mozilla :: #content

21 Apr 2017
07:39bakukanru: ping. So far a Blob doesn't know its principal. This is something that we can definitely add. And we can send it to the parent process when we move blobs across processes.
07:45kanrubaku: I see. I filed bug 1358285
07:45firebothttps://bugzil.la/1358285 NEW, nobody@mozilla.org Would be nice if MemoryBlobImplDataOwnerMemoryReporter can report the file's origin
12:46smaugbaku: does IPCBlob have Move ctor?
12:47bakusmaug: it has a copy ctor.
12:48smaugok, so using Move() doesn't make much sense, I guess
12:56farrebkelly: ping
12:56bkellyfarre: hi
12:56farrefarre: I know your queue is closed, but I thought I'd ask about bug 1339909 anyway
12:56firebothttps://bugzil.la/1339909 NEW, afarre@mozilla.com Avoid ThrottleTrackingTimeoutsCallback keeping the Window object alive
12:57bkellyfarre: I would just like to cancel the timer if the inner window is not active any more
12:58bkellyfarre: and I think it would be fine if the inner window goes into the bfcache before this timer fires, then you can just reschedule the timer from scratch when it comes out of bfcache
12:58bkellyIMO
12:59bkellyfarre: oh, are you asking me to review?
12:59farrebkelly: yep :) I do it as you suggest
13:00bkellyfarre: I'm not at my computer yet, but I can look later this morning
13:00farrebkelly: np
13:00bkellyfarre: can you just NI me?
13:00farrebkelly: will do
13:01farrebkelly: thanks
13:02smaugfarre: and sorry, my queue is closed. I'm trying this "keep queue open only when I'm not working"
13:03smaug(just in case you were trying to get review from me first)
13:04farresmaug: also not a problem. btw, do you know if we have a bug for idleDispatch with timeout?
13:04smaugfarre: I don't think so. Ehsan made some other bug depend on the parser bug I filed
13:05smaugbut would be better to have some new bug
13:05* farre is filing
13:53bkellyso, my phone with my totp authenticator died... might lose duo/bugzilla access for a while (when my current sessions expire)
14:04smaugmccr8: I don't think the ghost issue is addon related
14:04smaugI can see it now
14:04smaugin new nightly
14:04smaugdon't recall which nightly I was using before
14:33bkellywow, firefoxos 2.5 is better than I remember it to be
14:38bkellyasuth: are you still taking mail app bug reports?
14:39asuthbkelly: People keep wontfixing everything in the component, but I'm open to hearing about hijinks.
14:39asuthbkelly: You're probably experiencing a very unhappy backfill problem if it's been years.
14:40bkellyasuth: no... just a glitch when it needs to reauth multiple accounts... the second account wouldn't start the reauth process... killing the app and restarting fixed it
14:40bkellybut I can't auth my mozilla account because my duo authenticator is on my dead phone :-(
14:41asuthbkelly: oh, huh, yeah, that might work better in 2.6.
14:41bkellyits strange to say this, but I feel like the keyboard on this experia with fxos 2.5 is better than the keyboard with android on my n5x
14:42bkellymaybe because its had so long to rest
16:20smaugmccr8: curious, are you perhaps also trying to reproduce the ghost window issue?
16:21smaugbkelly: I really liked ffos2.5. Easier to use UI than in Android
17:04smaugmccr8: so far no luck reproducing the ghost issue
17:04mccr8smaug: yeah I haven't been able to repro it
17:45ehsanasuth, ping
17:45asuthehsan: pong
17:46ehsanwhat's a .metadata-v2 file under profile/storage
17:46ehsanand what's a ._.metadata-v2 file there?!
