mozilla :: #developers

27 Apr 2017
00:07pulsebotCheck-in: - Gerald Squelart - Bug 1337062 - Transfer initial gfxVars with SendSetXPCOMProcessAttributes - r=dvander
00:08dvandergerald: did you want me to poke at that DeviceInitData talos regression?
00:09geralddvander: Any help is welcome. I was going to work on it this afternoon (noon here)
00:10dvandergerald: okay, ill do a try push with that logging to see
00:14pulsebotCheck-in: - Wes Kocher - Bug 1341703 - Re-skip a frequently failing reftest on stylo a=bustage
00:20pulsebotCheck-in: - Wes Kocher - Backed out changeset 8d6b492ebb36 (bug 1348751) for browser_all_files_referenced.js failures a=backout
00:27pulsebotCheck-in: - Daosheng Mu - Bug 1355648 - Part 1: Check VR controllers if losing tracking; r=kip
00:27pulsebotCheck-in: - Daosheng Mu - Bug 1355648 - Part 2: Handle GampadPose losing tracking situation; r=kip,qdot
00:27pulsebotCheck-in: - Daosheng Mu - Bug 1355648 - Part 3: Set position and orientation IsValid properties for GamepadPose tests; r=kip
00:28pulsebotCheck-in: - Daosheng Mu - Bug 1359670 - Notify GamepadManager when controllers are removed in VRPuppet manager; r=kip
00:28pulsebotCheck-in: - Mantaroh Yoshinaga - Bug 1264125 part 10 - Remove unnecessary whitespace and align the colum of variable definition on ComputedTiming.h. r=whitespace-only DONTBUILD
00:30pulsebotCheck-in: - 52 changesets - Merge inbound to central, a=merge CLOSED TREE
00:35pulsebotCheck-in: - 72 changesets - Merge m-c to inbound, a=merge
00:44pulsebotCheck-in: - 57 changesets - Merge m-c to autoland a=merge
00:59pulsebotCheck-in: - Eric Rahm - Bug 935809 - Part 0: Include assertion header in DoublyLinkedList.h. r=waldo
00:59pulsebotCheck-in: - Eric Rahm - Bug 935809 - Part 1: Convert breakpoint lists to DoublyLinkedList. r=jimb
00:59pulsebotCheck-in: - Eric Rahm - Bug 935809 - Part 2: Convert onNewGlobalUpdateWatchers to DoublyLinkedList. r=jimb
00:59pulsebotCheck-in: - Eric Rahm - Bug 935809 - Part 3: Remove JSCList. r=jimb
01:03pulsebotCheck-in: - Jared Wein - Bug 1347413 - Remove the unused browser-fullZoom:zoomReset and browser-fullZoom:zoomChange notifications. r=dao,mccr8
01:24pulsebotCheck-in: - Chris Pearce - Bug 1358373 - Handle underestimating how many shmems the CDM needs to return decoded video frames to Gecko. r=gerald
01:25pulsebotCheck-in: - Boris Zbarsky - Bug 1359859. Use the right entry global for XBL constructor/destructor/field execution. r=bholley
01:37pulsebotCheck-in: - Sean Lee - Bug 1349489 - Part 1: Move the codes from FormAutofillHandler.collectFormFields to FormAutofillHeuristics.getFormInfo.; r=MattN
01:37pulsebotCheck-in: - Sean Lee - Bug 1349489 - Part 2: Add test fixtures for autofill field name heuristics.; r=MattN
01:38pulsebotCheck-in: - Jessica Jong - Bug 1357981 - Allocate input type=file related member variables only when needed. f=smaug, r=baku
01:38pulsebotCheck-in: - Gerald Squelart - Bug 1355506 - Ensure the ChromiumCDMProxy stays alive until the parent has completed its Shutdown(). r=cpearce
01:44pulsebotCheck-in: - Jonathan Watt - Bug 1359631 - Convert Toolkit's icons/autocomplete-search.svg to use context paint to improve performance. r=dao
01:44pulsebotCheck-in: - Jonathan Watt - Bug 1359631 - Convert Toolkit's extensions/utilities.svg to use context paint to improve performance. r=dao
01:46ihsiao|sheriffdutyerahm: ping
01:47ihsiao|sheriffdutyerahm: build bustage like
01:56geraldihsiao|sheriffduty: Hi! One of the above commits will also solve some intermittents. Once the main bug is resolved&fixed, is it preferable to mark these as duplicate, or to close them as resolved&fixed (with a comment linking to the bug where the fix is)?
