[00:02] http://mozillalinks.org/wp/2008/04/firefox-reached-29-share-in-europe/ [00:06] jyväskylä [00:08] Could someone look at bug 198074? I have seen that behaviour (flash over content) before, and I suppose it's a fairly known bug. [00:08] Launchpad bug 198074 in firefox "Flash not showing properly" [Undecided,Confirmed] https://launchpad.net/bugs/198074 [00:09] thats a dupe [00:09] search for "flash always on top" [00:09] thats the master bug and should already have plenty of dupes [00:09] ok [00:09] maybe its eve nclosed on firefox side [00:09] search on flashplugin-nonfree [00:09] (firefox side for that bug was fixed) [00:12] lol http://www.screenage.de/blog/2008/04/24/ubuntu-landed-on-berlin-metro-system/ [00:12] yeah heard about that before ;) [00:12] Cool :) [00:13] cool stuff [00:17] asac: fta: upstream might have a patch-in-hand by tomorrow for bug 215728. They're hitting freeze real soon - is it feasable to have it pushed to a ppa to help upstream test it? [00:17] Launchpad bug 215728 in firefox-3.0 "High CPU Consumption" [Undecided,Confirmed] https://launchpad.net/bugs/215728 [00:18] jcastro, sure [00:18] fta: I told him to specifically attach it to the lp bug, so if you guys could watch it closely I would appreciate it [00:19] jcastro, do you mean, b5 + this patch in a ppa or the latest upstream snapshot containing this patch ? [00:19] expect a patch [00:20] my ppa is very close to upstream, ie, far ahead of b5 (hardy) [00:20] fta: he'll see when he gets there, he will put the details in the bug [00:21] the only problem with my own ppa is that lang packs are not compatible (our max version is too strict) [00:22] fta: campd is upstream. [00:22] fta: I've pinged asac about it, but he's probably asleep now, so pls ensure he's in the loop [00:22] and I have a headache. [00:23] jcastro: i am still here :) [00:23] woo [00:23] so I don't know how much jorge has relayed; [00:23] but this is a shittastic bug, that doesn't seem to manifest on other platforms [00:23] hi campd [00:23] there's no Great fix, but I'm working on some Better fixes [00:23] campd: why is sqlite so bad on linux? [00:23] dunno! [00:24] (and so much better on win?) [00:24] so the current fix is to tune cache size? [00:24] yes. [00:24] temporarily, during the update [00:25] when I say "current fix" I mean "the one I'm still working on" [00:25] campd: what is done during update? isn't that incremental? [00:25] so it's incremental, yeah [00:25] but unfortunately it needs to update an index [00:25] and the data it's adding is a) plentiful and b) deliciously random [00:26] so it ends up needing to insert into all sorts of different spots in the index [00:26] there's practically no decent locality [00:26] so the index is physically clustered? [00:26] not sure what you're asking; [00:26] hehe not sure either [00:27] maybe our words don't match [00:27] the real problem is that the index is Big (at certain points in the update process it has more than a million hashes) [00:27] and we throw a bunch of effectively-random data at it [00:27] I'm making up numbers here, but if the index has 5,000 pages [00:28] and we add 10,000 urls [00:28] we're going to be hitting most of those 5,000 pages, multiple times [00:28] and if the cache is only 1,000 pages, you're going to get a Lot of Thrash [00:29] this is mostly a problem during the Initial Population of the database, which encompasses the first few update requests [00:29] once you have the whole database, you're adding closer to 1,000 urls a week [00:29] which will be (relatively) easier on the cache [00:30] so the proposed fix is basically "let the cache get much bigger if it needs to" [00:30] "which will happen during those initial updates" [00:30] (I'm also considering adding "try to do these updates when the user is idle") [00:30] so how much mem do we expect then? [00:30] not entirely sure yet [00:31] (suggestions welcome!) [00:31] campd: one more question. you are talking about db indexes like in "create index ..."