[00:01] fta: do we want to add the info that we added a new branding to the upstream version? [00:01] e.g. +nobinonly+awesome ;) [00:02] imho, the version is already long enough, it's getting ridiculous [00:04] yeah [00:05] ok .. i am pushing the awesome thing to ppa ;) [00:05] how does it look like ? [00:06] the idea is that you get the generic branding when you dont have firefox-3.0-branding installed [00:07] which is now recommended by firefox [00:07] if you install awesome-browser it will force the branding to got away [00:07] e.g. conflicts firefox-3.0-branding [00:08] fta: at best you try yourself. i definitly need a review on this chunk :) [00:08] yes, i figured that out reading your commits, but do you have a screenshot of awesome-browser ? [00:09] the package name is awesome-browser ... the UI just says "Web Browser" [00:09] you can look at the icons in the branding branch ;) [00:09] i can try to get a screenshot ;) [00:09] but let that upload finish first ;) [00:10] https://code.edge.launchpad.net/~mozillateam/firefox/awesome-browser-branding [00:12] fta: http://people.ubuntu.com/~asac/screenshots/awesome1.png [00:13] woow, where does this icon come from ? [00:13] from our artist ... he made that during last sprint when this idea became more real [00:14] but i want it to be improved ... but well [00:15] but its not complete ;) [00:15] the credits page has still the rotten old blue one [00:16] thought it was complete [00:16] bad bad ;) [00:16] but lots of time to get that done i think [00:19] what i wonder is why this stupid sentence about the firefox mark doesnt come from the branding chrome [00:19] makes no sense to put it in a non-branding place imo [00:19] file a bug :) [00:37] hmm ... debian/patches/lp185622_system_path_default_browser.patch is too stupid to easily point to a different path [00:38] bug 185622 [00:40] no chance [00:40] thing is away [00:41] using the same path for both sounds tempting, but would cause pain elsewhere i guess [00:41] e.g. gnome preferred application dialog [00:41] which hardcodes application names by path afaik [01:26] fta: do you know if its possible to tweak $0 ? [01:26] (in sh) [01:31] i don't think it is [01:32] how sad [01:33] why do you want to fake the name ? [01:35] because its cumbersome to teach firefox always that we are /usr/bin/somthing [01:35] and not /usr/lib/... when it tries to update default browser [01:35] same for "restarT" [01:35] where it tries to exec the binary in pkglibdir [01:35] which is far from perfect for the upgrade case [01:36] and since we have a script $0 is not even right [01:36] before resolving links [01:36] so i hoped we could fake $0 for the binary we invoke from our /usr/bin script [01:37] if $0 would be correct, then i could implement a real fix [01:37] and not this stupid hard code shuffeling we currently do [02:44] ok thats good enough for today [02:44] off === BobChao_ is now known as BobChao [10:43] Kamping_Kaiser: https://code.edge.launchpad.net/~asac/firefox/firefox-3.0.head.awesome [11:09] hi [11:09] mozilla bug 452754 [11:09] Mozilla bug 452754 in jemalloc "Deadlock in jemalloc when using fakeroot" [Normal,New] http://bugzilla.mozilla.org/show_bug.cgi?id=452754 [11:32] plop [11:43] seems most core devs in cairo are either redhat or mozilla [11:43] http://cairographics.org/summit/2008/schedule/ [11:44] http://cairographics.org/summit/2008/notes/ [11:51] "Most of the new API is uncontroversial, but Carl dislikes the 3 lcd_filter API entries---it seems we should be able to address the use cases that have come up without any new API here." [11:55] fta: true [11:56] typical "Let's choose Fedora rawhide as our initial reference platform to make the whole test suite work and then extend from there, (Fedora, Ubuntu, Win32, OS X, etc.)" [11:57] i'm surprised people like macslow are not involved in cairo [12:09] fta: well. i think his experience is on a different level [12:10] its a difficult question [12:11] they are talking opengl in cairo now, seems he fits in, but well, i'm just reading his blog [12:13] he is more on a user side ... working on new ways