02:04ihsiao|sheriffdutygerald: the latter, close them as resolved&fixed (with a comment linking to the bug where the fix is)
02:05geraldWill do. Thank you.
02:11pulsebotCheck-in: - Iris Hsiao - Backed out changeset 0b8bf5cb743f (bug 1358373) for build bustage. a=backout
02:14pulsebotCheck-in: - Iris Hsiao - Backed out 4 changesets (bug 935809) for build bustage. a=backout
02:15pulsebotCheck-in: - Emilio Cobos lvarez - Bug 1359995: Add back the null-check on curData. r=heycam
02:24pulsebotCheck-in: - Gabriel Luong - Bug 1317215 - Add tests for the LayoutActor and GridActor. r=pbro
02:30dmajornjn: ping
02:30pulsebotCheck-in: - Iris Hsiao - Backed out changeset 6208b116de63 (bug 1349489) for eslint failure
02:30pulsebotCheck-in: - Iris Hsiao - Backed out changeset 4ff3acec9515 (bug 1349489). a=backout
02:42ihsiao|sheriffduty!seen gl
02:42firebotgl was last seen 2 days and 11 hours ago, saying 'pbro: it can probably be changed' in #devtools.
02:46pulsebotCheck-in: - Gabriel Luong - Bug 1317215 - Part 2: Fix eslint errors with browser_layout_getAllGrids.js and devtools/server/tests/browser/head.js. r=me
02:59pulsebotCheck-in: - Chris Pearce - Bug 1358373 - Handle underestimating how many shmems the CDM needs to return decoded video frames to Gecko. r=gerald
03:02pulsebotCheck-in: - JW Wang - Bug 1358969 - let MediaDecoder::CurrentPosition() return a TimeUnit. r=kikuo
03:03pulsebotCheck-in: - JW Wang - Bug 1358972. P1 - remove unused code. r=kaku
03:03pulsebotCheck-in: - JW Wang - Bug 1358972. P2 - remove MediaDecoder::mEstimatedDuration and its friends which is never modified. r=kaku
03:06ihsiao|sheriffdutybz: hi
03:06ihsiao|sheriffdutybz: could you help to check mochitest failures like
03:28pulsebotCheck-in: - JW Wang - Bug 1359725 - add timeout to LoadInitData() in eme.js. r=gerald
03:38pulsebotCheck-in: - Phil Ringnalda - Backed out changeset ddcbb81610fd (bug 1355198) for Android test_ext_proxy.html bustage
03:43philorihsiao|sheriffduty: take him out, take them all out - bz's not going to manage to fix a Windows-PGO-only crash at this time of night, and browser_layout_getAllGrids.js isn't going to start looking any better
03:43wagleanyone know what the code in xul that determines the gaps and placement of items in a window? tk calls it "packer" iirc
03:43ihsiao|sheriffdutyphilor: ok
03:46pulsebotCheck-in: - Iris Hsiao - Backed out changeset d19a04e55bef (bug 1359859) for mochitest failure. a=backout
03:49njndmajor: pong
03:50dmajornjn: I have a confusing dmd build error
03:50njnit happens to the best of us
03:50njnwhat is it?
03:51dmajornjn: so I made this change, basically separating out some bits of a file:
03:51ihsiao|sheriffdutyoh, browser_layout_getAllGrids.js
03:52philoryeah, gl's not having a good night
03:52dmajornjn: and my reward was
03:52dmajorwhich is what I'd expect if stackwalk_windows.cpp were not included. Does that mean DMD doesn't define OS_ARCH==WINNT?!
03:53ihsiao|sheriffdutyphilor: pinged gl
03:53njndmajor: the build system is responsible for setting OS_ARCH, no?
03:53njn(in the, which is what I assume you meant)
03:53dmajorI, uh, er, <hand wave>!
03:54njndmajor: have you tried building locally on a windows machine?
03:54njnI think you can insert a print() call in the
03:54dmajornjn: no, mostly because I&#39;ve forgotten how to build DMD. can you remind me?