? [00:31] yeah [00:31] although the actual data itself shares the cache [00:31] campd: do you drop them before doing the import batch? [00:31] nope. [00:31] performance there would be dismal [00:32] once that big initial update is complete, we're adding like 10-100 urls at a time [00:32] to a database with like 500,000 urls [00:32] campd: i don't say "drop index" [00:32] regenerating that index would be Pure Pain [00:32] ah, what are you suggesting? [00:32] i mean: "drop index while doing import batch ... then when finish create it" [00:32] right [00:32] so regenerating that index over 500,000 urls [00:32] thats what we did when we imported huge datasets in the past [00:32] when you're only adding 10-100 [00:33] will be prohibitive [00:33] campd: well ... if its just the initial batch you could do that only then [00:33] yeah, that's a possibility [00:33] although the raw data is still an issue there [00:33] campd: do you know in advance how many you import? [00:33] nope [00:33] (we could Guess, but we don't really know) [00:34] then you might preparse X and cache them in memory, then decide if to just insert ... or insert as batch (e.g. without index) [00:34] yeah, it's possible [00:34] in my tests, dropping the index didn't help a whole heck of a lot [00:34] even during initial updates [00:34] ok, so the issue is not the index ? [00:34] not really [00:34] I mean, the index is part of the issue [00:35] yes, but if this doesn't really help, then it must be something else [00:35] it's the io [00:35] it's the fact that our current page cache is around 2000 [00:35] right ... so the raw data IO alone is too slow? [00:35] yeah [00:35] that would be really bad ;) [00:35] and we're kinda stupid about it [00:35] where "stupid" is "we do too much of it, because our cache is too small" :) [00:36] so what I'm looking at doing is basically choosing a cache size based on system memory; [00:36] are you using transactions? [00:37] yes [00:37] (which is why the page cache is useful at all; otherwise that page cache would be synced to disk after every operation) [00:40] so anyway, my current thinking is that a) currently our page cache is Just Too Small, Period [00:40] b) with a decent-sized page cache, the IO becomes negligable [00:40] c) we can reclaim the page cache memory after a successful update [00:41] so similar to what our bookmark/history code does [00:41] I'm going to look at sizing the page cache based on available system memory [00:41] during an update process. [00:41] if that is a proven pattern then i agree that this is the path to use [00:42] s/to use/to go/ [00:42] so what we'll do is choose some percentage of system memory [00:42] campd: how do you get "available system memory"? [00:42] PR_GetSystemMemory() :P [00:42] I'll check what that does [00:42] but it is Physical [00:42] yeah thats what i wonder [00:43] so that page cache grows to the size needed to contain the update; [00:43] asac, bug 151956 ... People complain that flashplugin-nonfree/mozilla-plugin-gnash are not installable from live cd, because multiverse repo is not enabled. We might include to ubufox to check if multiverse is enabled and to present the correct plugins list, maybe with a message "please turn on multiverse for more plugins". Of course, if that's possible... [00:43] Launchpad bug 151956 in firefox "Gutsy RC1 LiveCD can't find flashplugin-nonfree when using install plugin on firefox" [Undecided,Incomplete] https://launchpad.net/bugs/151956 [00:43] it's not Necessarily the size we dicide; [00:43] decide [00:43] so as an example [00:44] let's say we choose 7%; [00:44] (places uses 6%) [00:44] on a 512mb machine, that'll set aside 35megs [00:44] during the Big Scary Initial Updates, we might use up that whole 35 megs [00:45] during the update process (which, because we have very little IO, will only take like 10 seconds) [00:45] and we'll try to fit those updates into when they're idle if possible [00:45] once they're caught up to the current state of the list, we'll still ALLOW 35 megs [00:46] but it'll probably end up being much smaller, because we're only updating 10-100 urls [00:46] so it'll probably be a negligable increase [00:46] so that's the PLAN [00:46] I don't have a patch yet [00:46] Jazzva: multiverse _should_ work [00:46] and someone smarter than me might have a better solution :) [00:46] asac: Hmm, ok. [00:47] I'm idly curious why it's so much worse on linux === fta_ is now known as fta [00:47] campd: do you use same cache size on win? [00:47] yep. [00:48] I haven't done as vigorous testing there, but we've gotten no complaints [00:48] while ubuntu users seem to have quite a few :) [00:48] but regardless, this is a fix we should be doing everywhere. [00:48] yeah. but we use in-source sqlite ... so at least it should not be related to us using a system library [00:48] right. [00:49] that was actually my Very First Question, but I was able to verify on my linux box with the mozilla tree [00:49] so, yeah [00:49] personally i don't experience this [00:49] how can that be? [00:49] how big is your urlclassifier3.sqlite? [00:49] 22M [00:49] huh [00:49] then that's a good question :) [00:50] maybe those users are using some kind of strange filesystem [00:50] like fat16 [00:51] well, I'm seeing crappy performance [00:51] campd: could you confim this? [00:51] ah i see [00:51] (albeit with a contrived test case) [00:51] yeah ... but real user-experience impact? [00:51] haven't actually confirmed that, no [00:52] but my contrived test cases are pretty close to what you'd expect in real life [00:52] (I'm actually adding the real data, just at a different pace than the normal updates) [00:54] man, I have a gigantic headache [00:55] ok thanks for coming here. if you have a patch we can try let us know [00:55] yeah [00:55] I was talking to jorge about that; [00:55] I'm going to try to have a patch together tomorrow, discuss the approach with some other mozilla folks; [00:55] once we decide for sure which way we're going [00:55] I'd *love* to get as much testing as possible [00:56] well, we apparently have plenty "volunteers" here :) [00:56] yeah. [00:57] apparently asac? LOL [00:57] no those forum users that complain ;) [00:57] i don't see this bug unfortunately [01:00] Jazzva: actually i am not sure about that multiverse thing on CD [01:00] (and while I'm here, congrats on the release) [01:00] (it's been working pretty darn well for me here) [01:01] Jazzva: i always thought one can update sources.list ... and we had a bug about something similar against apturl, but i was assured that this was fixed. [01:01] Jazzva: so its worth to investigate i guess [01:01] campd: thanks! :) [01:02] well, whatever slightly-beta version I have now [01:02] campd: only thing thats unfortunate is that we missed RC1 by a week or so [01:02] I can't get to the damned mirrors to update :) [01:02] yeah [01:02] well tbh, I'm kinda glad you did [01:02] in a selfish using-your-users-as-testers sort of way [01:02] 'cause this was flying under our radar previously [01:03] but yeah, RC1 would have been good [01:04] i am not glad obviously ;) ... we spend a considerable amount of work to get xulrunner going for all apps and all. it was fun, but still having a beta browser in final release is kind of a decision that causes discussion for us :) [01:04] asac: Only thing I need is a Hardy cd :). I asked commenter to see if that's happening on hardy final [01:05] campd: i think we can sell this as a long term decision, but RC1 would have been so much easier to sell ;) [01:05] yeah, I know what you mean [01:05] asac: how good are you with scripting for greasemonkey? [01:05] I'm mostly thinking "if ubuntu hadn't been releasing, would this bug have caught our attention in time for RC1" [01:05] * gnomefreak thought of this while im eatting :( [01:06] campd: yeah ;) ... thats the good side. and i am confident that shipping 3 b5 was the right thing to do. [01:06] yeah. [01:06] this issue aside [01:06] (which, honestly, can be worked around by turning off the malware detection) [01:06] its still open how the press reacts on this, so far its mostly trollish users that question this [01:06] Does anyone know if TB supports groups in the address book. I looked at the help section and it was no help. [01:06] (which ffx2 doesn't have) [01:07] campd: right. i don't feel bad about this particular issue [01:07] most people just see "Beta" and run screaming ;) [01:07] nod [01:07] but ffx3 >>>>>> ffx2 [01:07] *particularly* on linux [01:07] exactly [01:07] as said. press didn't really trash us for that from what i can see [01:07] Sergeant_Pony: i havent tried but as i recall you can add groups whileing setting up accounts so i dont see why not but again i havent played too much with tb2 for a while other than everyday usage [01:08] run screaming from beta *shock*? Where's the fun if it's all stable? [01:08] for now [01:08] asac: you can reply that Google was still in beta for years... ;) [01:08] right [01:08] weve done it before and it worked out fine this time should be no differnet [01:08] i have no problem to argue for that [01:08] Volans: If Google was only :)... [01:08] Volans: it still is [01:09] *the only one [01:09] aside google searches people have had many issues with thier products [01:09] gnomefreak: yes, but bars are raised. people look closer [01:09] asac: true but isnt b5 rc1 than release (rough order) [01:09] previously the important thing was "latest" ... now it has to be "latest and most stable" :) [01:09] maybe next month and