support e10s-on and e10s-off. Are we going to be running the full suite for each mode? Can our infrastructure even support that?

allison

• mostly on pto this week, will miss meeting
• can now do browser/ reviews related to e10s
• Giving up on waiting for bugzilla tweaks shim, looking for a replacement good-first-shim. So far, addSystemEventListener is the leading candidate since it should be functionally the same as addEventListener which is currently shimmed
• felipe has hopped onto the social api bugs. I sent him my wip to help him along. That set of bugs is probably good for gavin's team, especially with a headstart

jimm

• non-e10s related ipc work from last week complete.
• M1 update
  • bug 691601 (e10s support for gFormSubmitObserver) - landed!
  • bug 942707 (pdf viewer doesn't work with e10s enabled) - dealing with some cpow security changes currently.
• M2, P1
  • bug 903022 (P1, Save pages to disk) - pretty much done, finishing final review changes, should land tomorrow.
  • bug 1046512 (P1, Marionette breaks in e10s enabled Nightly) - wip

billm

• working on IPC, almost done
• CPOW fixes that will break a lot of stuff
  • now behave same as chrome object wrappers
  • contentWindow only accessible from add-ons
    • browser code should call contentWindowAsCPOW
• bug 950745 - urgent message deadlock asserts
  • Some add-ons call dispatchEvent themselves, which might be dangerous.
  • moving asserts from dispatchEvent to the dangerous functions

johns

• refactoring plugins in parent process
• will post plugin patches for jimm to review widget code next week

dtownsend

• rejoining Firefox desktop team to help organize e10s efforts
• investigating bug 999239 switching between remote/non-remote (e.g. about:config) pages
  • billm says this code is scary
  • in the short term, dtownsend will just clone tab history
  • eventually run more about: pages remotely

handyman

• bug 1051842 - OSX crash with password fields. review almost done.
• bug 1008759 - M1 gfx card blacklist fallout
• resuming swap frame loader bug
• next: bug

alex_tz

• I am finally seeing results: I managed to load pages using the background thread
• I have already asked Steve Workman for feedback on my first patch.

juanb

• I have a meeting conflict this week. Will skip the call.

bjacob

• Getting 1:1 help from ehsan about printing stuff, moving forward!
• De-tangling the old code is pretty difficult - there's a lot of detangling going on. UI and back-end for printing... it's all the same thing. :(