03:54njnbasically: ac_add_options --enable-dmd
03:55dmajoroh. that&#39;s quite simple!
03:56dmajornjn: I&#39;ll try that in the morning. any other advice?
03:56njndmajor: nothing comes to mind...
03:56dmajornjn: okay, I&#39;ll see what I can find. thank you
03:57pulsebotCheck-in: - Iris Hsiao - Backed out changeset 379d1f58868f (bug 1317215) for devtool failures at browser_layout_getAllGrids.js. a=backout
03:57pulsebotCheck-in: - Iris Hsiao - Backed out changeset b3f163644c78 (bug 1317215). a=backout
04:15pulsebotCheck-in: - Jeremy Chen - Bug 1359313 - whitelist -moz-border-*-colors related bindings to satisfy heap write analysis. r=bholley,heycam
04:22pulsebotCheck-in: - Nicholas Nethercote - Bug 1357298 - Remove the &#39;layers.frame-counter&#39; option. r=mstange,mattwoodrow.
04:45pulsebotCheck-in: - Andrzej Hunt - Bug 1356693 - Pre: use IOUtils.safeStreamClose() in IOUtils r=walkingice
04:45pulsebotCheck-in: - Andrzej Hunt - Bug 1356693 - Pre: MergeCursor can handle null cursors r=Grisha
04:45pulsebotCheck-in: - Andrzej Hunt - Bug 1356693 - infer: fix RESOURCE_LEAK&#39;s in base r=walkingice
04:45pulsebotCheck-in: - Andrzej Hunt - Bug 1356693 - infer: fix RESOURCE_LEAK&#39;s in services r=Grisha
04:45pulsebotCheck-in: - Andrzej Hunt - Bug 1356693 - infer: fix RESOURCE_LEAK&#39;s in stumbler r=walkingice
04:45pulsebotCheck-in: - Andrzej Hunt - Bug 1356693 - infer: fix RESOURCE_LEAK&#39;s in geckoview r=walkingice
05:37pulsebotCheck-in: - Boris Chiou - servo: Merge #16626 - Implement frames timing function (from BorisChiou:animation/timing_function/frames); r=emilio
05:39pulsebotCheck-in: - Boris Chiou - Bug 1359719 - Update mochitest expectations for frames timing function. r=xidorn
05:40pulsebotCheck-in: - Chris Manchester - Bug 1353090 - Convert `output` to use `print` in r=ted
05:40pulsebotCheck-in: - Chris Manchester - Bug 1353090 - Remove code handling multiple files from r=ted
05:41pulsebotCheck-in: - Luke Chang - Bug 1348751 - [Form Autofill] A utility library for handling full name and separated names, r=MattN
05:47pulsebotCheck-in: - Henri Sivonen - Bug 1359353 - Make the backing buffers of XPCOM strings available as mutable slices. r=mystor.
05:53pulsebotCheck-in: - Henrik Skupin - Bug 1335778 - Make switchToFrame synchronous for in-process frames. r=ato
05:53pulsebotCheck-in: - Henrik Skupin - Bug 1335778 - Synchronize navigate() for trigger methods using generators. r=ato
05:53pulsebotCheck-in: - Henrik Skupin - Bug 1335778 - Harden test_page_timeout_fail against intermittent failures. r=ato
05:53pulsebotCheck-in: - Henrik Skupin - Bug 1335778 - Make element click command check for page load and wait. r=ato
06:10pulsebotCheck-in: - Rob Thijssen - Bug 1359760 - enable: a11y, awsy, cpp, j, jit, jp, mn, ss suites on tc win10; r=jmaher
06:12pulsebotCheck-in: - Jason Duell - Bug 711886 - Fail Websocket if server replies with non-matching subprotocol, r=mcmanus
06:20pulsebotCheck-in: - Andrea Marchesini - Bug 1354308 - Entries API must support patches containing &#39;..&#39;, r=froydnj
06:23azakianyone here running multiple versions of firefox on linux? like nightly, release, dev edition, etc? wondering if it&#39;s normal for all of them to just end up using nightly&#39;s taskbar icon..
06:24nigelb&quot;task bar&quot;?
