/srv/irclogs.ubuntu.com/2010/05/02/#ubuntu-devel.txt

=== dendrobates is now known as dendro-afk
* TheMuso sighs. Redhat writing their own init daemon.01:11
TheMusoOr Lennart is at least.01:11
=== dendro-afk is now known as dendrobates
psusiwho was it that maintains udisks?  got an interesting problem with dvd+rw... it seems udisks is confused by the fact that a blank dvd+rw is not "blank" so it just ignores the media and the drive vanishes from nautilus02:01
imbrandonTheMuso: i thought fedora ( and rh by proxy ) used upstart ?02:16
TheMusoimbrandon: they do02:17
imbrandonTheMuso: hum, then that is diserning02:18
=== dendrobates is now known as dendro-afk
respireideas for ubtuntu03:26
respiresome guy made a script called rmkernel, it deletes unused kernels since they hog space its very simple03:26
respireubuntu should do it on boot and if nobody used a kernel for a week ask them if they want to delete it or to never ask again03:27
respire(skip with timeout)03:27
respirealso there was this program i saw that was supposed to help delete unused ubuntu programs03:28
respireit was awful, rubbish a total failure03:28
respireim prepared to help and in fact mostly write a replacement if someone else will too :)03:28
respirewhy is there no services config tool for the sysv stuff03:41
respireyou could make a gui for that easy03:41
psusipitti, finally got those logs you asked for in bug #55892603:52
ubottuLaunchpad bug 558926 in udisks "Open files on ejected media cause automount of new media to break" [Undecided,Invalid] https://launchpad.net/bugs/55892603:52
George_EHey, just wanted to thank everyone for making Ubuntu 10.04 a tremendous success!04:44
George_EYou guys rock!04:44
George_ELucid boots in literally 10 seconds on my computer - compare that to Vista which takes > 5 minutes.04:46
jdongGeorge_E: glad to hear! believe it or not there's STILL room to improve with our bootup sequence!04:56
George_EReally.... I didn't think that was possible. Well, keep up the good work. Love the new look too!04:56
jdongGeorge_E: http://jdong.mit.edu/~jdong/ureadahead-pack.svg this image is a semi-accurate look at the files being accessed while booting, on a fresh Ubuntu Lucid setup04:57
jdongGeorge_E: so... once we figure out a way to "defragment" (move all those files to one spot on disk), you can expect another speedup04:57
George_EOh good. Also shutdown time is wicked fast. Vista takes around 1:30 - 2 minutes to shutdown... :(04:58
virtualdon a fresh install?04:59
George_EWell... not exactly.04:59
jdongvirtuald: yeah, I did 3 more fresh installs just to verify04:59
jdongGeorge_E: well we should actually give Vista a bit of credit for the bootup/shutdown systems05:00
George_EHow so?05:00
jdongGeorge_E: Vista actually has an AI algorithm for placing blocks on disk next to each other to minimize seeking05:00
jdongGeorge_E: and Windows has had a parallel bootup (like Upstart does for us) since NT.05:00
jdongGeorge_E: the reason Vista is slow is likely because it's close to 20GB while we're 2GB or so altogether...05:01
jdongif you normalize to that size discrepancy, I bet Vista is achieving faster MB/s disk reads at bootup05:01
jdongthough that doesn't excuse its monstrous size05:01
George_EVista has a lot of legacy support too. That has an effect on startup time.05:01
jdongit loads a lot more services too05:02
jdongbut I do feel their bootup and application launch systems are somewhat "smarter" than us, though now we've got most of the tools needed to catch up05:02
jdongas soon as the EXT4 defragmentation API stabilizes a bit more, we're in business05:02
George_EYes. Ext4 is a nice step forward.05:02
jdongthat graph was generated with data from our existing bootup accelerator (ureadahead)...05:02
virtualdso you're working on bug 1 :)05:03
ubottuhttps://bugs.launchpad.net/ubuntu/+bug/1 (Timeout)05:03
jdongso we already KNOW exactly which blocks needs to be moved together05:03
George_Eubottu: Haha. That is a *real* bug.05:04
virtualdjdong: does this affect ssd's at all (performance wise)?05:06
jdongvirtuald: it certainly does...05:07
jdongvirtuald: SSDs have some minimal block size which is fairly large, sometimes as big as a couple hundred KB's05:07
virtualdo.O05:07
jdongvirtuald: the smallest unit you can ask it to read is a "block"05:07
jdongso if you've got a bunch of text files that have configuration data stored everywhere on the disk...05:07
jdongyou might be reading WAY more data than you really need to05:07
virtualdi see05:08
jdongso I think a combination of good placement on disk + proactive/predictive prefetching into cache is important for everyone05:09
ionjdong: And after that, reordering blocks based on post-boot usage patterns as well. :-)05:10
jdongion: yup indeed.05:11
* George_E is not so sure SSDs are all they're cracked up to be :)05:11