half or 2 depending on blockers [01:10] gnomefreak: yes. two month from now it will be forgotten [01:10] but people reading this might perceive this as "not mature" and could a long lasting bad idea [01:10] brand perception [01:11] most firefox bugs is profile, extentions, addons, but there are a few or more that are code itsself from what ive seen [01:11] asac: we really could have not made it default since we carry 2.0 with hardy [01:12] gnomefreak: the idea is to not change the default throughout the lifetime [01:12] but i dont see that big of an issue atm. stop using flash cut ff bugs to 1/4th of what they are [01:12] and ffox 2 would be really hard to support at some point [01:12] upgrading to ffox 3 would also give us bad reputation [01:12] asac: true thinking on LTS [01:13] its a catch 22. people are either gonna complain its not default or complain that it is since its not final [01:13] Hi all :) [01:13] hi l3on [01:14] IMHO 2months until final is better than not having it in for 3 years [01:15] are they planning on releasing xul and ff3 at same time? [01:15] gnomefreak: they don't release xul [01:15] they release it all together in ffox [01:15] we split [01:18] something really wrong here [01:19] still sucky network in PA? [01:19] 20:19 -NickServ(services@mozilla.org)- Nick gnomefreak isn't registered. [01:19] and shitty ns [01:19] but yes i have to deal with ISP tomorrow [01:19] gnomefreak: thats on mozilla server :) [01:19] not freenode [01:20] oh crap [01:20] If I'm in Mozilla Bugs, will I receive every comment for every bug to which mozilla team is subscribed? Just to know, so I don't subscribe myself for no reason :) [01:21] Jazzva: mozilla-bugs gets mail for bugs to which mozilla-bugs is subscribed [01:22] Jazzva: mozillateam accordingly only to bugs to which mozillteam is subscribed [01:22] but we don't subscribe mozillteam [01:22] hmm... ok [01:22] Thanks :) [01:22] Jazzva: but i am not sure if we get mail for firefox-3.0 [01:22] i think we don't [01:22] no idea how things are setup [01:23] i get mail for firefox + thunderbird [01:23] and all bugs to which mozilla-bugs is subscribed [01:23] Ok ... then I'll just subscribe myself to every one... Hope it won't produce too much mail :) [01:23] haha [01:23] it will [01:23] Hope gmail filters are good enough :) [01:23] gnomefreak: can you sort out how to subcribe to get all bugmail for all our packages [01:24] i looked a few days, but couldn't figure ... and i just don't remember [01:24] gnomefreak: do you still remember how this was done? [01:24] 02:23 < asac> gnomefreak: can you sort out how to subcribe to get all bugmail for all our packages [01:24] 02:24 < asac> i looked a few days, but couldn't figure ... and i just don't remember [01:24] 02:24 < asac> gnomefreak: do you still remember how this was done? [01:25] no idea if he got it [01:25] i can if i stay around long enough tomorrow [01:25] :) [01:25] i didnt get it til now === gnomefre4k is now known as gnomefreak [01:25] gnomefreak: not urgent [01:25] i just thought you remembered ;) [01:26] ok i will look tomorrow but IIRC you have to subscribe from main bug page on each package [01:26] gnomefreak: so jazzva needs to go to firefox-3.0 and subscribe to all bugmail there? [01:26] i think thats possible [01:26] but being mozilla team you are already subscribed to all mozilla bug mail [01:26] No, not all bugmail :) [01:26] just thought we subscribed mozilla-bugs in such a way [01:27] Jazzva: ? [01:27] we did [01:27] i did [01:27] Jazzva: thought thats what you wanted? [01:27] Jazzva: you are a memeber of team right? [01:27] gnomefreak: so maybe mozilla-bugs is not yet subscribed to firefox-3.0 and xulrunner-1.9? [01:27] asac: I thought that too ... then I realised I'm needing just the one I commented :) [01:27] Umm, right :) [01:27] I am receiving new bug reports, though... :) [01:27] asac: ff3 it should be i get them xul i dont get iirc so i will fix it [01:28] ok thanks [01:28] asac: is the packages up to date on the mozilla team LP page? [01:28] gnomefreak: remember that there are still lots of ffox 3 reports going to firefox package because of confusion [01:28] and are we talking extentions as well? [01:28] so maybe thats what you are seeing [01:28] worth to check i guess [01:28] asac: ill look into it int he morning and subscribe mozilla-bugs to all our packages [01:29] damn [01:29] extentions i may need help with them as i saw way too many to remember [01:29] * asac wonders if that is any better for his own