06:24azakiat first i thought maybe it replaced the normal firefox icons, but it doesn&#39;t seem like it has. the icons in /usr/share/pixmaps and /usr/share/icons seem to have different names.
06:24azakinigelb: task list? window list? whatever you want to call it..
06:25nigelbThe gnome left bar?
06:25azakino, i&#39;ve enabled the gnome window list, the one you get in &quot;classic mode&quot; at the bottom.
06:25nigelbAh, I don&#39;t use classic. No idea.
06:26nigelbBut in the past, when I did this, I used a desktop file.
06:26nigelband you could specify icon in the desktop file.
06:26nigelbThat would make them have different icons.
06:26nigelbPerhaps that&#39;s what you need?
06:26azakiwell, the left bar, and the launchers in the gnome search thing all have different icons already
06:26azakiso it doesn&#39;t seem like that&#39;s the problem
06:27nigelbOh boy.
06:27nigelbThis is going to be one of those really subtle gnome bugs.
06:27azakiso you&#39;re convinced it&#39;s gnome then? hm
06:29nigelbI wouldn&#39;t be surprised if it&#39;s gnome.
06:29azakithey also have the correct icons in system monitor it seems. i just checked
06:30nigelbyeah, so it&#39;s not from the firefox end.
06:30nigelbJust somethng wonky in gnome classic.
06:31azakiwell, you can enable the window list without actually using gnome classic. but it&#39;s a supported gnome extension that ships out of the box, since it&#39;s used for classic mode
06:31pulsebotCheck-in: - Brian Birtles - servo: Merge #16625 - SMIL support for Gecko (from birtles:smil-support); r=hiro,heycam
06:32kanruazaki: it&#39;s something with the WM_CLASS x11 property
06:32nigelblet me try it out :)
06:32kanruazaki: WM_CLASS default to the exectuable name
06:33kanrubug 849571 and bug 1151753
06:33pulsebotCheck-in: - Brian Birtles - Bug 1355348 - Add FFI for getting SMIL override styles; r=heycam
06:33pulsebotCheck-in: - Brian Birtles - Bug 1355348 - Move IsFlattenedTreeDescendantOf helper to nsContentUtils; r=heycam
06:33pulsebotCheck-in: - Brian Birtles - Bug 1355348 - Add SMIL restyles in the stylo pretraverse; r=heycam
06:33pulsebotCheck-in: - Brian Birtles - Bug 1355348 - Move ServoComputedValuesWithParent to separate file; r=hiro
06:33pulsebotCheck-in: - Brian Birtles - Bug 1355348 - Add constructor to AnimationValue; r=hiro
06:33pulsebotCheck-in: - Brian Birtles - Bug 1355348 - Factor out method to remove minus sign from property values; r=hiro
06:33pulsebotCheck-in: - Brian Birtles - Bug 1355348 - Make nsSMILCSSValueType store an AnimationValue instead of a StyleAnimationValue; r=heycam
06:33pulsebotCheck-in: - Brian Birtles - Bug 1355348 - Convert stylo assert in nsComputedDOMStyle::DoGetStyleContextNoFlush to a warning; r=hiro
06:33pulsebotCheck-in: - Brian Birtles - Bug 1355348 - Don&#39;t return early when sampling SMIL animations with Servo backend; r=heycam
06:34azakinow that i think of it, when using dev edition, it always used the icon for firefox-release, i thought it was just normal, but i just installed firefox-nightly yesterday and then suddenly everything uses nightly icon, which is odd. =o
06:34azakikanru: i&#39;ll check those bugs out
06:38nigelbRight, so different desktop files are still the right way to go.
06:40azakihm, wait, i already have different desktop files.. but i am launching firefox from terminal
06:40azakii like to monitor the console errors and stuff sometimes
06:41azakiso i&#39;m guessing that&#39;s why it&#39;s happening =\
06:42gastonhmmm so changing a value in libref/init/all.js doesnt change the *default* value of the pref, if unset in profile ?
06:47azakinigelb: is it hard to change the wm_class for the different versions? the bug report kanru linked was around 2 years old. it seems the problem reported in unity in that bug is worse than the issue i&#39;m experiencing.
06:47nigelbno idea.