to present the desktop [12:17] fta: can you review the awesome merge? [12:17] i have to check with others on the name. dont like the idea that we need another name transition :( [12:19] its https://code.edge.launchpad.net/~asac/firefox/firefox-3.0.head.awesome [12:21] sure, i will [12:25] awesome bumb [12:41] + This package ships the firefox branding bits. If you remove this package you [12:41] + your user experience will become that of the awesome-browser. [12:41] asac, ^^ topo [12:41] typo [12:41] haha [12:42] a typo reporting a typo :P [12:46] asac, so everyone will have /usr/bin/awesome-browser* ?? why ? couldn't it be only in the awesome-browser-3.0 package ? [12:50] wtf is awesome browser? [12:50] asac, why all those common-*-arch rules instead of a proper Makefile.in patch ? [12:51] ah, unbranded build, meh [12:51] i'd say, rebranded [12:54] fta: the right fix would be intrusive [12:54] fta: you would require to teach firefox to biuld multiple brandings [12:55] so its a bit of a hack that we respin the branding [12:55] at least i dont see a simple way to fix the build system [12:55] but i dont want to rule out that thats not an issue [13:12] too bad you can't install both [13:15] fta: cant install both? [13:15] yeah ... its the same browser [13:16] i am not sure how to properly do it that you can install both without shipping two distinct binary trees [13:16] firefox and xul would probably need to learn about custom "icons" directories [13:16] and custom branding packages [13:16] that have a different namespace [13:33] asac, asac_, why don't you move all the awesome files to the awesome-browser package ? [13:34] fta: because then firefox-3.0 on its own wouldnt work anymore [13:35] so we would need make firefox-3.0 depend on firefox-3.0-branding | awesome-browser-3.0 [13:35] might be an option ... but i fear that we get upgrade issues [13:35] dependencies become more and more complex [13:36] i always feel bad about making bi-directional depends [13:36] like [13:36] firefox-3.0 depends on -branding ... and -branding depends on firefox-3.0 [13:37] the other idea is to implement this in a consistent way: [13:37] Top Level: firefox, awesome-browser [13:38] depends on firefox-3.0, firefox-3.0-branding [13:38] or awesome-browser-3.0, awesome-browser-3.0-branding [13:38] but that makes things really complex [13:38] more package bloat imo === asac_ is now known as asac [13:40] why i did it in this minimal fashion is too keep more options. we can always add more packages [13:40] but removing them always causes painful transitions [13:43] ideally, the branding should be like a theme :) [13:44] what does mozilla think about that whole rebranding idea ? [13:47] fta: well. i think they are happy when we still use their branding [13:47] one of the counter arguments you always get from them when telling them that their trademark thing isnt good is that we (linux distros) are the one that fail to provide an easy way for downstreams [13:47] to use a free branding [13:52] fta: but in the end we wont really know until we release i think ;) [14:16] fta: some folks say that awesome might be a bit too casual ... they suggest zuul-... what do you think? [14:32] fta: colin said that "webbrowser" as package name would be ok most likely [14:33] i try to get more feedback on that [14:33] i think that would be better if other browser-lovers could accept that ;) [14:33] where is this discussion taking place ? [14:34] i asked him privately [14:34] ill ask the other browser lovers in -desktop [14:34] and note sure [14:35] -desktop has been dead for a week now [14:36] Hi! [14:36] fta: right [14:36] seb is gone ;) [14:37] asac, what's up ? [14:39] XioNoX: not much ;) ... time just disappeared. how many days have you not been here? [14:39] 1? [14:39] ;) [14:40] since monday [14:42] so 4 days [14:47] XioNoX: wow. time just slips ;) [14:47] When is the feature freeze ? [14:47] xi we are past FF [14:47] ;) [14:48] but doesnt matter for these features [14:48] i have a FFe i guess [14:48] ? [14:48] to upload on weekend [14:48] XioNoX: there are still a few issues with