[00:05] http://en.wikipedia.org/wiki/DWDM for an intro, when you're done with your study ;) [00:07] fta: yeah thats it [00:08] http://www.snotr.com/video/76 [00:08] optical cross connects ... yay [00:08] are there any "pure" optical solutions? [00:08] not in IP [00:09] routing is still difficult [00:09] yeah ... for static routings i guess [00:10] well, but big lines can have quite static routings ... at least for a few hops :) [00:11] transmission is possible in full optics (DWDM), but IP inside (L3) still needs optic-electric-optic conversions [00:12] wave lengths commutation vs packets routing [00:12] yeah [00:12] thats what i mean :) [00:12] * asac feels unqualified to do any serious discussion on this topic [00:15] and i feel tired after so much sports yesterday and today [00:15] fta: bike? [00:16] yes [00:16] with +27°C [00:17] fta: i envy you [00:17] weather could be better here for two weeks already [00:18] btw, i'll be traveling for the 1st half of july [00:19] holiday? [00:19] no [00:19] well, half [00:26] fta: ok ... so starting next week? [00:26] yep [00:28] fta: do you look into install.rdf to get the addons version in check-extensions script? [00:29] yes, but there are exceptions :( [00:29] fta: for non-amo extensions? [00:30] useragentswitcher => config.properties [00:30] fta: ok. so to properly check the version we need to spin the package :/ [00:30] fta: but well [00:30] webdeveloper => config.properties + config_common.properties + config_seamonkey.properties + ... [00:30] we can make that pluggable [00:31] e.g. custom_extensions_version_command :) [00:31] fta: is webdeveloper packaged by unpacking amo .xpi? [00:31] donno [00:32] thats painful [00:32] indeed [00:33] fta: but webdeveloper is not synched from AMO apperantly [00:33] so falls in the "not-yet-covered" upstream source category [00:38] it's not really difficult [00:48] lol http://home.kairo.at/blog/2008-06/soccer_impact_on_internet_usage [00:54] yeah cool ;) [01:37] fta, read about wdm. Interesting :). [01:37] But no, we don't mention optical telecom at all... [01:37] Thanks for the link :) [05:53] howdy [05:54] I'm noticing a pretty big issue in Firefox-3.0 3.0+nobinonly-0ubuntu2 [05:54] Simply, jemalloc seems broken === asac_ is now known as asac === fta_ is now known as fta [12:41] good ... kaze built [13:48] is this a known bug? https://bugs.launchpad.net/ubuntu/+source/firefox/+bug/242322 [13:48] Launchpad bug 242322 in firefox "invalid rendering of pages - wrong size of some links" [Undecided,New] [14:00] LimCore: please search bugzilla.mozilla.org [14:01] LimCore: ffox 3 still has a some rendering issues, so there might be a bug open there [14:01] LimCore: go to advanced search page and select the layout component [14:01] or gfx [14:01] to get a search on a valid bug base :) [14:07] ok I will look in time then [14:10] cool. thanks [16:33] Hey, asac... About extension merges. Since we're not doing it with debian, do we wanna keep lower version (for example debian has 0.14-1, and we use 0.14-0ubuntu1), or not? [16:52] ha ... i am not allowed to contact gmail.com anymore because i am in germany [16:53] really really stupid [16:53] Jazzva: given that we override their changes we should use higher version imo [16:54] ok... just to correct then. I thought that maybe we could use lower version, and then indicate in merges that we mostly diverged from them for now, but that wouldn't be easier :) [16:55] Then I should also include debian's changelog entry [16:58] yes, use higher version so it disappears from the MoM list [16:58] ok [17:01] yay. two more merges that I'm doing for sure :). Though, both are extensions :) [17:53] Hi asac :-) [17:53] (and hello everybody) [18:08] hi Yannig [18:08] I found a problem with Firefox po :-) [18:08] It cannot be re-imported into Launchpad :p [18:11] On 2008-06-15 14:49+0000 (7 days 55 minutes ago), you uploaded a file [18:11] with Occitan (post 1500) (oc) translations for firefox in Ubuntu Hardy [18:11] package "firefox-3.0" to Launchpad. [18:11] We were unable to import the file because