06:47nigelbI&#39;m not the expert :)
06:50azakihm, i just confirmed that changing the wm_class with --class firefox option does work, as described in this bug
06:51azakifirefox --profilemanager --new-instance --class firefox
06:51azakilike this basically ^
06:54azakii guess maybe i&#39;ll write a bash script to launch it with that option, as a workaround for now.
06:55azakisomething tells me wayland will have replaced x11 long before this bug is fixed. =p
06:55kanruI think the fundamental issue is gtk stopped supporting setting wm_class directly so if we want to fix it we have to take other routes.
06:56kanrualso some notes from
06:56azakiand wm_class seems x11-specific so presumably this wouldn&#39;t happen with a native wayland port of firefox.
06:58azakikanru: hm, i see.. but then how does the --class command line option work?
07:01kanruthe code is here
07:02kanruI guess
07:03kanruThe icon issue has annoyed me so I did some research. But this is as far as I went.
07:08azakikanru: ah. i love me some magic ;D
07:08azakithank you for the help
07:08azakii kind of asked with quite a bit of pessimism, since i didn&#39;t expect anyone else would&#39;ve actually seen such a subtle issue before. i really appreciate the help. =)
07:17pulsebotCheck-in: - Tobias Schneider - Bug 1321865 - Enable IntersectionObserver tests. r=jet
07:17pulsebotCheck-in: - Tobias Schneider - Bug 1321865 - Enable IntersectionObserver. r=jet
07:18pulsebotCheck-in: - Tobias Schneider - Bug 1313927 - Split no-root test from test_intersectionobservers.html into its own mochitest, so that we can skip that part on Android. r=jet
07:21jetdholbert: thanks! ^^^
07:21dholbertjet: sure! :D
07:24* dholbert zzz
07:25pulsebotCheck-in: - Alex Chronopoulos - Bug 971528 - Allow stereo capture in AudioCallbackDriver. r=padenot
07:25pulsebotCheck-in: - Alex Chronopoulos - Bug 971528 - Expect stereo input in MediaEngineWebRTCMicrophoneSource. r=padenot
07:35kanruazaki: you&#39;re welcome :)
07:46birtlesMy push seems to have triggered a new heap-write hazard
07:47birtlesLooks like it doesn&#39;t know what to do with the pointer-to-lamda called from Gecko_GetStyleAttrDeclarationBlock
07:47heycambirtles: maybe change it to macros or a templated thing, or just duplicate the code?
07:47heycamhopefully it will be happy with that
07:48birtlesheycam: ok, I suppose that&#39;s better than adding it to ignoreContents
07:49birtlesI don&#39;t know if I&#39;ll be able to setup a build environment to test it locally before I head out the door in 10mins though...
07:49birtlesMaybe I can just push a fixup to autoland and cross my fingers
07:50heycambirtles: do that, and if it doesn&#39;t work, I&#39;ll bump the expected hazard count, and you can look at it later
07:50birtlesheycam: ok, great, thanks
07:57Tomcat|sheriffdutybirtles: autoland is closed
07:57Tomcat|sheriffdutywe are pinging the taskcluster tram
07:58birtlesTomcat|sheriffduty: ok, thanks -- I have a patch ready to push that I think should fix it
08:03birtlesI really should submit a PR to mozilla-tree-status to add autoland to the list of repositories
08:05Standard8well autoland doesnt matter too much, just tell it to land and itll wait until it opens :-)
08:05* Standard8 had 3 bugs land patches with that last night
08:08birtlesexcept if you are pushing directly to autoland
08:10birtlesTomcat|sheriffduty: I was hoping to push this patch to (hopefully) fix the hazard bustage:
08:11birtleswould you mind pushing that when autoland re-opens?
08:11birtlesIf that doesn&#39;t work heycam was just going to update the hazard count until I can work on a proper fix tomorrow
08:11Tomcat|sheriffdutybirtles: we are not allowed to push manually to autoland anymore
08:11Tomcat|sheriffdutyexecept mergeds
08:12birtlesTomcat|sheriffduty: I think stylo team can in order to fix breakage
08:13birtles(because backing out the changeset will cause further breakage due to the inbound servo changes that it corresponds to)
08:15Tomcat|sheriffdutybirtles: ok in that case i will land
08:15Tomcat|sheriffdutyand take hte blame :))
08:15birtlesTomcat|sheriffduty: thank you!