the altplugin dialog as you can guess ;) [14:48] ill try to flash out as many things as possible for the upload [14:49] flashout ? [14:49] fix [14:49] ok [14:49] I can't realy do things this week [14:49] the restart thing is also broken on major upgrades, but thats a problem in the guts of the firefox restart mechanism [14:49] but newt week it will be ok [14:49] which ill have to patch [14:49] XioNoX: sure [14:49] XioNoX: ill get those bits all together and upload [14:50] next week we will certainly see bu reports so we can work together on fixing them maybe [14:50] cool [14:50] sure :) [14:50] i hope i didnt promise too much [14:50] (by uploading this weekend) [14:52] ok [15:02] oh ... i think ill get thrown out of debian :( [15:02] iceowl was removed from lenny due to undermaintenance [15:02] anyone wants to fix iceowl? [15:02] i can add you as uploader so you can directly upload to debian in future ;) [15:09] fta: so do you understand why jemlaloc causes that dealock? [15:11] i guess because pthread calls malloc [15:12] http://mxr.mozilla.org/mozilla-central/source/memory/jemalloc/jemalloc.c#5341 [15:14] http://mxr.mozilla.org/mozilla-central/source/memory/jemalloc/jemalloc.c#5365 [15:18] asac, see what i mean? [15:19] fta: why does fakeroot trigger pthread thing? [15:20] no [15:20] i see that unlike apparently triggeres a reentrant malloc [15:21] s/unlike/unlink/ [15:21] do you have fakeroot dbgsym ? [15:21] jemalloc triggers its own malloc, which calls unlink, provided by fakeroot, which itself need calloc [15:22] so jemalloc has 2 locks in one allocation [15:22] no fakeroot dbgsym :( [15:23] yes. so its not the pthread comment you pointed to [15:24] i pointed you to the deadlock [15:24] fta: try to disable MALLOC_PAGEFILE [15:24] that appears to be a new feature [15:24] its not in 1.9 branc [15:24] (e.g. no unlink there) [15:24] fta: the first line was the comment about recursion and phread ;) [15:25] 16:11 < fta> i guess because pthread calls malloc [15:25] 16:12 < fta> http://mxr.mozilla.org/mozilla-central/source/memory/jemalloc/jemalloc.c#5341 [15:25] thats what you said ;) [15:25] anyway. doesnt matter [15:25] isnt the MALLOC_PAGEFILE code supposed to be for windows? [15:25] sorry, i'm tired, and distracted [15:26] no problem ;) [15:28] was the update today supposed to change the icon for firefox? [15:28] fta: http://mxr.mozilla.org/mozilla-central/source/memory/jemalloc/jemalloc.c#168 [15:28] try to disable that there for UNIX too [15:36] well, if anyone cares, the update changed firefox's icon to some planet (awesome browser?) and broke compatibility with imwheel [15:37] asac, go back to work ;) [15:45] asac, if you need something, just send me an email [15:46] natrooki probably didnt have firefox meta package installed [15:52] i don't have it either [15:53] isnt firefox-3.0-branding the first recommends of firefox-3.0 ? [15:53] i mean, the firefox meta package [15:53] yeah [15:56] lets check whats going on ;) [15:57] ok .. lets see defualt install (removing firefox-3.0 first) [15:59] ok [15:59] 1. just install firefox: http://paste.ubuntu.com/41578/ [15:59] (OK) [15:59] 2. just install firefox-3.0: http://paste.ubuntu.com/41579/ [15:59] (OK) [16:00] ok installing from intrepid to test upgrade [16:00] (for now just firefox-3.0 [16:02] hmm ... will take a bit [16:02] have to first upgrade everything [16:07] damn, i have to hack inside libxml2 [16:07] why? [16:09] namespace problems while merging xml trees [16:09] xmlReconciliateNs() is not good enough [16:10] does it wrong namespaces or jsut in an uncommon presentation? [16:10] it makes me produce un-optimized svg content [16:10] becase of namespaces? [16:10] yes [16:10] well ... i can assume that lookups with namespaces might take a bit longer [16:11] but is it worth hacking and maintaining it in a xml parser lib? [16:11] it creates new ns at each merge, instead of reusing the global ones [16:11] yeah. but you could probably santize that in a second step outside hte parser [16:11] (if you think its worth it at all) [16:11] i just extend the api on my side, i don't want to