mozilla :: #developers

14 Mar 2017
00:01pulsebotCheck-in: - Xidorn Quan - Bug 1346669 - Make font-synthesis not a subprop of font shorthand. r=jfkthame
00:05pulsebotCheck-in: - David Keeler - bug 1239344 - remove error alert for successful PKCS12 operations r=Cykesiopka
00:05Kwanokay, sweet, after a little bit of fighting the virtualenv to get nose working/installed correctly, I seem to still be passing all tests, PR coming up
00:13Kwan...curse you flake8
00:19bradwertharyx: sorry, I'll fix patch for 1343695 and attempt autoland again. Is there anything I need to do to remove what I've already pushed?
00:21mrbkapnjn: ping?
00:25njnmrbkap: pong
00:27mrbkapnjn: Hey, I'm debugging a shutdown crash we're seeing on OSX and I think that AutoProfileRegister is (indirectly) the culprit. Are you the right person to talk to?
00:27njnmrbkap: probably
00:28mrbkapnjn: (if not, I'll find a new reviewer for a related patch I just attached)
00:28njnmrbkap: mstange too, maybe
00:28pulsebotCheck-in: - Brad Werth - Bug 1337305 Part 1: Refactor GetRuleNodeForElement to just return the StyleContext, so caller can either get a RuleNode or a raw servo rule. r=shinglyu,xidorn
00:28pulsebotCheck-in: - Brad Werth - Bug 1337305 Part 2: Change GetCSSStyleRules to return RuleNodes for Gecko rules, and nothing otherwise (instead of asserting). r=xidorn
00:29mrbkapnjn: The crash that I'm debugging is
00:30mrbkapnjn: We're freeing an nsThread object from the PRThread destructor and the NSLogRefcount is causing us to try to recreate the PRThread object.
00:30mrbkapnjn: Meanwhile, we're extremely late in shutdown, i.e. running the atexit handlers.
00:32mrbkapnjn: I'm pretty convinced that this is happening because registering a thread with the profiler causes us to create an nsThread for it and then nobody gets around to cleaning it up before XPCOM shutdown, leading to situations where we end up deleting it too late.
00:32mrbkap(does that make sense?)
00:33njnsort of...
00:33njnmrbkap: I wonder if is relevant here
00:33njnthat's a patch I'm planning to land later today, which fixes a thread leak in the profiler
00:34njnI introduced that leak in the middle of last week
00:36njnmrbkap: do you know the name of the thread on which the problem occurs?
00:36njnif it's "SamplerThread", then there's a good chance this fix will help
00:36mrbkapnjn: No, I'm going to re-push to try to figure that out.
00:36mrbkapnjn: Your patch fixes a recent regression, right?
00:36njnmrbkap: yes, from the middle of last week
00:36mrbkapnjn: if so, I don't think so; this intermittent has been around for quite some time.
00:37pulsebotCheck-in: - 25 changesets (bugs 1337133, 1346527, 1013882, 1322112, 1336791 and 7 other bugs)
00:39mrbkapnjn: Unfortunately, with my latest logs, all I know is that this isn't a thread that was created via nsIThread.