09:04leplatremhi there :)
09:05leplatremquick question... the build takes ages on my machine (65min) with &quot;only&quot; 30% miss rate on ccache
09:06leplatremis that normal? I&#39;ve got 4 core @ 2.60GHz ...
09:07Standard8leplatrem: which platform, and hdd or ssd?
09:08leplatremStandard8, linux 64 on ssd :/
09:08leplatremmy mozconfig contains only 1 line for ccache
09:08Standard8leplatrem: also, if youre not touching c++ have you tried artifact builds?
09:09leplatremwell I touch c++ from times to times
09:10leplatrem(or telemetry stuff like histograms definitions)
09:10Standard8well the only thing Id say is that Id probably expect them to be about 30 mins, unless theyre debug builds - but thats just based on what the tree builders use. I use a Mac so cant really compare
09:12leplatremStandard8, okay.. thanks for your feedback
09:12flodbobowen: you just NI a spammer ;-) (in Bugzilla)
09:13Standard8leplatrem: someone else might have a better response, alternately try asking in #build (maybe more around PST morning though)
09:13bobowenflod: oh... whoops
09:14flod(mostly to explain why i marked your question ass off-topic, so it won&#39;t show up by default)
09:14floddo&#39;h, as, not ass
09:14flodhe&#39;s also a fast one, based on the number of comments, usually they stop at 2 or 3
09:14Piketypo fight!
09:15bobowenflod: thanks
09:42Gijsif a git changeset copies file a to file b, removes most of the content of file a, and makes minor changes to file b, how do I get a diff of the previous version of file a (prior to the removals) and the new file b (after any changes after the copy) ?
09:42Gijsin reviewboard when using hg, that&#39;s the only thing that shows up (ie, it shows you the changes to file b after the copy)
09:42Gijsbut github just shows the entire file as &quot;hey, this is new&quot;
09:42Gijsdespite git knowing it&#39;s a rename
09:43Gijsthis is really frustrating when trying to review the code, because rather than just reviewing changes now I need to go through the entire file
09:50PikeGijs: git doesn&#39;t know the rename, git makes them up based on similarity of files before and after
09:51Pikecopies and renames, that is
09:51GijsPike: sure, but it shows that it&#39;s picked up this rename
09:51jgrahamGijs: Try git diff -M05 or something (that sets the threshold for considering files to be a rename at 5%)
09:52Gijsie when I pull this branch I see:
09:52Gijs rename recipe-client-addon/test/browser/{browser_bootstrap.js => browser_ShieldRecipeClient.js} (86%)
09:52flodinteresting, I&#39;ve always seen GitHub doing the smart thing and just showing the changed parts
09:52jgrahamGijs: Oh 86% is way over the threshold
09:52Gijsjgraham: &quot;git show&quot; doesn&#39;t take the -M option (or ignores it, rather)
09:53Gijsdoesn&#39;t that mean it&#39;s 86% similar?
09:53GijsI&#39;ve since just gone and checked out file a from master and done a non-git-diff
09:53Gijswhich works but is ugly
09:53jgrahamI was using git diff, but I don&#39;t think that&#39;s the problem
09:53Gijsjgraham: sure, but I don&#39;t know how to get a diff from an extant cset in git
09:53Gijsjgraham: people always told me to &quot;just use git show&quot;
09:54jgrahamYeah git show will show the commit but presumably git diff SHA^..SHA will work (if there aren&#39;t merge commits or whatever)
09:54Gijsjgraham: ie what&#39;s the magic incantation to make &quot;git diff&quot; show me an already-committed cset?
09:54GijsI have to go lookup the shas?
09:55jgrahamWell unless it&#39;s HEAD or something presumably you need taht for git show too
09:55Tomcat|sheriffdutybirtles: landed
09:55jgrahamGijs: I suspect that haing a and b both in the repo is what&#39;s confusing things here
09:56Gijsjgraham: why is that confusing?
09:56* Gijs mutters about git being like hg but way clunkier
09:56Gijshg diff -c .
09:56Gijsboom, done.
09:57Gijsjgraham: so the cset I&#39;m &quot;on&quot; is the cset I&#39;m interested in
09:57Gijsno merges
09:57Gijsbut I&#39;ve never figured out what the parent cset is called in git
09:58Gijsbut I thought HEAD was like tip ?