17:46bzehsan: ping
17:47ehsanbz, hi
17:47bzehsan: So I'm trying to make sense of https://people-mozilla.org/%7Emlayzell/bhr/20170405.html
17:47ehsanbz, yes
17:47bzehsan: so I looked at the fourth thing on the list, the one that has nsRefreshDriver::Tick
17:47ehsanasuth, see above ^
17:47ehsanbz, yep
17:47bzehsan: And that takes me to https://people-mozilla.org/%7Emlayzell/bhr/20170405/3.html
17:47bzWhere there are a bunch of stacks, many of which don't involve nsRefreshDriver::Tick
17:48ehsanbz, correct
17:48bzok
17:48ehsanbz, did you note the description about how these are captured in my email? :)
17:48bzSo what am I looking at? ;)
17:48ehsanhere is what happens
17:49ehsanfirst we notice that the main thread has been unresponsive for 128ms
17:49ehsanthen we capture a pseudostack trace
17:49ehsanthis is what you see first
17:49ehsan(the RefreshDriver::Tick one)
17:49ehsanthen we wait
17:49bzaha
17:49ehsanafter 8 seconds of unresponsiveness we capture a native stack
17:49ehsanwhich is unrelated (probably) to the former trace
17:49bzI see
17:50ehsanbz, mystor is fixing this btw
17:50bzok
17:50ehsanbut this is just how things are at the moment
17:50asuthehsan: the metadata files store a custom binary serialization of the origin and their timestamp serves as the last access time of the directory
17:50bzmost of these native stacks are just stting there under WinUtils::WaitForMessage
17:50asuthehsan: v2 is the new post-app removal format
17:50ehsanasuth, yeah... pay close attention to the name of the second file :)
17:50bzAs in, "we just have nothing to do"
17:50mystorbz: After 1346415 we're going to capture them at the same time
17:51asuthehsan: you should not have a metadata file in the root of your storage/ directory, if you're seeing one. And it shouldn't have a "._" prefix in front of it. The temporary file name these days is .metadata-v2-tmp
17:51ehsanasuth, this is a profile I moved from an old machine, just wanna make sure if I screwed something up or if gecko did
17:51mystorbz: (though not in the exact same pausing of the thread yet, I'm workinf on fixing that in a follow-up right now)
17:51ehsanbz, yes that is another stupid issue we should fix
17:51mystorbz: (on nightly specifically)
17:51bzok
17:52ehsanasuth, does gecko ever create files with the "._" in the front?
17:52asuthehsan: I'm not aware of any temporary file creation logic that would think that's a good idea...
17:52ehsanasuth, it's not a good idea, cause right now it breaks all idb unfortunately
17:53ehsanasuth, can I delete all of these files?
17:53ehsanasuth, (I also don't understand why the idb code is so scared of files it doesn't recognize, and is just willing to completely refuse to load any idb whatsoever as soon as it sees such a file...)
17:53asuthehsan: yes, although, obviously, the state of your profile is maybe a bit concerning. it wouldn't be surprising if there's other things in there that make quota manager freak out
17:54ehsanasuth, yeah... we'll see, but I need to first undig myself out of this hole :(
17:54ehsanasuth, it'll be interesting if I end up with any data left after this is finished :(
17:54ehsansigh
17:54* ehsan picks up a shovel and walks towards his profile directory
17:56ehsanasuth, ah here is a fun discovery
17:56ehsanasuth, when I cat the ._.metadata-v2 file, I see some garbage characters with "Mac OS X" inside...
17:56ehsanand this is a profile moved from OSX to Linux
17:56asuthehsan: yeah, https://searchfox.org/mozilla-central/source/modules/freetype2/src/base/ftrfork.c#568 seems to use it for resource fork stuff?
17:56ehsanso I bet this is a hidden file of some sort that osx creates
17:57ehsanso if you copy a profile from osx to linux (and presumably windows) you'll lose all idb
17:57asuthIDB knows to ignore certain OS specific stuff
17:57asuthMaybe we need to teach it a few more things
17:58ehsanasuth, should I file a bug?
17:58asuthehsan: yeah, and if you want to hack it up... https://searchfox.org/mozilla-central/source/dom/quota/ActorsParent.cpp#1413
17:59ehsanasuth, I can just delete these files, no?
17:59ehsan(I'm running a moco nightly)
17:59asuth(noting that IDB probably needs to be taught too... https://searchfox.org/mozilla-central/source/dom/indexedDB/ActorsParent.cpp#18326)
17:59asuthehsan: yeah, you can delete anything matching a "._" spec for sure
18:00ehsanasuth, this error comes from http://searchfox.org/mozilla-central/source/dom/quota/ActorsParent.cpp#4255
18:00ehsanasuth, ok that unblocks me for now, thanks very much :)
18:00* ehsan files the bug
18:03ehsanasuth, https://bugzilla.mozilla.org/show_bug.cgi?id=1358571
18:03firebotBug 1358571 NEW, nobody@mozilla.org Copying a profile from OSX to Linux (and probably Windows) will cause the user to lose all IDB data
18:08asuthehsan: https://lists.apple.com/archives/applescript-users/2006/Jun/msg00180.html and https://apple.stackexchange.com/questions/14980/why-are-dot-underscore-files-created-and-how-can-i-avoid-them suggest those files come into existence when the underlying FS doesn't support the meta-data, also possibly you used finder to look at the dir.