00:43pulsebotCheck-in: - Kaku Kuo - Bug 1346120 part 1 - Extract BlankMediaDataDecoder so it can be shared; r=jwwang
00:43pulsebotCheck-in: - Kaku Kuo - Bug 1346120 part 2 - Implement NullDecoderModule; r=jwwang
00:44pulsebotCheck-in: - Kaku Kuo - Bug 1346120 part 3 - Use NullDecoderModule while suspending a video element's decoder; r=jwwang
00:44pulsebotCheck-in: - Kaku Kuo - Bug 1346120 part 4 - Only set ImageContainer if there are valid new images in VideoSink::RenderVideoFrames(); r=jwwang
00:44pulsebotCheck-in: - Kaku Kuo - Bug 1346120 part 5 - Revert the blank decoder to create green frames; r=jwwang
00:44pulsebotCheck-in: - Kaku Kuo - Bug 1346120 part 6 - Test drawImage gets a non-single-color image from suspended video; r=jwwang
00:49pulsebotCheck-in: - 1 changesets - servo: Merge #15931 - style: Kill SharedStyleContext::default_computed_values (from emilio:die-defaultvalues-die); r=mbrubeck
00:51squibwell, it only took the whole afternoon but i actually figured out how to make an hg bookmark and push it to mozreview (to be fair 80% of that time was spent messing with my zsh config to show the current bookmark in my prompt)
02:00pulsebotCheck-in: - Makoto Kato - Bug 1345336 - Part 1. Remove empty zoneStrings entries. r=Waldo
02:00pulsebotCheck-in: - Makoto Kato - Bug 1345336 - Part 2. Update ICU date file. r=Waldo
02:00pulsebotCheck-in: - Makoto Kato - Bug 1345336 - Part 3. Update CLOBBER. r=Waldo
02:09pulsebotCheck-in: - Eric Rahm - Bug 1344081 - Switch to Base64Decode in nsHttpNegotiateAuth::GenerateCredentials. r=jduell
02:12pulsebotCheck-in: - JW Wang - Bug 1346681 - Label runnables in dom/media/gmp/GMPVideoEncoderParent.cpp. r=jesup
02:12pulsebotCheck-in: - L. David Baron - Bug 1346692 - Clarify reftest documentation about combining <failure-type>s. No review.
02:38pulsebotCheck-in: - Iris Hsiao - Backed out changeset eead99c3e41a (bug 1275287) for mochitest failures in browser_ext_optionsPage_browser_style.js
02:43pulsebotCheck-in: - Henry Chang - Bug 1346757 - Change the downloadError callback timing. r=francois
02:47pulsebotCheck-in: - Nicholas Nethercote - Bug 1346356 (part 1) - Fix SamplerThread leak in profiler_start(). r=mstange.
02:47pulsebotCheck-in: - Nicholas Nethercote - Bug 1346356 (part 2) - Merge SamplerThread::Join() into ~SamplerThread(). r=mstange.
02:47pulsebotCheck-in: - Nicholas Nethercote - Bug 1346356 (part 3) - Do most NotifyObserver() calls when gPSMutex is unlocked. r=mstange.
02:48pulsebotCheck-in: - Nicholas Nethercote - Bug 1346356 (part 4) - Clean up notification in the profiler. r=mstange.
02:58pulsebotCheck-in: - Kaku Kuo - Bug 1346498 part 1 - extract the MediaDecoder::NotifyCompositor() method; r=jwwang
02:58pulsebotCheck-in: - Kaku Kuo - Bug 1346498 part 2 - implement the VideoDecodeMode mechanism in MDSM; r=jwwang
02:58pulsebotCheck-in: - Kaku Kuo - Bug 1346498 part 3 - implement the UpdateVideoDecodeMode() policy in MediaDecoder; r=jwwang
02:58pulsebotCheck-in: - Kaku Kuo - Bug 1346498 part 4 - remove mIsVisible cannonical-mirror pair; r=jwwang
02:58pulsebotCheck-in: - Kaku Kuo - Bug 1346498 part 5 - don&#39;t check mHasSuspendTaint in HandleVideoSuspendTimeout(); r=jwwang
02:58pulsebotCheck-in: - Kaku Kuo - Bug 1346498 part 6 - remove mHasSuspendTaint cannonical-mirror pair; r=jwwang
02:58pulsebotCheck-in: - Kaku Kuo - Bug 1346498 part 7 - remove outdated comments in HTMLMediaElement.cpp; r=jwwang
02:58pulsebotCheck-in: - Kaku Kuo - Bug 1346498 part 8 - extract the HTMLMediaElement::NotifyDecoderActivityChanges() method; r=jwwang
02:58pulsebotCheck-in: - Kaku Kuo - Bug 1346498 part 9 - move all policy codes into MediaDecoder::UpdateVideoDecodeMode(); r=jwwang
03:04pulsebotCheck-in: - Kaku Kuo - Bug 1346116 part 1 - initialize MediaDecoder::mIsDocumentVisible and MediaDecoder::mIsElementVisible at HTMLMediaElement::FinishDecoderSetup(); r=jwwang
03:04pulsebotCheck-in: - Kaku Kuo - Bug 1346116 part 2 - consider a video is in-tree or not in the suspend-video-decoding policy; r=jwwang
03:04pulsebotCheck-in: - Kaku Kuo - Bug 1346116 part 3 - a test case for not suspend not-in-tree videos; r=jwwang
03:15pulsebotCheck-in: - L. David Baron - Bug 1305754 - Don&#39;t rebuild the whole path in the rule tree when resolving style with replacement. r=heycam
03:18zombiewhat&#39;s the order of magnitude of the data we send in sync on child process creation?