of errors in its format: [18:11] PO file: duplicate msgid ending on line 78 === gnomefreak changed the topic of #ubuntu-mozillateam to: Welcome to the Ubuntu Mozilla Team: | Mailing List: ubuntu-mozillateam@lists.ubuntu.com | Please help Mozilla QA tracker: http://tinyurl.com/4mooo2 | Firefox 3 released to hardy-updates! | Next meeting will be TBA. You can find the agenda for the meeting at: http://tinyurl.com/2ekzoq [18:21] is anyone here? [18:21] Jazzva: thanks for minutes they look good [18:21] jdhore: whats up? [18:22] gnomefreak, lemme copy what i said way earlier...1 sec [18:22] howdy [18:22] I'm noticing a pretty big issue in Firefox-3.0 3.0+nobinonly-0ubuntu2 [18:22] Simply, jemalloc seems broken [18:22] gnomefreak, no problem :) [18:22] asac: do you have anything on UWN by chance? [18:22] jdhore: jemalloc? [18:22] care to elaborate? [18:23] gnomefreak, the new firefox3 memory allocator that makes Firefox suck a lot less [18:23] jdhore: let me try this. what is the problem [18:24] Basically, after only running Firefox for a short time, it's getting up to ~500MB of RAM in use whereas i NEVER had the nightly binaries from Mozilla hit over ~250MB RAM in use [18:24] hi, [18:24] jdhore: does this also happen in safe mode? [18:24] gnomefreak, yes [18:24] I just booted an outdated hardy [18:25] and FF3 is pretty unstable [18:25] Let's see if reboot can help [18:25] jdhore: does this include flash and java pages as well as text pages? [18:25] gnomefreak, some flash, but not much and no java (I don't even have JRE installed) [18:26] and only 3 extensions [18:26] (all of which are FF3 compatible, i did no overrides) [18:27] jdhore: camplatible or not doesnt mean they work but sinc eyou claim it happens in safe mode it doesnt matter on extensions at all. [18:27] true [18:27] I remembered that just after i mentioned that [18:27] jdhore: please file a bug using Help > Report a Problem and follow the instructions on filing a bug than give me bug link please [18:28] * gnomefreak reading about any upstream issues with 1.9 atm [18:28] gnomefreak, It's not there...or it's greyed out [18:28] jdhore: its greyed out? [18:29] asac: !!! where are you? [18:29] jdhore: its not greyed out here [18:29] kinda hard to tell (grey text on black background = not easy to read) [18:29] jdhore: is this Ubuntu's version of firefox? [18:29] Yes [18:29] but i should mention that i'm on Debian Testing, not Ubuntu [18:30] jdhore: that would do it [18:30] lemme guess [18:30] jdhore: we dont recommend nor do we advocate using binaries from Ubuntu on Debian nor vice versa [18:31] jdhore: you will encounter problems some small some bigger than you will ever see again [18:31] you guys put libjemalloc (or whatever it uses) in the system so it depended on a shared library which Debian doesn't have? [18:31] jdhore: suggestion compile it from source use ubuntus source and compile for debian [18:31] gnomefreak, i didn't use binaries, i built it from source in the Debian-approved way [18:31] (used the ubuntu sources and cowbuilder) [18:32] jdhore: that is very possible is it in the control file for Ubuntu firefox [18:32] i didn't notice it in the control file [18:32] jdhore: see if its missing from iceweasel source [18:32] I noticed a few patches in debian/patches relating to jemalloc, but i didn't want to remove them [18:32] jdhore: what version of xulrunner are you using? [18:33] 1.9+nobinonly-0ubuntu2 [18:33] jdhore: they are needed for our build dont remove them [18:33] jdhore: debians? or built from our source? [18:33] yours [18:33] (built from source) [18:34] jdhore: see if iceweasel has same issue but if i had to guess there is gona be alot more depends that you will need [18:34] i dont have adebian chroot handy atm [18:35] iceweasel doesn't have the same issue, they explicitly compile jemalloc out [18:35] - Use MOZILLA_NO_JEMALLOC instead of LD_PRELOAD to disable jemalloc. [18:35] jdhore: than that sounds like a depend issue [18:35] jdhore: your missing a few deps [18:36] gnomefreak, can you give me a full list? cuz...I assume