jdongGeorge_E: they're expensive and kinda sorta band-aid our current seek performance grief05:11
jdongGeorge_E: but I honestly don't see any reason why everyone needs to be moving to SSD's05:11
jdongGeorge_E: I think optimally, SSDs should play a role like ZFS's L2ARC cache hierarchy05:12
jdongwhere fast SSDs sit in front of a slower but larger HDD pool05:12
jdongand a good replacement algorithm selects the best data to place in the SSDs05:12
George_EThey have benefits (lower seek times, etc.) but they're not perfect (slower write times, etc.).05:14
jdongGeorge_E: huge cost disadvantage, larger block size, uneven wearing issues, unproven reliability track record, and so on :)05:15
George_EThe uneven wearing issue is my biggest concern. (And a concern with all flash-based devices)05:16
jdongGeorge_E: yes and the algorithms that take care of that seem to nullify quite a bit of the SSD performance advantage over time05:16
temugenjdong: as an L2ARC the reliability/wearing wouldn't be as much of an issue if replacement were easy05:17
temugenof course that doesn't justify the cost factor any more05:17
jdongtemugen: and hopefully the L2ARC algorithm wouldn't be finicky to keep replacing its contents :)05:17
George_EIf you need fast access to non-system data, you can always use a ramdisk... but for system use... hmmm.05:17
jdongGeorge_E: well the ramdisk effect is what RAM caching is supposed to do05:18
jdongGeorge_E: the L2ARC would be a second level of cache that sits between RAM and mechanical disks05:18
George_ETrue. Though it's only on a small scale.05:18
=== aburch is now known as aburch_
=== aburch_ is now known as aburch
=== StevenK_ is now known as StevenK
Damascenehi, joaopinto08:12
joaopintohello Damascene08:13
Damascenedid any one add a session about the vte issue?08:14
TrueTomIs there a way to offer the installation of a package based on an udev event / newly pluged-in USB device?09:00
RAOFTrueTom: That's a fine question.  I'm pretty sure the answer is “yes”, but you'd need to write the udev rule.  A logical place to put a generalisation of this code (if it isn't there already) would be Jockey.09:03
Damascene /j #launchpad09:12
TrueTomRAOF: Hm, the "branding" of Jockey suggest to only use it for proprietary drivers... I want to install 'irda-utils' when an USB-IrDA device is pluged-in, which would look weird since it's free...09:33
switchgirlanyone found a fix for Bug 56954311:24
switchgirlit's very urgent11:24
ubottuError: Could not parse data returned by Launchpad: The read operation timed out (https://launchpad.net/bugs/569543)11:24
switchgirlwell?11:50
switchgirli guess not11:50
jpds!weekend11:55
ubottuIt's a weekend. Often on weekends the paid developers and a lot of the community may not be around to answer your question. Please be patient, wait longer than you normally would or try again during the working week.11:55
mdkeswitchgirl: that could be bug 552227 - there is a workaround described in the comments there, give it a try and mark your bug as a duplicate of that if it works11:56
ubottuLaunchpad bug 552227 in gwibber "gwibber-accounts crashed with KeyError in get_account_data()" [Undecided,Confirmed] https://launchpad.net/bugs/55222711:56
mdkeswitchgirl: if it's not the same, then you'll need to wait for the developers to fix it unless you can find a fix yourself11:57
switchgirlmdke, umm i tried deleting the couch12:00
switchgirlnothing happened12:00
mdkeswitchgirl: this was the workaround I had in mind - https://bugs.launchpad.net/ubuntu/+source/gwibber/+bug/552227/comments/5512:01
ubottuLaunchpad bug 552227 in gwibber "gwibber-accounts crashed with KeyError in get_account_data()" [Undecided,Confirmed]12:01
mdkeswitchgirl: not sure if it's the same issue you have or not though12:02
switchgirlmdke, no it's not i can auth facebook12:22
switchgirlnot do the next step12:22
switchgirlie not create the account12:22
switchgirlsame with twitter12:22
switchgirlnot with flikr12:22
switchgirli use google dns12:23
mdkeswitchgirl: I wondered if the cause might have been the same. Anyway, I don't know much about gwibber so can't help further12:45
switchgirlok mdke where can i poke the dev's?12:46
mdkeswitchgirl: you can try #ubuntu-desktop during a working day, I guess. But it's one among many other bugs so if you don't have a fix, you'll have to be patient12:47
switchgirlmdke, its a buggy anoying thing for fresh install's of ubuntu that loads are coming across12:48
mdkeswitchgirl: there are plenty of such bugs, I'm afraid. We have 85000 bugs, with over 1000 ascribed with "High" importance12:48
mdkegwibber seems to have quite a few crashes unfortunately12:50
switchgirlmdke, i dunno whyit was included in the 10.4 release then12:59
kklimondabecause it's linux and now windows. we include stuff that mostly works and let users help test it. (my personal opinion, probably noy an official one ;) )13:04