03:18zombiefor things we would like to avoid, like prefs?
03:19pulsebotCheck-in: - 12 changesets (bug 1344629)
03:21pulsebotCheck-in: - Blake Kaplan - Bug 1347007 - Register with the profiler after doing so for the ThreadManager. r=njn
03:42ihsiao|sheriffdutyxidorn: heycam hi, do you know the stylo mochitest failure like
03:42xidornihsiao|sheriffduty: I&#39;ll update the expectation
03:43ihsiao|sheriffdutyxidorn: ok, thank you
03:48xidornihsiao|sheriffduty: could you backout dmajor&#39;s bug 1344629?
03:49xidornwhich breaks stylo builds
03:49ihsiao|sheriffdutyxidorn: sure
03:50ihsiao|sheriffdutydmajor: stylo build bustage like
03:53pulsebotCheck-in: - Iris Hsiao - Backed out 12 changesets (bug 1344629) for stylo build bustage
03:53xidorndmajor: I&#39;ll try to run a stylo build with your patches and see what should we do
04:02pulsebotCheck-in: - Xidorn Quan - Bug 1341102 - Update stylo test expectations.
05:25pulsebotCheck-in: - BharatR123 - Bug 1302888 - Replace GetPresContext() with nsContentUtils::GetContextForContent() to obtain nsPresContext* in dom/animation r=hiro
05:32pulsebotCheck-in: - Masayuki Nakano - Bug 1318312 part.1 Add automated tests for checking focus move at using Selection API r=smaug
05:32pulsebotCheck-in: - Masayuki Nakano - Bug 1318312 part.2 Mark Selection as &quot;called by JS&quot; when every Selection API which may cause changing selection is called by JS r=smaug
05:32pulsebotCheck-in: - Masayuki Nakano - Bug 1318312 part.3 Selection should move focus at every selection change when it&#39;s called by JS r=smaug
05:35ajacan someone tell me how to needinfo someone in a bug i just filed?
05:35bz_sleepaja: Bug is already filed?
05:36bz_sleepaja: click the &quot;Edit bug&quot; button at top right
05:36* aja might not have the magical powers
05:36bz_sleeper, actually...
05:36bz_sleepJust scroll all the way to the bottom
05:36bz_sleepunder the comment textarea, do you have a checkbox labeled &quot;Need more information from&quot; ?
05:37ajabz_sleep, Tomcat would be :Tomcat ?
05:37bz_sleepNo problem
05:37* bz_sleep sleeps harder
05:42pulsebotCheck-in: - Nico Grunbaum - Bug 1346913 - rewrite waitForRtpFlow using async await;r=jib
05:51pulsebotCheck-in: - 1 changesets - servo: Merge #15935 - Binding update (from upsuper:binding-update); r=Wafflespeanut
05:52xidorndmajor: I think you can land it now ^
05:52dmajorxidorn: cool, thanks
05:56pulsebotCheck-in: - 12 changesets (bug 1344629)
06:02pulsebotCheck-in: - Carsten &quot;Tomcat&quot; Book - Backed out changeset 985d9b83325f (bug 1337043) for causing bug 1337043
06:11pulsebotCheck-in: - Timothy Nikkel - Bug 1346510. Don&#39;t allow the surface cache to unlock the animated frames of an animated image (when discarding of animated images is disabled). r=aosmond
07:06xidornit is probably the first time I&#39;m seeing &quot;Mozilla Autoland&quot; commenting on a bug :)
07:06Tomcat|sheriffdutyyeah i guess this is new
07:06xidornit took me a while to realize that it isn&#39;t a real person who has surname &quot;Autoland&quot;
07:09Tomcat|sheriffdutyhmmm this sounds like a nice irc nick :)
07:09Tomcat|sheriffdutyxidorn: i guess i even commented in this bug resulting in this autoland comment
07:10Tomcat|sheriffdutywas bad to land checkin-needed stuff and don&#39;t get notified of merge conflicts etc
07:10Tomcat|sheriffdutyerr conflicts (not merge conflicts :)
07:10Tomcat|sheriffdutywhen pushing things with autolander
07:10xidornright, I sometimes missed landing patches because of that as well
07:11xidornwe really need a bot to walk through bugs with unlanded r+ed patches, and notify people whether they want to do anything
07:14ewongTomcat|sheriffduty: isn&#39;t your workload gonna go sky high if the new proposed review process goes through?