09:59jgrahamNo HEAD is whatever&#39;s checked out
10:00jgrahamNot &quot;whatever the commit with the highest arbitary sequence id&quot; is, or whatever the weird hg semantics are
10:03Gijsthe last thing that got committed or pulled in, yes :)
10:04Gijsjgraham: so is there some easy way to just do a diff from HEAD to its parent?
10:06jcristaugit diff HEAD^..
10:07Gijsok, so the -M thing seems to just not work, it still shows the entire file as new
10:11jgrahamGijs: Maybe try -C instead of -M
10:12jgrahamSince this seems like a copy rather than a rename
10:17Gijsjgraham: success! That works. Why are there two options instead of just 1? :(
10:57Tomcat|sheriffdutyjesup: ping
10:57jesupTomcat|sheriffduty: pong
11:04irfanHas anyone worked with the firefox datachannels? I&#39;m working on a C++ client that speaks webRTC datachannel and having it interop with js/firefox. My only problem is that my sctp send buffer doesn&#39;t flush when I send from the C++ side. I noticed that firefox doesn&#39;t have this problem. Does anyone know what could be wrong?
11:05jesupirfan: not sure, but see our impl (using the usrsctp code) in netwerk/sctp/datachannel/*
11:05jesuplibrary is in netwerk/sctp/src
11:09irfanjesup: Thanks. I&#39;ve been looking at it for a while.. but no luck so far :(
11:10Tomcat|sheriffdutyTrees reopen
12:02luisbgis there a way to filter Bugzilla email by component? any settings in the bugzilla side to make the component part of the subject?
12:06Standard8luisbg: theres headers you can use. Are you in gmail?
12:07Standard8luisbg: may help
12:08luisbgStandard8: awesome! thanks for the pointer
12:11luisbgStandard8: Step 1 in that list is not available anymore, if you know somebody who might be able to change that wikipage
12:11luisbgand Step 2, I think
12:11freddybluisbg: HTML is enabled by default (Step 1)
12:12freddybstep 2, I&#39;m not sure. Maybe its on by default :|, either way it makes sense to tell folks in #bmo that their docs are outdated
12:12freddyboh x-headers in body is not on by default. I thought it&#39;s about x-headers in general
12:13freddybyou can filter based on x-headers without them being in the body now.
12:13freddybI think.
12:13freddyblooking at my filters that doesnt seem true :|
12:13Kwanluisbg, freddyb: that stuff is in now
12:16Kwanpulsebot: ping
12:18luisbgswitching &quot;Include X-Bugzilla- headers in BugMail body&quot; to On
12:18luisbgKwan: thanks! the wiki page should link to that URL now (and keep step 1 and 2 since it is good to know)
12:30pulsebotCheck-in: - Daosheng Mu - Bug 1358053 - Remove DataSourceSurface assertion when doing snapshot in WebGL; r=jgilbert
12:30pulsebotCheck-in: - Nevin Chen - Bug 1329156 - Add telemetry for customtab customization usage. r=walkingice
12:31pulsebotCheck-in: - Bob Silverberg - Bug 1344087 - Update browser_ext_sessions_getRecentlyClosed_tabs.js to use a URL with an identical title, r=mixedpuppy
12:31pulsebotCheck-in: - Nicholas Nethercote - Bug 1358074 (part 7) - Split PS in two. r=mstange.