kklimondaand not*13:05
mdkekklimonda: not really - the focus of a long-term support release like this one is stability. Probably the reason is that gwibber mostly works for those people who tested it13:33
kklimondamdke: is it realle stability? I know it's supposed to be but the amount of changes (or maybe their nature) that has been pushed into this LTS says otherwise. it does seem to me that we use users as testers, which I personally don't mind. gwibber mostly working (because it does work for me most of the time indeed) looks like an example - users are going to test it and we'll have more data to fix13:39
kklimondait in the 10.10 release.13:39
imbrandonyes its really stability , there will always be cases that arent caught, change != not stable13:40
=== dendro-afk is now known as dendrobates
joaopintokklimonda, I agree with you, there were a significant changes on this release which increase the likely of problems, but I don't see how it could have been done better, not keeping the release goals and schedule14:55
joaopintoit has been stable for most people involved in the development phase14:56
=== almaisan-away is now known as al-maisan
=== al-maisan is now known as almaisan-away
=== dendrobates is now known as dendro-afk
joaopintoanyone noted a bug report about the gnome theme randomly failing to be set on the first login ?16:36
=== dendro-afk is now known as dendrobates
=== dendrobates is now known as dendro-afk
=== dendro-afk is now known as dendrobates
Bernardohi all16:54
BernardoI am trying to port the patches for the poulsbo drivers and xorg 1.7.x from mandriva to lucid, but I am stuck initializing DRI. Anyone can help me?16:55
imbrandonBernardo: you might try #ubuntu-x ( and durring the week for even better response )17:22
=== dendrobates is now known as dendro-afk
=== dendro-afk is now known as dendrobates
Bernardoimbrandon: thanks!17:36
imbrandonBernardo: np17:37
anilHi All17:38
anilthis is my first time here17:38
anilI wanted to know, if Ubuntu 10.04 LTS now support Indian Languages natively? Thanks17:38
anilAlso any GUI builder recommended for ppl coming from Windows Dev to Linux Dev? Thanks17:39
joaopintoHello anil, the best channel for support in #ubuntu :)17:45
imbrandonanil: as far as the dev gui builder try monodevelop ( sudo apt-get install monodevelop ) but there are lots of them17:46
anilthanks Imbrandon and joapinto17:50
anilsorry got into the wrong room17:50
anilCheers!17:50
=== yofel_ is now known as yofel
joaopintoanil, check the quickly project, I personally don't recommend monodevelop, my tests shown itsn't not stable17:50
joaopintoerm, gone17:50
joaopintoor is it quicky17:50
imbrandonquickly17:50
joaopintoops :)17:50
imbrandonbut monodevelop isnt stable ? really ? i use it every day without issue17:51
joaopintoimbrandon, go to help, search for ()17:57
joaopintothe last time I have tried it crashed, the bug is open for more than 1 year :)17:57
joaopintoI didn't test run it on lucid yet17:58
Laneyworks here18:03
joaopintook, time to check it, maybere there is an usable RAD tool now :P18:04
joaopintoLaney, thanks for testing18:10
joaopintotime to close the bug18:10
=== dendrobates is now known as dendro-afk
joaopintostill not stable, 1 minute to find a new bug, http://www.ubuntu-pics.de/bild/55823/screenshot_002_3WwGI6.png18:14
joaopintoon a plain solution creation wizard18:15
joaopintoimbrandon, http://www.ubuntu-pics.de/bild/55824/screenshot_003_9BOWZ8.png <- I am missing something at the top :) ?18:16
joaopintoam I...18:17
BAKTUNHelp! Ubuntu 10.04-64.bit - exFAT driver link please. My English is not good. Please Help.19:18
=== radoe_ is now known as radoe
MachinTrucChosehiya19:21
BAKTUNDo not help me?19:22
MachinTrucChoseNot sure if this is the right channel to discuss Ubuntu suggestions. Is it? I'm trying to gauge the feasability of something.19:22
MachinTrucChoseand I don't think the people in the main channel would be the sort to udnerstand the technical difficulties19:22
micahg!brainstorm19:24
ubottuPost your ideas for Ubuntu at http://brainstorm.ubuntu.com and vote for the ones you like!19:24
MachinTrucChosethat's probably for the better, yeah19:25
MachinTrucChoseis the brainstorm account list separate from the ubuntuforums one?19:28
MachinTrucChoseevidently19:28
MachinTrucChoseallright thanks for your help19:28
=== mako_ is now known as mako
=== cr3_ is now known as cr3
xaver__Hello all, hello Keybuk i have the problem with mountall too. I tryed to use your fix (2.15~ppa1) but nothing changed :( I use 64bit kubuntu, i tryed the standard kernel and 2.6.33 and i use cryptdisks-early. At the moment i use the whorst fix (dpkg --force-depends -i mountall_1.0.2_amd64.deb). The Problem startet with the update to 2.14.21:18
=== NullPointerExcep is now known as dmb

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!