07:16Tomcat|sheriffdutyewong: not nessesary, like &quot;Direct commit access to repositories will be strictly limited to sheriffs and a subset of release engineering.
07:16Tomcat|sheriffdutyis what we do on autoland repo already
07:16Tomcat|sheriffdutyand it &quot;works&quot;
07:16ewongTomcat|sheriffduty: when I read that.. that&#39;s the first thing I thought
07:17Tomcat|sheriffdutyewong: when gps told us in london about the plan for autoland i was wondering if that would work, but it actually works better than expected
07:17Tomcat|sheriffdutyand for checkin-needed as example it even saves time for sheriffs
07:18ewongTomcat|sheriffduty: so m-i is going to be pretty much obsolete, right?
07:20Tomcat|sheriffdutyyeah i personally think so
07:20Tomcat|sheriffdutybut we also survived the shutdown of fx-team and the world didn&#39;t stopped :)
07:20ewongbut tbh, not every pushed to fxteam.. whereas *everyone* pushes to m-i
07:20Tomcat|sheriffdutyewong: and the other positive side is we would get rid of merge-commits
07:21ewongs/not every/not everyone/
07:21Tomcat|sheriffdutywho can be in general also dangerous
07:21ewongTomcat|sheriffduty: oh yeah.. merge commits
07:29pulsebotCheck-in: - Xidorn Quan - Bug 1346734 - Disable the leaking reftest.
07:30Tomcat|sheriffdutyxidorn: ++
07:31ewongTomcat|sheriffduty: so autoland don&#39;t have r= flags?
07:31xidornewong: it has, but it doesn&#39;t block you from landing without r= flag
07:31xidornwhen you have permission to push to it directly :)
07:32ewongxidorn: never done an autoland push.. so dunno what the difference is between that and m-i..
07:33ewongI can push to m-i.. but not autoland?
07:33kanruewong: only sheriffs can push to autoland
07:34Tomcat|sheriffdutyand autolander (the tool)
07:34xidornewong: no difference, you can push to m-i without r= flag as well
07:34ewongso flag a checkin-needed and that&#39;s in?
07:34ewongxidorn: you can? I didn&#39;t know that.
07:34ewongI&#39;ve always thought all pushes required r+
07:34* xidorn does that all the time :)
07:35xidornfor trivial changes, I mean
07:35Tomcat|sheriffdutyewong: pushing a checkin-needed looks like via the autolander
07:36Tomcat|sheriffdutythere is a land commits button and this lands then changes in the autoland repo
07:36Tomcat|sheriffdutyno hg import or something
07:36Tomcat|sheriffdutyalso its there much easier to push to try
07:37Tomcat|sheriffdutyewong: and the nice thing for checkin-needed for sheriffs is that this works also for bugs with x patches to checkin
07:37Tomcat|sheriffdutylike this 10 part bugs :)
07:37pulsebotCheck-in: - Tomislav Jovanovic - Bug 1313648 - Add previousVersion to onInstalled details r=aswan
07:38freddybah, there is a specific interface for going through checkin-needed? neat.