the debian/control file is missing a few [18:36] firefox-3.0 Depends: fontconfig Depends: psmisc Depends: debianutils Depends: xulrunner-1.9 Depends: libatk1.0-0 Depends: libc6 Depends: libcairo2 Depends: libfontconfig1 Depends: libfreetype6 Depends: libgcc1 Depends: libglib2.0-0 Depends: libgtk2.0-0 Depends: libnspr4-0d Depends: libpango1.0-0 Depends: libstdc++6 Depends: zlib1g [18:36] why does ubuntu's firefox-3 disable jemalloc? [18:36] jdhore: they do not include xulrunner deps [18:36] and most likely that is where the issue is [18:37] laptopnenolod: we dont [18:37] laptopnenolod: we built with it and its built in xulrunner not ff [18:37] laptopnenolod, i'm talking to them right now about that [18:37] hmm [18:37] laptopnenolod, it seems we're missing some dependencies which was the big list he pasted when you first joined [18:37] laptopnenolod: debian builds without it [18:38] jdhore, no, we have all of the deps i think [18:38] xulrunner-1.9 Depends: libatk1.0-0 Depends: libc6 Depends: libcairo2 Depends: libdbus-1-3 Depends: libdbus-glib-1-2 Depends: libfontconfig1 Depends: libfreetype6 Depends: libgcc1 Depends: libglib2.0-0 Depends: libgtk2.0-0 Depends: libhunspell-1.2-0 Depends: libidl0 Depends: libnspr4-0d Depends: libnss3-1d Depends: libpango1.0-0 Depends: libsqlite3-0 Depends: libstartup-notification0 Depends: libstdc++6 [18:38] Depnds: lib0x11-6 Depends: libxft2 Depends: libxrender1 Depends: libxt6 Depends: python2.5 Depends: zlib1g Conflicts: Breaks: devhelp Breaks: epiphany-gecko Breaks: midbrowser Breaks: yelp [18:38] jdhore: also neeed the build deps and they are in the debian/control for both packages to see what you are missing [18:39] gnomefreak, right, we built agains the debian/control files, so if those deps are in it, we built with them [18:39] fta: i dont have a ff3 source handy ar eyou around? [18:39] jdhore: did you leave any out? [18:40] gnomefreak, we imported the mozilla specific deps from ubuntu. so we have them [18:40] ones that maybe debiuan doesnt use or doesnt have right version of? [18:40] gnomefreak, the way cowbuilder (what i used to build) works is that it won't build the package unless all deps in debian/control are met [18:40] gnomefreak, our package set derivates majorly from debian's [18:40] (Debian Testing's) [18:41] jdhore: what about flags? [18:41] yes, that too [18:41] will it skip build flags? [18:41] if cant be met? [18:41] no [18:42] are you running it in cowbuilder? [18:42] * gnomefreak wonders if its like pbuilder only sets everything up in basic temp chroot [18:43] cowbuilder is almost exactly like pbuilder [18:43] (except you can chroot into the build environment) [18:44] if you built it with everything i dont se ehow its a Ubuntu issue since ours works fine built and ran on Ubuntu, im trying to cross out everything but running environment since i cant do that atm [18:44] jdhore: can you run it in cowbuilder? [18:45] no [18:45] I have no X in cowbuilder [18:45] if you can chroot into it you should be able to run it. mind you do it witha new profile [18:45] nor do i really want to [18:46] yes i know the feeling but since you need to test it in the build enviornment to see what the issue is i dont see another choice that will work. maybe build a Ubuntu chroot and test it in there. than remove it once test is done but again new profile [18:47] The build environment is identical to the running environment [18:47] jdhore: as of right now our build works without xulrunner issues from what i can tell ( on a basic system without extensions and such [18:47] jdhore: i dont think so [18:47] jdhore: build-environment needs to have Ubuntu packages not just one or 2. [18:47] well...fewer packages in the build environment, but other than that, yes, 100% identical packages and package versions [18:48] oh [18:48] crap [18:48] if you dont have the packages needed to build on Ubuntu it wont build on Ubuntu [18:48] we have the packages [18:49] if i tried to build a hardy package on gutsy but used gutsy packages it will still fail to run on hardy since it wasnt built on