bug performance [01:30] asac: Lots of mail coming your way after? [01:30] Jazzva: no idea :) ... my mailbox is always full with unread mails [01:30] thats all that i know [01:30] so i think so [01:31] but my mail setup really sucks in this regards [01:31] Work on it :) [01:31] *thinks* [01:31] ill look at it right now for a little bit and see if i cant figure an easy way of doing it [01:31] well ... i am sure that i don't have the time to process all New and Incomplete bugs. [01:31] maybe confirmed [01:32] at best only triaged and in progress ones :) [01:32] but thats a dream obviously [01:32] asac: you come up with greasemonkey scripts for comments and ill do the bug work all day and night ;) [01:32] Subscribing to all bug reports and forwarding it to another folder, hoping I'll find something interesting, or just to keep plain, old bug-browsing? [01:32] gnomefreak: i can give you those python scripts [01:32] Jazzva: tb and filters :) [01:32] asac: how do i run them? [01:32] gnomefreak: they are far easier to write and far easier to run [01:32] I don't think I am subscribed now ... only ~10 bug-reports came to my mail today. [01:33] gnomefreak: just chmod a+x .py [01:33] then ./xyz.py BUGID [01:33] you have to copy your cookies.txt file into the same directory you run them from [01:33] to be authenticated [01:34] and remember to replace at least Alexander with your name in the .py files [01:34] but i can imprve that [01:34] oh fuck my ISP [01:34] 02:33 < asac> gnomefreak: just chmod a+x .py [01:34] 02:33 < asac> then ./xyz.py BUGID [01:34] 02:33 < asac> you have to copy your cookies.txt file into the same directory you run them from [01:34] 02:33 < asac> to be authenticated [01:34] asac: yeah i saw that but is that how to run them? ./script bug # [01:34] 02:33 -!- gnomefre2k [n=gnomefre@adsl-221-44-39.rmo.bellsouth.net] has joined #ubuntu-mozillateam [01:34] 02:33 < asac> and remember to replace at least Alexander with your name in the .py files [01:34] 02:34 < asac> but i can imprve that [01:34] 02:34 < gnomefre2k> oh fuck my ISP [01:34] gnomefre2k: that might be different dependeing on what the script should do. the ones i posted all take just the bugnumber as argument [01:34] ok i can do that easy [01:35] ah ok [01:35] but i could also write something like: === gnomefre2k is now known as gnomefreak [01:35] ./post-bug-generic.py bugid --status=Incomplete --comment="Here comes your comment" [01:35] :) [01:36] or [01:36] ./post-bug-generic.py bugid --status=Incomplete --summary="New summary that is comprehensible" --tag=flash --comment="Here comes the comment" [01:37] oh no [01:37] am i here? [01:37] you are [01:38] sounds miserable [01:38] i feel with you gnomefreak :) [01:38] thanks [01:38] * gnomefreak really gonna lay into them [01:38] this is bs [01:38] package name for xul 1.8 is just xulrunner right? [01:38] if any of you need some bug script let me know. i can write it given some time ahead. [01:38] gnomefreak: yes [01:39] xulrunner-1.9, firefox-3.0 is the hot combo [01:39] everything else is just legacy [01:39] https://edge.launchpad.net/~mozillateam [01:39] yes, thats right [01:39] ubuntu-php-firefox-human is no longer around right? [01:39] gnomefreak: just for ff2 [01:40] php? [01:40] same with firefox-themes-ubuntu/ [01:40] that's what I thought you were asking about [01:40] https://edge.launchpad.net/ubuntu/+source/ubuntu-php-firefox-human/ [01:40] edgy was last afaik [01:40] edgy is EOS in a day or 2 [01:40] \o/ [01:40] hail on that [01:41] one upload less on firefox security updates [01:41] i will remove it from page should we keep themes package until ff2 EOS? [01:41] yes [01:41] we have that packagein universe [01:41] however we dont really care ... its the artwork team that should fix it imo [01:41] k [01:41] otoh havin git in that list doesnt hurt [01:42] its never been us afaik [01:42] its always someone else that did it [01:42] yeah [01:42] git? [01:42] where do you see git? [01:42] and given that all the scripts in there are writtin in php i wont like to work on it more than absolutely necessary [01:43] ah [01:43] gnomefreak: bad spacing: [01:43] kompozer is us or tony? [01:43] < asac> otoh having it in that ... [01:43] gnomefreak: tony [01:43] so should i remove it? [01:43] i don't want to put my name on something like that ;) ... ifpossible [01:43] and LP plugin package [01:43] gnomefreak: well ... he is in mozillateam :) [01:43] good point [01:44] no idea if thats appropriate [01:44] but i don't mind [01:44]