[00:14] hmmm, after looking at scrollbar stuff, i really want to poke my eyes out already [01:29] chrisccoulson: did you want the 9.0 final builds to go to precise or are we just uploading 10 beta 1 when they're tagged? [04:44] micahg: All my ppa build completed successfully for seamonkey 2.6. I sent up builds for lucid through precise. [04:44] joelesko: that's good news :), that means we should be ready to go on tuesday for precise at least and maybe oneiric [04:44] micahg: Let me know where the test builds are when you create them. [04:45] micahg: I'm using it on maverick and oneiric and not running across any problems. Is there some test scripts you use? [04:45] joelesko: ok, precise I'll just upload, ubuntu-mozilla-security will be where the oneiric build will go if I don't run out of time, otherwise, we can push next week when I'm off [04:46] micahg: great. thanks. I sure would like to close those open bugs. [04:47] joelesko: yeah, we use test-browser.py from lp:qa-regression-testing (you need to run it in a clean env as it'll blow away your profile), I need to make a smaller set of test cases for the universe packages like chromium and seamonkey [04:48] joelesko: take a look at this blueprint which has the list of the limited test cases: https://blueprints.launchpad.net/ubuntu/+spec/security-p-catch-all, it's under security roundtable wednesday [04:48] joelesko: you'll need to test the binaries that are produced in the PPA though since that's what will be released [04:49] micahg: That's the plan. I will pull then when ready. I have time this week. [04:49] cool [04:50] lightning is current broke, right? [04:50] shouldn't be [04:50] you'll need to use the version from the thunderbird-next PPA [04:51] I asked someone else because I didn't see the xpi on https://developer.mozilla.org/en/Calendar/Calendar_Versions [04:51] https://ftp.mozilla.org/pub/mozilla.org/calendar/lightning/tinderbox-builds/comm-beta/ [04:52] thanks. Is thunderbird schedule for release soon as well? [04:52] we have 1.1 built in thunderbird-next (will be in ubuntu-mozilla-security tomorrow sometime) [04:52] er, only for oneiric that is [12:48] any news concerning chromium translations? https://bugs.launchpad.net/ubuntu/+source/chromium-browser/+bug/888617 [12:48] Launchpad bug 888617 in chromium-browser "Chromium translations on LP no longer updated" [Undecided,Confirmed] [12:48] I remember we spoke that once daily builds are active, translation templates would be next === mdeslaur_ is now known as mdeslaur === micahg_ is now known as micahg [14:03] artnay: I have no idea what's involved in the chromium translations [14:07] chrisccoulson: hi. I thought I had spoken with micahg but it must have been you. [14:07] chrisccoulson: 14:48 artnay : any news concerning chromium translations? https://bugs.launchpad.net/ubuntu/+source/chromium-browser/+bug/888617 [14:07] Launchpad bug 888617 in chromium-browser "Chromium translations on LP no longer updated" [Undecided,Confirmed] [14:07] chrisccoulson: 14:48 artnay : I remember we spoke that once daily builds are active, translation templates would be next === bdrung_ is now known as bdrung [14:51] artnay, if i can do anything about this i'll take al ook at it today [15:07] bjsnider: that would be great, thanks in advance. [18:14] micahg, alright it seems that the change i mentioned recently does result in working chromium dbg packages, and it also seems like that's the only way to get it done, ie there's no more build flag. [19:40] timeless: hi! [19:45] timeless: I blame the slowness on resource:///components/nsPrompter.js:68 and chrome://global/content/bindings/browser.xml:273 on IRCCloud and Grooveshark fighting over Flash [19:46] timeless: I've fetched firefox-trunk-mozsymbols [19:52] can you figure out if/what it did, or what it claims to do? [19:52] (ask it for a description/contents) [19:57] timeless: sorry crashed Flash prevented me from seeing your messages [19:57] can you figure out if/what it did, or what it claims to do? [19:57] (ask it for a description/contents) [19:57] timeless: any chance you can get a fix for https://bugzilla.mozilla.org/show_bug.cgi?id=699133 so I can at leas submit the crash? [19:57] Mozilla bug 699133 in General "No "Send crash report" link for Flash crashes on non visible Flash elements." [Normal,New: ] [19:58] "This package contains the Firefox symbols in a format expected by Mozilla's [19:58] Breakpad. Eventually this package should go away and the symbol upload be [19:58] implemented in soyuz (or other builders that build this package)" [19:59] sorry on the multi-line [19:59] doesn't bother me [19:59] ok, can you figure out if it really has symbols? if it does, the contents should be significant in terms of size [20:00] on this very channel I asked why the builds from this team PPA don't have about:crashes though, so that's another thing I'd need [20:01] if the package gives you symbols, you can use gdb [20:01] which is