hardy [18:49] all of them with the same names, and probably similar versions, they just have none of the Ubuntu packages [18:49] *patches [18:50] jdhore: example if you built on ubuntu's nss than most likely it wont run on debians. you need to try taking debian packages and renaming them and change build-options to see if it will build and run [18:50] IT RUNS [18:50] IT BUILDS [18:50] This is not the issue [18:51] The issue is that jemalloc doesn't bloody work [18:51] jdhore: BUT IT FAILS TO WORK PROPERLY [18:51] hint do you think Mike stopped that just in package? [18:51] i would assume theres a lib or 4 that is needed for that [18:51] I'd say more like you guys bloody screwed something up [18:51] does debian have any of them at all [18:52] jdhore: but it wortks here [18:52] jdhore: so we couldnt have we left upstream build-options as is you change them [18:52] Why the hell diedn't you guys just leave jemalloc working stock from Mozilla and be bloody happy [18:52] ? [18:52] we didint disable it [18:52] it works fine [18:52] you made it shared it looks like [18:53] debian build disables it [18:53] we didnt change upstream anything for that. [18:53] jdhore: take a look at our patches and tell me what one changes it in any way [18:53] Drop LDFLAGS workaround now that jemalloc is no longer a static lib. [18:53] We still ship jemalloc as a shared lib [18:53] xulrunner-1.9 (1.9~rc1+nobinonly-0ubuntu1) [18:54] jdhore: did you happen to get final [18:54] and yes i can see him doing that [18:54] jdhore: did upstream apply the patch ? [18:54] yes, i did get final, and i looked though the changelog from there till final and i saw no patches or entries that reverted that patch [18:55] How the hell do i know? [18:55] * gnomefreak really getting tired of this static vs shared upstream topics [18:55] It is static in Firefox upstream, i know this for a fact [18:55] jdhore: grab final build from us and look to see if patch is still there or to use old patcha nd see if the upstream file has been changed to match [18:56] it is [18:56] jdhore: thats not what i meant [18:56] jdhore: this isnt the first time this bug has been brought up [18:56] static vs shared [18:56] IT HAS NOT BEEN REVERTED IN THE CHANGELOG. THERE IS NOTHING NEWER THAN THAT EVEN REFERRING TO JEMALLOC. [18:56] its an upstream bug [18:56] ah [18:56] jdhore: changelog doesnt tell you everything BTW [18:57] its kind of like a set of notess [18:57] notes [18:57] true, but there are still patches in the debian/patches directory in final that mention you guys doing shit with jemalloc [18:58] jdhore: asac and fta are more caught up on the static/shared issues, since noone likes that we build shared it has been in upstream bugs but numbers escape me its been a while and i was concerned only in aspect of Tbird-3 [18:59] right [18:59] so now it becomes a huge pain in the ass cuz now we have to build with Mozilla source...yay... [19:00] jdhore: its because we build shared and they want us to build static as far as i know. i have been busy with "real life" to keep track of the builds off hand. this was a moot subject when i left and still shouldnt effact our build since noone in Ubuntu sees it [19:00] jdhore: you should do that anyway [19:00] jdhore: that way you dont add ubuntu/debian/RH bugs in your package [19:00] right, i guarantee that's where the problem is since mozilla builds static and their nightly binaries worked fine [19:00] we dont build from debian ;) [19:01] true, but it's a huge pain for packages as...large/expansive as Firefox and XULRunner [19:01] jdhore: try building them every week for months [19:01] it becomes natural [19:01] heh...true :P [19:02] * gnomefreak toatally forgot about shared/static :( [19:04] but i still cant say its that that is causing the main issue since our build doesnt show any leakage signs that i have heard about so you may stioll have the issue after changing build flags. [19:05] hrm [19:05] but our build is built to run on ubuntu nothing else :) [19:05] asac: looks like ff3 segfaults on ppc [19:05] asac: if built with -O2 [19:06] or -Os [19:08] armin76: hes not here right now if you see him please smack him for me ;) [19:09] * armin76 smacks gnomefreak for not fixing it [19:09] :) [19:09] -02 should be fine, so it leads me to think its the lack of support for PPC [19:09] ever since it moved to ports i dont think we have one to test on anymore [19:10] IIRC asac was only one with a PPC for the longest time [19:15] jdhore: so whats your issue about jemalloc? [19:16] asac, according to a friend, building it shared is the stupidest thing in the world [19:16] if you make a library which overrides a function in glibc(), the library will be ignored [19:16] the only way to override functions in glibc is by static linking [19:17] Basically, if you try to use shared linking, it will ignore jemalloc, it will only not if you use static linking [19:17] something bothers me why did we add that to changelog if mozilla didnt comply with what was said :( [19:18] we dont build with --disable-* nor --enable-* with static or shared so it would build as Mozilla had wanted from waht i can tell [19:18] jdhore: doesnt it depend on the load orderß [19:19] orderß what is that last char? [19:19] ? [19:19] gnomefreak, there were manual defines (or something) in one of the debian/patches [19:19] jdhore: non of these patches in debian/patches are applied. look in debian/patches/series [19:19] asac, no, not if you're overriding the default malloc [19:19] jdhore: that would have been workaround no? [19:20] asac, nope, only 3 weren't applied [19:20] jdhore: i am pretty sure you can use LD_PRELOAD ... so why wouldnt it depend on the load order? [19:20] and i can guarantee at least one of them has nothing to do with jemalloc [19:20] jdhore: not sure what you are talking about [19:21] asac, because that's for normal libraries, not libraries overriding glibc's native calls [19:21] jdhore: name the patches of concern and i can tell you more [19:21] 1 sec [19:22] drop_bz418016 in xulrunner looks like the patch of concern [19:22] jdhore: that one is not applied [19:22] LMAO i show more in series than i do in patches [19:24] jdhore: actually i dont understand how malloc should be technical different to other library functions ... that said, I think its still used. [19:24] if its not i dont care much either :) [19:24] as then it doesnt hurt and we are just tracking upstream behaviour here [19:28] jdhore: ok works here :) [19:28] hmm? What does? [19:28] jdhore: http://paste.ubuntu.com/22397/ [19:28] overloading malloc :) [19:28] in a shared lib [19:29] but the paste above in a file "test.c" [19:29] gcc -fPIC -shared -o libtest.so test.c [19:29] then make a main.c with: [19:29] http://paste.ubuntu.com/22399/ [19:30] and build gcc -ltest -L. -Wl,-rpath=$PWD/ main.c [19:30] then run [19:30] ./a.out [19:30] :) [19:32] can we tryt o keep patches under 10,000 lines :( [19:33] gnomefreak: the hunspell patch is just so big because it moves files :) [19:33] this is the old gconf patch [19:33] its huge [19:33] gnomefreak: right [19:33] gnomefreak: its in mobile build [19:33] this looks liek the patch that we used as workaround [19:33] oh [19:33] i have to move the gconf part to some shared component and punch that into a separate package [19:34] ok [19:35] jdhore: anyway ... this all isnt really relevant as jemalloc isnt used anymore in final afaict :) [19:35] uhh [19:35] Yes it is [19:36] jdhore: nope [19:36] $ strace -f -eopen 2>&1 firefox | grep jemalloc [19:36] ... see: deadly silence [19:36] not loaded [19:36] right [19:36] and there's where the problem is [19:36] jdhore: thats upstream [19:36] jdhore: export LD_PRELOAD=/usr/lib/xulrunner-1.9/libjemalloc.so [19:36] because according to the Mozilla devs, it is still used in FF3 final [19:36] then run [19:36] strace -f -eopen 2>&1 firefox | grep jemalloc [19:36] open("/usr/lib/xulrunner-1.9/libjemalloc.so", O_RDONLY) = 3 [19:36] open("/usr/lib/xulrunner-1.9/libjemalloc.so", O_RDONLY) = 3 [19:37] jdhore: no its not [19:37] beta4 was last post i can see by devels on jemalloc [19:37] i saw that patch landing before