18:08asuthehsan: anything special about your FS in use? and/or are you willing to confess to having used a GUI?
18:17ehsanasuth, I use HFS+ (the thing that came with the machine, I never changed it)
18:17ehsanasuth, and I mostly have used iTerm to move in my profile dir but I had the machine for 2 years so I may have had Finder open in some of these folders at some point :)
18:18ehsanasuth, since these files were only created in ~20 or so folders the second theory is pretty likely
18:19smaugmccr8: hmm, almost if we aren't calling nsDocument::Destroy. I have a CC log, but still don't know how to reproduce
18:19asuthehsan: This sounds like another cat related bug. Dragging and dropping .metadata-v2 files between directories. I'm sure you purged the rogue file by now, but it would be interesting to know what origin was in the file.
18:19mccr8smaug: weird
18:20smaugoh, hmm, ::destroy doesn't unbind anymore
18:22ehsanasuth, I've attached one file to the bug. the rest still exist on my old machine
18:23ehsanyou can ni? me if you need them
18:23ehsanthere is also ._.metadata files btw
18:25asuthehsan: no, just idle curiosity for identifying where the profile/storage/.metadata-v2 which shouldn't have been there came from. Unless I miss-parsed and you never had one there, just the ._ variants in reasonable locations in the profile dir.
18:30ehsanasuth, I can't say that I haven't
18:45bkellydid moz sf lose power today?
18:56mccr8that's a very specific question
19:17bkellymccr8: apparently a lot of sf lot power today... just wondering if our office was caught in it
19:53mccr8Ah, yeah I just saw that.
20:01smaugmccr8: hmm, I think I need to try to do some regression range hunting using my main profile
20:01smaugghost windows do show reasonable easily
20:01mccr8smaug: yeah if you can reproduce it, that would be great.
20:01smaugbut still no idea how to reproduce
20:02smaugI mean, I can see them in my normal profile
20:02smaugwhich has hundreds of tabs etc, not all loaded
20:02smaugbut trying to reproduce using a new profile... no luck
20:34bkellysome ipdl change has totally broken my patches :-(
20:34bkellyerror: &#39;Write&#39; is not a member of &#39;IPC::ParamTraits<mozilla::gfx::RectTyped<mozilla::CSSPixel> >&#39;
20:34bkellywtf... I don&#39;t care about gfx types
20:34bkellyI really should have taken today off
20:38gandalfqDot: if you, for whatever wrong reason, wander here, there&#39;s a minor update to the client/server patch that I&#39;d like you to take a second look at before I land it. :)
20:39gandalfqDot: and if you didn&#39;t want me to bother you, you shouldn&#39;t be on IRC during PTO :p
20:51bkellybillm: is it normal that searchfox hasn&#39;t updated since yesterday?
20:52billmbkelly: you might need to reload. the date I&#39;m seeing on pages is from today.
20:52bkellyah yes
20:52bkellyinteresting
20:53bkellyI had never visited this page in searchfox before... wonder how it got cached
20:53billmbkelly: the update happened half an hour ago
20:53billmbkelly: actually, it might have gone live pretty recently
20:53bkellycool
20:53bkellyso I&#39;m not (completely) crazy
20:54billmbkelly: a while back I switched things so it would update while I was actually at work. that way if something breaks I can fix it right away :-). I learned that the hard way.
20:54billmbkelly: it kinda sucks for everyone else though since Pacific is basically the last time zone
20:55bkellyah
20:55bkellybillm: you could get a commitment from asuth to help and do it in eastern time zone :-)
20:56bkellyalthough his sleep schedule may match pacific
20:57billmbkelly: that&#39;s true. somehow I need to figure out how to give other people AWS credentials first though.
20:58asuthI&#39;m pretty sure you can crowd-source it by committing the username and password to github.
21:09billmasuth: heh. by the way, I&#39;m really sorry I haven&#39;t gotten back to the context menu patch. I got really busy with stuff and haven&#39;t had any time for searchfox.
21:10asuthbillm: Yeah, that&#39;s about where I am too. Have been meaning to follow through on my proposal...
21:10asuth(a &quot;uses&quot; straw-man UI for the mega menu)
21:17* bkelly finally gets his tree rebuilt 15 minutes before the end of the day.
21:17bkellywell, end of the week :-\
22 Apr 2017
No messages
   
Last message: 7 days and 4 hours ago