07:38freddybI didnt know
07:38Tomcat|sheriffdutyfreddyb: i think this is bascially the ui for landing patches on autoland
07:39Tomcat|sheriffdutyso not checkin-needed specific :)
07:39freddybah ok
07:40pulsebotCheck-in: - Matthew Wein - Bug 1275287 - Use the same browser_style styling for options that we use for popups r=mixedpuppy
07:41pulsebotCheck-in: - Kaku Kuo - Bug 1309494 part 0 - make the seek operation at StateObject::HandleResumeVideoDecoding() observable; r=jwwang
07:41pulsebotCheck-in: - Kaku Kuo - Bug 1309494 part 1 - let video controler handle the &quot;mozexitvideosuspend&quot; event; r=jaws
07:41pulsebotCheck-in: - Kaku Kuo - Bug 1309494 part 2 - implement the show throbber mechanism; r=jaws
07:42pulsebotCheck-in: - Munro Mengjue Chiang - Bug 1345038 - Label runnables in dom/media/imagecapture/CaptureTask.cpp; r=alfredo
08:17pulsebotCheck-in: - Gabriele Svelto - Bug 1345654 - Enable client-side stack-walking on beta and release for main process crashes r=bsmedberg a=gchang
08:18pulsebotCheck-in: - JerryShih - Bug 1310681 - fix the missed closing parenthesis. r=me
08:18pulsebotCheck-in: - Chih-Yi Leu - Bug 1315037 - Use NS_ENSURE_SUCCESS for nsresult diagnosis. r=erahm
08:18pulsebotCheck-in: - Ya-Chieh Wu - Bug 1338476 - Add &quot;pdfium.enabled&quot; as a pref for turning on/off mortar pdf. r=evelyn, r=Gijs
08:18pulsebotCheck-in: - Samael Wang - Bug 1339707 - Replace NS_DispatchToCurrentThread to DocGroup / TabGroup dispatch method and set nsITimer target. r=smaug
08:22pulsebotCheck-in: - Yoshi Huang - Bug 1300671 - set firstPartyDomain on about: pages. r=smaug
08:24pulsebotCheck-in: - Yoshi Huang - Bug 1301649: Use MOZ_ASSERT_IF instead. r=smaug
08:25gastonanyone knows whats the longterm plan with pdf.js now that i see pdfium commits ?
08:26pulsebotCheck-in: - Honza Bambas - Bug 1320458 - Make logging by sandboxed child processes to a file work on Windows, r=aklotz a=gchang
08:26pulsebotCheck-in: - Michael Layzell - Bug 1337073 - Improve the precision of the IPC_SYNC_LATENCY_MS probe and rename it to IPC_SYNC_MAIN_LATENCY_MS, r=billm a=gchang
08:26pulsebotCheck-in: - vincentliu - Bug 1343851 - Add nullptr check for SurfaceFactory in StartVRPresentation. r=mtseng a=gchang
08:26pulsebotCheck-in: - Mats Palmgren - Bug 1344395 - Make checkbox/radio controls have a default (non-zero) size also when -moz-appearance:none on Android/Gonk (since that&#39;s the default on these platforms). r=dholbert a=gchang
08:26pulsebotCheck-in: - Mats Palmgren - Bug 1344395 - Reftest for checkbox/radio controls with width/height:auto. a=test-only
08:26pulsebotCheck-in: - Shu-yu Guo - Bug 1345145 - Remove some frontend tracelogging that&#39;s in functions that are too hot, causing regressions, and whose per-call execution times are too low to be useful. r=h4writer a=gchang
08:26pulsebotCheck-in: - Makoto Kato - Bug 1346098 - Don&#39;t expose intlUtils without ENABLE_INTL_API. r=smaug a=gchang
08:36Tomcat|sheriffdutyewong: ^ this affects also tbird trees
08:37ewongTomcat|sheriffduty: thanks for the heds up
08:39ewongTomcat|sheriffduty: the TCW is now?
08:40Tomcat|sheriffdutyewong: made this more a little more clear :)
08:41ewongTomcat|sheriffduty: ah. thanks. coz the bug didn&#39;t say when.. and I only just pushed something to try.. oh well.. guess I&#39;ll fix something else
08:41Tomcat|sheriffdutyewong: yeah linked now also the annoucement in the bug :)
08:42* ewong wanders off to some other stuff
08:45kanrufirebot: seen karlt
08:45firebotkarlt was last seen 6 hours and 57 minutes ago, talking in #auckland.
09:00bakuTomcat|sheriffduty: good morning. eta for m-i?
09:01Tomcat|sheriffdutybaku: if the 90 minute estimate from jgraham was right it should take another 60 minute
09:02bakuTomcat|sheriffduty: ok thanks
09:03jgrahamYeah, not 100% sure about that because a) there was some initial overhead I forgot to count and b) there ought to be less data than when I did a trial run
09:03jgrahamHopefully those things exactly cancel out
09:59jyawhat are win8 treeherder machine running ?