release that disabled it with the hint "dont use by default and let distros whatever they want" [19:37] chances that i am wrong exist, but I'd say its < 10% [19:37] asac, I followed the svn commits up to the day before final [19:38] jdhore: mozilla isnt in svn :) [19:38] and i never saw a ptch that disabled jemalloc by default [19:38] asac, i know, before they switched over to hg [19:38] jdhore: before that it was in cvs ;) [19:39] hmm...thought it was svn...Well...no matter, i followed bonzai which shows all cvs commits [19:42] asac, you lose, sir [19:42] 18:36:06 < jdhore> Is jemalloc still used on Linux on FF3 final? [19:42] 18:42:16 < dmose> jdhore: yes [19:43] Memory cycles are broken and collected by an automated cycle collector, a new memory allocator reduces fragmentation, hundreds of leaks have been fixed, and caching strategies have been tuned. [19:43] but fails to say what [19:45] i got it from http://www.mozilla.com/en-US/firefox/3.0/releasenotes/ [19:45] gnomefreak, this was coming from dmose who is one of the lead FF3 Linux people [19:45] jdhore: i wasnt looking at this channel i was reading [19:45] gnomefreak, i know, it's not very technical in the release notes, but i'm sure i would've been corrected when i said jemalloc in firefox3's dev channel [19:46] than why did mike decide default was not best? [19:46] who's mike? [19:46] debians Maintainer [19:47] jdhore: [19:47] $ mv /home/asac/.mozilla /home/asac/.mozilla.bak [19:47] asac@hector:/tmp/qw/firefox$ pwd [19:47] /tmp/qw/firefox [19:47] asac@hector:/tmp/qw/firefox$ ls libjemalloc.so [19:47] libjemalloc.so [19:47] asac@hector:/tmp/qw/firefox$ strace -f -eopen firefox 2>&1 | grep jemalloc [19:47] nothing [19:47] sure there are more than just him but as i recall he is head Mozilla maintianer for debian [19:47] so they definitly ship shared and dont use it [19:47] yes [19:47] jdhore: and they dont use static either :) [19:47] because you bloody disabled it or something [19:47] jdhore: thats upstream build :) [19:47] jdhore: look ad the pwd in my paste [19:47] its just upstream firefox ... nothing else [19:48] er [19:49] jdhore: ok i stand corrected. in any case. its optional and we dont change any configure flag for that [19:50] again, not according to dmose who is a ff3 dev [19:50] i can surely look into this. but imo it doesnt really matter performance wise [19:50] i know him :) [19:50] Actually, it quite does [19:50] It dropped my RAM usage of firefox in half the day they enabled it [19:50] and no, i'm not being sarcastic [19:51] asac: did you grab 3.1? [19:51] jdhore: try what i said [19:51] ? [19:51] jdhore: export LD_PRELOAD=/usr/lib/xulrunner-1.9/libjemalloc.so [19:51] then firefox [19:52] (with our build) [19:52] ill use it a bit for now [19:52] jdhore: in any case: its shared, not static [19:52] Jazzva: did you find out about LP blogging? [19:52] and it works ... if it works for upstream ;) [19:53] gnomefreak, it doesn't exist. I filed a wishlist bugreport [19:53] Jazzva: ok thanks i was gonna ask if you didnt get around to it [19:53] * gnomefreak scared to touch email [19:53] gnomefreak, it's ok :). Do you need a bug report num? [19:54] Jazzva: yeah please i would like to follow it [19:55] damn i forgot we set -t :( [19:55] gnomefreak, bug 242211 [19:55] Launchpad bug 242211 in launchpad "Wishlist: Enable announcement feature for team pages" [Undecided,Confirmed] https://launchpad.net/bugs/242211 [19:55] Jazzva: thank you again ;) [19:55] np :) [19:56] 629 in one box [19:58] jemalloc fails to compile with glibc-2.3 fyi [19:59] ha i found a bug on it :) [20:00] armin76: just got that email [20:00] asac: jdhore http://pastebin.mozilla.org/466509 [20:01] gnomefreak: whats that? [20:02] thats someone building it failing on glib atleast hope it isnt [20:02] jemalloc doesn't support glibc-2.3 [20:03] ah [20:03] gnomefreak: try http://people.ubuntu.com/~asac/miscpatches/jemalloc_in_xul.patch [20:03] in xulrunner-1.9 package [20:03] ok i have to do email first [20:03] just add it to debian/patches/ and debian/patches/series [20:04]