12:31pulsebotCheck-in: - Bob Silverberg - Bug 1359806 - Fix sessions.restore to return an object instead of an array, r=mixedpuppy
12:31pulsebotCheck-in: - Bob Silverberg - Bug 1358127 - Fix so it doesn&#39;t return the contents of tag folders, r=mak
12:31pulsebotCheck-in: - Masatoshi Kimura - Bug 1358758 - Use CSSIntRect for nsIFrame::GetScreenRect. r=kats
12:37pulsebotCheck-in: - Emilio Cobos lvarez - servo: Merge #16629 - Bug 1336863: Ensure we don&#39;t restyle with outdated rules (from emilio:dirty-doc); r=heycam
12:40pulsebotCheck-in: - Petr Sumbera - Bug 1359841 - xpcom/base/nsStatusReporterManager.cpp should cast getpid() return value to int. r=alchen
12:40pulsebotCheck-in: - Kevin Jones - Bug 1358735 - Ensure userContextId is always set in TabStateCache. r=mikedeboer
12:40pulsebotCheck-in: - digitarald - Bug 1318095 - Strawman for adding markers for navigation timing. r=smaug
12:44pulsebotCheck-in: - Christoph Kerschbaumer - Bug 1345615: Allow websocket schemes when using &#39;self&#39; in CSP. r=freddyb,dveditz
12:44pulsebotCheck-in: - Christoph Kerschbaumer - Bug 1345615: Test websocket schemes when using &#39;self&#39; in CSP. r=freddyb,dveditz
12:45pulsebotCheck-in: - Emilio Cobos lvarez - Bug 1359779: Also look at pseudo frames in nsComputedDOMStyle::DoGetStyleContextNoFlush. r=heycam
12:46pulsebotCheck-in: - Henrik Skupin - Bug 1357407 - Enable logger output for framescript when remoteness is enabled. r=ato
12:53pulsebotCheck-in: - Vangelis Katsikaros - Bug 1356872 - Hard to distinguish IP from Port in IPv6 address. r=ntim,rickychien
12:54whimboosmaug: hey. i have a question regarding remoteness and loading a URL with a file:// protocol
12:55smaugbobowen would know the sandboxing part of that way better
12:55smaugwhimboo: but ask
12:56whimboosmaug: lets see.. we have a frame script installed which handles all navigation. When I try to load a file:// URL the framescript gets reloaded
12:56pulsebotCheck-in: - Vangelis Katsikaros - Bug 1356869 - Add cookies and set cookies columns. r=ntim,rickychien
12:56whimbooall is fine when its from non-remote to remote
12:56whimboobut if the formely loaded page is already remote, only the window id is changing
12:57whimbooso it looks like we replacing the frame script only due to security reasons?
12:57whimboowe then send back the new window id to the main script, and there I&#39;m trying to find the window via getOuterWindowWithId()
12:57whimboobut that returns null
12:58smaugwhimboo: when you have first http:// or some such loaded, it is in child process Foo and when file:// is loaded, the Foo is replaced with Bar
12:58philorRyanVM: oh, that should be fun, pdf.js leak on beta-55
12:58smaugwhimboo: in other words, file:// urls are loaded in different processes
12:59whimboosmaug: k, sounds reasonable
12:59whimboosmaug: but why can&#39;t I find the window with the known id?
12:59yuryphilor: how it leaks?
12:59smaugwhimboo: which window ID and where?
13:00whimboosmaug: in the framescript we retrieve it via winUtil.outerWindowID
13:00smaugwhere does the &quot;main script&quot; run?
13:00philoryury: for Windows, for ASan
13:00whimboosmaug: the main script runs in teh chrome process
13:00smaughow would parent process know about the window ID
13:01smaugwhen the window ID is in child process
13:01philoryury: that&#39;s current mozilla-central run as though it had been merged to beta, thus the &quot;fun&quot;
13:01whimboosmaug: we send a message to the chrome process to let it know about the id
13:02smaugwhimboo: sounds like then your code which I&#39;m not familiar with
13:02whimboosmaug: it works all fine for usual http: urls
13:02smaugwhere is the id sent?
13:03smaugwhimboo: is it possible that you try to access the ID in parent process before the ID from child process has been received ?
13:04whimboosmaug: no. the code is in the method which receives the data from the child
13:04whimbooand the id is correct in the parent process
13:04smaugthen I&#39;m lost what the question is :) I thought you said you don&#39;t have id in parent
13:04smaugwhat does getOuterWindowWithId return?
13:04smaugin parent
13:05smaugsince of course there isn&#39;t any Window object for the ID in parent
13:05whimbooi said I have the id, but that getOuterWindowWithId() returns null
13:06smauggetOuterWindowWithId() in parent process should return null
13:06smaugsince the Window object lives in child
13:06pulsebotCheck-in: - 12 changesets (bugs 1308343, 1358601, 1334676, 1351278, 1358056 and 5 other bugs)
13:06whimboosmaug: hm, should this be the case for all tabs which are remote?
13:07RyanVMphilor: feeling glad that we update that from upstream pretty regularly at least :