10:00jyaI get a permanent decoding failure there, but I can&#39;t reproduce it in a win8 VM I&#39;ve setup
10:02Tomcat|sheriffdutyjya: the #releng guys might know
10:02jyaasking there.. thanks
10:05pulsebotCheck-in: - Alfredo.Yang - Bug 1343461 - add a wrapper class to encapsulate rust/stagefright sample table. r=gerald
10:05gandalfany DOM peer online?
10:05pulsebotCheck-in: - Zibi Braniecki - Bug 1346674 - Migrate all uses of nsILocaleService::GetApplicationLocale to mozILocaleService::GetAppLocale. r=jfkthame
10:06padenotbaku, ^ ?
10:06bakugandalf: here
10:07pulsebotCheck-in: - JW Wang - Bug 1343459. Part 1 - Label runnables under dom/media/mediasource. r=jya
10:07pulsebotCheck-in: - JW Wang - Bug 1343459. Part 2 - remove the &#39;updateend&#39; handler which might fire before we register it. r=jya
10:07gandalfbaku: can you take a look at bug 1346819 ?
10:07gandalfI&#39;m updating a name of chrome only API and need to update Window.webidl
10:08gandalfcan I get a require r= from you on this?
10:09bakugandalf: r+
10:10gandalfbaku: I think I need it in the bug
10:10bakugandalf: done
10:11gandalfthank you
10:11gandalfTomcat|sheriffduty: there will be a build or two broken on autoland because of the bad order of landings
10:11Tomcat|sheriffdutygandalf: ok
10:11pulsebotCheck-in: - Zibi Braniecki - Bug 1346819 - Port SanitizeAsBCP47 to LocaleService. r=baku,jfkthame
10:11gandalfbug 1346819 landed after bug 1346674 :(
10:12gandalfsorry for that!
10:12freddybis there a way to maintain order with autolanding?
10:36pulsebotCheck-in: - Andrea Marchesini - Bug 1346718 - Add a check before using a BlobParent actor, r=qdot
10:38pulsebotCheck-in: - Andrea Marchesini - Bug 1346801 - No need to dispatch runnables to the main-thread for Entries API, r=qdot
10:38pulsebotCheck-in: - Andrea Marchesini - Bug 1346801 - Make Directory object always cloneable, r=qdot
10:49pulsebotCheck-in: - Zibi Braniecki - Bug 1346617 - Add LocaleService::SetRequestedLocales. r=jfkthame
11:29pulsebotCheck-in: - Do Gottwald - Bug 1343330 - Use fast-forward and fast-rewind characters instead of skip-to-end and skip-to-start on Windows 7. r=Felipe
11:30pulsebotCheck-in: - Andrea Marchesini - Bug 1274343 - Add parent-to-child pipe streaming to IPCStream - part 1 - renaming PSendStream to PChildToParentStream, r=smaug
11:30pulsebotCheck-in: - Andrea Marchesini - Bug 1274343 - Add parent-to-child pipe streaming to IPCStream - part 2 - PParentToChild, r=smuag
11:31DexterIs there any way to sign requests coming off a Firefox client? Or.. how can we make sure, on a service that we own, that requests come from legit Firefox clients? Is there any prior example for that in Firefox?
11:42Tomcat|sheriffdutydao: es lint failure
11:43pulsebotCheck-in: - Carsten &quot;Tomcat&quot; Book - Backed out changeset 8cd4168dbb04 (bug 1343330) for eslint failure
11:45WG9sDexter: Your signing idea has side-effects. If a site only accepts requests for a signed version, then you, as a devopers, would not be able to test a fix for an issue with that site using either your own build or a try build.
11:46DexterWG9s, I see
11:46Dexterthanks for chiming in
11:46bagdernot to mention the question what a &quot;legit&quot; version is
11:46Dextergood points
11:46Dexterso, let me start from the problem
11:47WG9sTHat to is a nightly build legit or not?
11:47Dexterit is
11:47pulsebotCheck-in: - 1 changesets - servo: Merge #15847 - Remove link to servo from treeherder job detail to reduce noise (from shinglyu:treeherder-cleanup); r=Wafflespeanut
11:47DexterSo, what I would like to do is making sure that requests that come to the service we&#39;re building are really coming from a Firefox client
11:47Dexterrather than brute-forced through curl, for example
11:55WG9sDexter: Ah I thought you were more worried about a hacked version of Firefox.