go upstream [16:12] ok. but thats not really hacking libxml in the narrow-sense ;) [16:12] the thing is, they have private data in their structures [16:12] at least not what i understood when you said it ;) [16:12] urgh. [16:12] what comes now sounds scary :) [16:13] kind of void* private fields [16:13] i would really try to find a way that relies on public data [16:13] whatever it costs [16:14] or get it upstream ;) [16:14] they might as well drop some private fields or so [16:14] but well ... i dont know the case ;) [16:14] so might be perfectly reasonable [16:54] moin. [16:54] [reed]: around ? [17:06] fta: ok so make firefox-3.0-branding | awesome-browser a depend of firefox-3.0 instead of receommends? [17:07] in that way i think that an upgrade to branding is enforced [17:07] yes [17:07] but then we could also consider to move the files to their own package [17:08] ok ... so most likely we end up firefox-3.0 depends on firefox-3.0-branding | webbrowser-3.0-branding [17:08] with firefox depending on firefox-3.0, firefox-3.9-branding [17:08] yes, that's what i wanted since the beginning :) [17:08] and webbrowser depending on firefox-3.0, webbrowser-3.0-branding [17:08] sure [17:09] but thats one more package [17:09] why should firefox depend on firefox-3.9-branding if firefox-3.0 already depends on firefox-3.0-branding | webbrowser-3.0-branding ? [17:10] (and firefox depends on firefox-3.0 | firefox-2) [17:10] do it like a tree, not like a graph [17:11] well... the user experience would be: sudo apt-get isntall firefox -> firefox branding [17:11] instsall webbrowser -> webbrwser branding [17:12] and if you install firefox-3.0 you get the firefox branding because its first in the depends options [17:12] but we also have a webbrowser-3.0-branding conflicts firefox-3.0-branding :) [17:12] and maybe even hther way around :) [17:15] fta: i have to think :) ... this is getting complex just for the sake of making it clean [17:15] but maybe thats just a feeling [17:15] it's tricky.. if a branding is gone, the corresponding launcher (script and desktop file) should go too, otherwise, it will create troubles. [17:15] fta: we cannot make the /usr/bin/firefox link go away [17:15] that will stay everywhere [17:16] it exists for compatibility reasons [17:16] and the launcher now gets converted i think [17:16] so people with webbrowser installed will start firefox expecting it and get webbrowser :( [17:17] and since both branding packages are mutually exclusive that isnt needed imo [17:17] well ... there will always be colleteral damage [17:18] fta: we can add the branding as an extension so that people that end up having webbrowser could theoretically find it in the "Get Ubuntu Extensions" dialog [17:18] not really as an extensions, but pretend to be an extension to get in that dialog [17:19] well... you will get firefox when firefox is isntalled [17:19] otherwise what is closest to firefox that is available as system install [17:19] which is similar to what we do with vi ;) [17:20] and people complain as well when they end up having just nvi installed and not wim [17:20] or vice versa :-P [17:44] bug #256261 [17:44] Launchpad bug 256261 in kdebase-workspace "keyboard won't work loginmanager starts" [Medium,Triaged] https://launchpad.net/bugs/256261 [17:50] asac, same with totem-xine vs totem-gstreamer, people keep complaining that xine should be default [18:05] rob said that using gstreamer is like bashing ones head against the wall :) [18:05] (rob from gnash) [18:05] he used that to explain why they forgot completely about gstreamer when rewriting the video backend [18:06] now they are on the devils brother ... ffmpeg ;) [18:06] too bad, gst is everywhere, and will even be in ff [18:06] yes. i said them that thi smeans end-of-dream ;) [18:07] that gnash can never ship on desktop CD ;) [18:07] now they recosinder i think [18:07] is liked ffmpeg when i was deep in mplayer [18:07] -is+i [18:10] <[reed]> emgent: yes [18:10] <[reed]> ? [18:18] <[reed]> emgent: you can message me, or just e-mail me at reed@mozilla.com === Knusper_ is now known as Knusper