good enough for most things [20:01] Uncompressed Size: 142 k [20:01] the package should be >100mb if it's useful [20:01] Firefox can't fix a flash crash anyways, why would you want to report it? [20:01] what is it actually? [20:02] micahg: firefox is software, mozilla is an entity [20:02] timeless: right, sorry, Mozilla can't fix a flash crash :) [20:02] mozilla as an entity can help adobe review crash reports and fix bug reports [20:02] *if* they're collected [20:02] ah [20:03] similarly, microsoft will help just about any software vendor review and fix reports in their crashing software [20:03] (via WER) [20:03] micahg: IIRC there was a time were an "electrolisys" project existed and they valued the plugin-container crashes [20:03] alex_mayorga: the project still exists [20:03] although it's mostly focused on the next step which is irrelevant for flash crashes [20:03] but yes, plugin crashes are still valuable [20:03] but that's offtopic [20:04] we're here to try to get you symbols so we can get js stack traces for slow scripts w/o using firebug/venkman :) [20:04] alex_mayorga: plugin-container crash != flash crash [20:04] so, i don't care how few bytes the package has, figure out what files it has [20:04] micahg: sure, but 99% of the time it is the plugin that's crashing, and on linux, that's probably flash :) [20:16] if you want symbols, you need firefox-dbg or firefox-trunk-dbg [20:16] those are about 180MB ;) [20:31] * alex_mayorga catches up as he got his first Pangolin crash [20:33] doing sudo apt-get install firefox-trunk-dbg [20:33] chrisccoulson: you told me the other day no about:crashes for the PPA, right? [20:34] alex_mayorga, yes. but we still have symbols if you want to use gdb [20:35] chrisccoulson: meet timeless of Mozilla's fame, perhaps something could be worked out to get the crashes on Socorro somehow [20:36] timeless: firefox-trunk-dbg is installed now, what would be my instructions? [20:41] for getting them uploaded, they should talk to ted / moznet [20:41] but, now that you have the -dbg package (yay), let's visit gdb [20:42] do you still have a slow script, or are we just doing it for practice? [20:42] (we can get a slow script if we need one) [20:42] timeless: I'm pretty sure I can reproduce the slow script if I try [20:43] ok, for now, let's ignore that, do you have firefox running? [20:43] (and can you get a terminal? :) [20:43] timeless, oh, i can upload them already. it's just that i'd be submitting about 400MB of symbols per day if i did it for all of our builds, which is why we only do it on beta + release builds currently [20:43] timeless: yup! [20:43] chrisccoulson: out of curiosity, do you pay for bandwidth? [20:43] (really curious) [20:44] alex_mayorga: alright, in the terminal: [20:44] ps aux|grep firefox-bin [20:44] * timeless has no idea how magically hacked the ubuntu firefox is [20:44] we'll see shortly :) [20:45] timeless, i don't pay for the bandwidth, and i upload the symbols automatically from a job in our datacenter anyway [20:45] timeless: edited a bit to ps aux|grep firefox-trunk [20:45] i'm more concerned about disk usage on the other end :) [20:46] alex_mayorga: don't :) [20:46] really, if they're underprovisioned, they need to fix it anyway [20:46] timeless: the ps you suggested comes back blank [20:46] they need to have different aging policy for different kinds of packages (nightlies, alphas, etc) [20:46] and i'm fairly certain they do [20:47] getting crash reports for your builds would enable them to understand more about how they're used too [20:47] if I use firefox-trunk I get 2373 as my main suspect [20:47] alex_mayorga: ok [20:47] gdb -p 2373 [20:47] timeless: bear in mind I'm using "Nightly" from ppa here [20:47] at this point, a piratepad will be helpful [20:47] alex_mayorga: yeah, sure [20:47] my steps are designed to let me slowly work out the quirks of your build [20:48] e.g. there's no -bin file, and thus there's no .sh script which would accept magic commands [20:48] if you used a mozilla official thing, we'd use firefox -d gdb -g [20:48] timeless: I see [20:48] (or specify a different debugger w/ -d ...) [20:48] you should be able to do that with ours too [20:48] although, we rename the binary for our nightly build to make it coinstallable with the release builds [20:49] (so, firefox-trunk -g will work fine) [20:49] ok, so, in that case, 2373 may be the wrong answer [20:49] if you have a firefox-trunk-bin [20:49] we generally want the -bin process if one exists [20:51] timeless: got some form of permission thing with your command, sudo gdb -p 2373 did the trick [20:51] 0x00007fa090e8c753 in __GI___poll (fds=, nfds=, [20:51] timeout=) at ../sysdeps/unix/sysv/linux/poll.c:87 [20:51] 87 ../sysdeps/unix/sysv/linux/poll.c: No such file or directory. [20:51] in ../sysdeps/unix/sysv/linux/poll.c [20:52] alex_mayorga: so... [20:52] if you're debugging a shell script, the wrong thing happened [20:52] firefox-trunk is