11:56Dexterwell, that too
11:56Dexteror it&#39;s probably safer to say that I&#39;m not sure what I should be worried about :)
11:57WG9sDexter: what you are seeing is one ofthe reasons that pople try to put security on any client-server type application on the client side are just being lame.
11:59bagderDexter: most sites do funky javascript dances to attempt to make sure of that
11:59bagderand curl users do funky things to work around said dances =)
12:03WG9sDexter: but, one way to help if it is really a security thing that must be enforced on the client side sould be requiring the clinet to provide a vaid certificate, but it would have to be a certificate provided by the server software vendor. this does not really work for the browser case. what you do on the server side should really be sufficient even if the client mis=behaves. of course... could somehow require the browser to have an add-on and have be requird to be signed by you.
12:03Dexterbagder, WG9s thank you both, this discussion is really interesting :)
12:04bagderhey, I write curl, I have thousands of users working around those things all the time =)
12:04Aryxare people trying to poison the data or do some scanning?
12:04hsivonenYoric: is handling a complete file or a part of a file per call?
12:04maliuping :sebastian
12:04DexterAryx, potentially, yes
12:05pulsebotCheck-in: - Jan de Mooij - Bug 1341071 - Fix Ion IC post barriers to be part of the IC stubs instead of adding them in IonBuilder. r=jonco,nbp
12:06sebastianmaliu: pong
12:08pulsebotCheck-in: - Do Gottwald - Bug 1343330 - Use fast-forward and fast-rewind characters instead of skip-to-end and skip-to-start on Windows 7. r=Felipe
12:08maliuI was trying to enable activity-stream and installed switchboard. Seems &quot;about:experiments&quot; page is broken on nightly?
12:10Yorichsivonen: Normally, complete file. There would be ways to abuse it into handling parts of a file per call, but I hope that nobody does that.
12:10pulsebotCheck-in: - Rahul Chaudhary - Bug 1342002 - Border color of overriden items in Rule view looks lighter than font color in dark theme. r=gl
12:11hsivonenYoric: is there some other call that always happens on EOF (in case I don&#39;t assume EOF in that method)?
12:13Yorichsivonen: Are you interested in EOF or in the fact that the file descriptor is closed?
12:13YoricI don&#39;t think that we actually have anything that is guaranteed to be EOF. But that line is only called after closing the file descriptor, if that&#39;s what you want.
12:14YoricThe reason for which we don&#39;t guarantee EOF is that it can be used to read only a slice of a file, e.g. a prefix.
12:15Yorichsivonen: On an unrelated note, how much effort do you want me to put into making bug 1344152 SSE? I have started learning SSE.
12:16sebastianmaliu: Oh yeah, seems like I need to switch to the new messaging API.
12:17hsivonenYoric: at the moment, I don&#39;t see bug 1344152 as necessary to land at this time with or without SSE2. But if it lands, SSE2 would be nice. I should be very few SSE2 ops as outlined in my comment.
12:17YoricAh, I haven&#39;t seen your comment.
12:18hsivonenYoric: I&#39;m interested in EOF. The current usage of our Unicode decoders just misbehaves if the file ends with an incomplete byte sequence in a multi-byte encoding
12:18hsivonenYoric: since encoding_rs support correct behavior, I&#39;d like to make the behavior correct where possible
12:18hsivonen(I&#39;ve already located at least one legacy interface where not possible)
12:19hsivonenYoric: I can retain the bug, of course, for now.
12:20maliusebastian: Can I help this one? Any good example of the api migration?
12:21sebastianmaliu: Sure. If you want to go ahead. I probably won&#39;t have time for this currently anyways.
12:21sebastianmaliu: Definitely a good example why a stable add-on api would be nice (webextensions) - however this add-on is using some internal apis to change the experiments and this won&#39;t be available as webextension API unfortunately.
12:22Yorichsivonen: If you want to check for EOF, this will need some patching.
12:26pulsebotCheck-in: - Sheldon Roddick - Bug 1307481 - Part 4: Implements the Inline Tooltip widget. r=jdescottes
12:26hsivonenYoric: OK. I&#39;ll leave it for a follow-up
12:26hsivonenYoric: thanks
12:30Yorichsivonen: np
12:36smaugtimdream: ping
12:36smaugtimdream: is Bug 1338961 enough for now?
12:36smaugtimdream: just thinking if we should design properly what to do with bug 1327097
12:36smaugwhat blink is doing is rather bad
12:37timdreamsmaug: yes, I believe bug 1338961 is enough for Fx53.
12:38smaugtimdream: bug 1327097 sounds like something to discuss with other browser vendors
12:39smaugI&#39;m pinging mounir in #whatwg, perhaps he has ideas how he would like to see this work in blink
12:39timdreamsmaug: I am fine if we want to design bug 1327097 properly. But I don&#39;t want to leave it there forever though.
12:40timdreamThe fact that we emit events and ignore its default prevented would be a web compat issue
12:40smaugit just feels that blink added some hack there without thinking it much
12:40pulsebotCheck-in: - jordan9769 - Bug 1311347 - Enable eslint of browser/components/sessionstore/. Initial changes by Sourav, updated by Standard8. r=jaws
12:40smaugthis is a bit related to form control elements too
12:40pulsebotCheck-in: - Mark Banner - Bug 1311347 - Enable eslint of browser/components/sessionstore/. Autofix changes. r=jaws
12:40pulsebotCheck-in: - Mark Banner - Bug 1311347 - Enable eslint of browser/components/sessionstore/. Manual fixes. r=jaws
12:40smaugthey have a bit weird event handling
12:42smaugtimdream: do we need to suppress *move events?
12:43smaughmm, for range, I guess
12:43timdreamsmaug: yes, because preventDefault() move events will cancel dragging of the seeker.
12:45smaugtimdream: would be it possible to suppress event only if &quot;user input transaction&quot; started at the video element. So, only if *down event happened on <video>, then we&#39;d suppress following moves and up
12:47timdreamsmaug: I have no strong opinion on that.
12:48smaugtimdream: let me try get some more feedback from blink folks
12:48smaugtimdream: I keep the patch in my review queue, but might not review it today
12:48* timdream ttyl
12:54pulsebotCheck-in: - Julian Descottes - Bug 1338106 - hide toolbox toolbar icons for narrow viewport;r=ntim
12:54pulsebotCheck-in: - Julian Descottes - Bug 1338106 - reduce min width of devtools toolbar command buttons to 24px;r=ntim
12:54pulsebotCheck-in: - Julian Descottes - Bug 1338106 - Add mask-image to fade out text of devtools-tabs;r=ntim
13:09pulsebotCheck-in: - Alin Selagea - Bug 1344297 - Turn off buildbot static analysis jobs on linux64 debug and macosx64 debug, r=kmoir
13:10mounirtimdream, smaug: i&#39;m here too if you want :)
13:22pulsebotCheck-in: - 132 changesets (bugs 15870, 929535, 1346812, 1335966, 1343640 and 62 other bugs)
13:23pulsebotCheck-in: - 36 changesets - merge mozilla-inbound to mozilla-central a=merge
13:26pulsebotCheck-in: - 134 changesets - Merge mozilla-central to mozilla-inbound
13:37pulsebotCheck-in: - 37 changesets - Merge mozilla-central to autoland
13:39beekillhi everyone
13:40beekillwhat is responsible for showing the spinner in the tab header when firefox is loading webpage?
13:44Gijsbeekill: 2 attributes set on the tab element, and then XBL propagates those
13:45beekillGijs: thanks, but how to set those attributes?
13:46Gijsbeekill: what&#39;s the context in which you&#39;re asking? In most cases you shouldn&#39;t need to set those yourself...
13:47RyanVMjya: OK, I&#39;m confused RE: bug 1345545
13:47RyanVMoh, nevermind
13:47RyanVMI see it now
13:47RyanVMbug 1299105 is the one I was thinking of
13:48beekillGijs: it&#39;s actually when firefox is restoring from previous session. I want to disable the spinner when we&#39;re restoring. But I don&#39;t see which code do that
13:49Gijsbeekill: we should already not be doing that -
13:50Gijsbeekill: the two attributes are &quot;busy&quot; and &quot;progress&quot;, fwiw
13:51beekillGijs: thanks :)