grayed out, locked not responsive to mouse or keyboard input [20:52] and it's moderately painful for me to figure out if that's the case [20:52] i'd typically use pstree |grep -C 10 2373 [20:52] if there are children other than plugin-container, then it's the wrong process) [20:53] micahg, did you forget to push a commit to lp:~mozillateam/thunderbird/thunderbird.oneiric? [20:53] (ie, the last release tag) :) [20:54] alex_mayorga: anyway, try "info shar" [20:54] * timeless tries to remember how to speak gdb w/o finding a linux env [20:55] timeless: http://piratepad.net/c3berohKLm [21:05] how do I get the dbg of firefox-trunk dependencies "automagically"? does anyone know? [21:06] (i've definitely seen/used a script that did it, but it was ages ago) [21:13] lol, i can hear my daughter snoring over the baby monitor [21:13] that's a good thing [21:13] chrisccoulson: lovely :) [21:13] yeah, i feel more relaxed when i can hear her ;) [21:14] earlier today i could hear someone else's daughter wailing through the corridor (much less pleasant) [21:14] yeah, that's not so nice :( [21:25] timeless: done fetching -dbgs here [21:27] :) [21:27] ok, time to get a slow script dialog [21:27] you can be lazy and use this one: [21:27] javascript:while(1); [21:28] (hardly useful, but hey...) [21:30] anyway.. you should get a slow script dialog for that (i did) [21:30] timeless: that in the URL bar? [21:30] sure [21:32] timeless: what now in gdb? [21:32] the slow script is there [21:33] ^C / ^{break} [21:33] * timeless is busy downloading symbols for firefox (which is hosting irc cloud) [21:34] btw: you don't generally want to debug your irc client. it tends to die :) [21:36] timeless: pretty much the same thing [21:36] Program received signal SIGINT, Interrupt. [21:36] 0x00007fa090e8c753 in __GI___poll (fds=, nfds=, [21:36] timeout=) at ../sysdeps/unix/sysv/linux/poll.c:87 [21:36] 87 ../sysdeps/unix/sysv/linux/poll.c: No such file or directory. [21:36] in ../sysdeps/unix/sysv/linux/poll.c [21:36] yeah, the top frame will always be useless for this [21:36] the hope is that in the middle you'll see more useful stuff, hold on [21:39] oh right. it's a bad idea to copy text from firefox and paste it into your debugger [21:39] (if your debugger is debugging firefox..) [21:40] what problem are you trying to debug? [21:41] chrisccoulson: http://forums.mozillazine.org/viewtopic.php?p=11568785#p11568785 [21:41] if the top of the stack is a poll(), then the rest of it is usually not all that useful :) [21:42] I'm seeing slow resource:///components/nsPrompter.js:68 and chrome://global/content/bindings/browser.xml:273 when CPU is raced [21:42] the other thing is that it's quite likely you're looking at the wrong thread [21:42] * timeless was debugging the debugger instead of firefox (oops) [21:42] that's the symptom I get at least [21:43] ok, now i have a debugger (64bit) debugging firefox (32bit) [21:44] * timeless gets slow script dialog (yay) [21:45] ok, so,.... the brute force command is: [21:45] thread apply all bt [21:45] the windbg equiv fwiw is: ~* kp [21:45] (thread) (all) (backtrace) (somewhat verbose) [21:46] in general, for a slow script, you're probably going to want the ui thread, which is generally thread 0 [21:46] but from memory gdb's concept of threads isn't virtualized, so you actually have to think about them by their underlying representation instead of "the zeroth thread" [21:49] anyway, we can try to work wih certain constraints [21:51] * timeless pokes alex_mayorga [21:53] * alex_mayorga pokes the mostly death laptop with a stick [21:54] timeless: I've got the nsPrompter.js right here [21:54] what do I do in dbg? [21:54] (gdb) thread apply all bt [21:54] hrm [21:54] (gdb) thread apply all bt 1000 [21:55] let's not let it be annoying [21:55] if possible, try to make gdb think you have many lines on your screen so that you don't have to hit to continue (don't ask me how to do that) [22:18] * timeless pokes alex_mayorga [22:44] timeless: trying but this thing is very unresponsive [22:44] how much ram do you have? :) [22:45] 4GB [22:50] how much is free? :) [22:52] total used free shared buffers cached [22:52] Mem: 4016412 3902520 113892 0 864 87176 [22:52] about 1 GB I think [22:53] more like 100MB, unless my eyes are deceiving me ;) [22:55] chrisccoulson: I trust any third party, rather than myself :) [22:55] http://piratepad.net/c3berohKLm should be it [22:55] I got to run now [22:56] would let it sit here see if has exploded when I come back ;-) [23:05] alex_mayorga: ok, so, the section i left in purple is the interesting bit [23:05] the next bit is to see if we can use the xpconnect stack dumping stuff, if not, we'll have to do it the hard way [23:14] alex_mayorga: i left notes at the bottom [23:14] they should work, i hope [23:16] (right, selecting "continue" for slow script for while(1) isn't a good idea...) [23:20] lol @ https://twitter.com/#!/deadsquid/status/148902812783349760