/srv/irclogs.ubuntu.com/2009/03/03/#ubuntu-devel.txt

phil_ps1general question regarding libraries....the answer will save me time...00:02
phil_ps1is it possible to have a program and its plugin both use the same instance of a library?00:03
phil_ps1there is a problem with globals...00:03
maxbprovided it's a dynamic library (.so) it should work fine, IIUC00:06
lifelessphil_ps1: its normal00:08
lifelessphil_ps1: its a problem for a progrm and plugin to use different instances00:08
phil_ps1I don't understand how to make them the same instances...00:09
phil_ps1*how to make them use the same instances...00:09
lifelessthey will by default00:09
phil_ps1okay, maybe this is not the problem.  I just read it was...have to verify it00:10
lifelessyou have to go to some effort [or some special edge cases] to avoid it00:10
phil_ps1read it on a bug report....https://bugs.launchpad.net/ubuntu/+source/pidgin-otr/+bug/24868400:10
ubottuLaunchpad bug 248684 in pidgin-otr "pidgin resolver process crashes with ldap accounts" [Undecided,Confirmed]00:10
lifelesspackaging is potentiall one such edge case :P00:10
phil_ps1asked here because people would know about libraries...sorry if its the wrong topic...00:11
lifelessphil_ps1: https://bugs.edge.launchpad.net/ubuntu/+source/pidgin-otr/+bug/248684/comments/700:11
ubottuLaunchpad bug 248684 in pidgin-otr "pidgin resolver process crashes with ldap accounts" [Undecided,Confirmed]00:11
lifelessphil_ps1: not libraries in general, one specific library, apparently.00:12
phil_ps1yeah that is what I read...trying to find a fix...without exhaustively removing globals00:12
phil_ps1okay, I will read more about libgcrypt...thanks lifeless00:13
lifelessphil_ps1: so, its nothing to do with libraries00:16
Skiess1graphics on my Lifebook C1020 default to vesa, which I think is wrong...00:16
lifelessphil_ps1: imagine you have two different uses for the gcrypt functions in one program, - two unrelated uses.00:16
phil_ps1okay00:17
=== bdefreese2 is now known as bddebian
lifelessphil_ps1: as long as that works, without the programing needing its own globals, it will work fine if the gcrypt stuff is in alibrary; if not it won't00:17
lifeless[IOW, globals are bad, get rid of them]00:17
phil_ps1yeah, pidgin dies when a whole bunch of threads are spawned...00:17
phil_ps1so there is a shared state00:17
phil_ps1yes, I agree, globals are bad00:18
phil_ps1(what does IOW mean, I am new. :)00:18
StevenKIn Other Words00:19
phil_ps1I should make a program that first finds globals...would help in my parallel computing research...00:19
phil_ps1thanks StevenK00:19
phil_ps1oh, wait I have one....00:19
phil_ps1by my professor...00:19
phil_ps1for grading student's assignments00:20
phil_ps1we enforce no globals, no public data, cyclomatic complexity of function < 10, line count of function < 50 lines00:20
phil_ps1maybe things in here will fix my problem...http://olympus.het.brown.edu/cgi-bin/dwww?type=file&location=/usr/share/doc/libgcrypt11-doc/html/gcrypt_2.html#SEC900:23
phil_ps1also my ibm desktop ubuntu installation doesn't work...00:24
phil_ps1I get as far as a login screen and then just a black screen00:24
phil_ps1is this because no video driver?00:24
slangasekTonio___: knemo seems to have the same problem, no FFe for a new package00:25
slangasek(and not even a needs-packaging bug for this one)00:25
lamontwhy is my jaunty debootstrap pulling in an MTA?00:25
lamontadmittedly, the script is also installing a boatloadd of build-deps00:25
Tonio___slangasek: yeah, I discussed this with Riddell... it was in the repos as a KDE3 app but was removed in the KDE4 transition...00:25
Tonio___slangasek: shouldn't be considered a new app for him, as well as kblogger00:25
slangasekRiddell: ^^ is knemo ok to re-add to the archive (FFe)?00:25
=== Tonio___ is now known as Tonio_
slangaseklamont: which MTA?00:26
Tonio_slangasek: I know a bug would have been nice, but I expected him to take care of that sooner, hehe :)00:26
slangaseklamont: oh, you're running a script - SEP :-P00:26
slangasekTonio_: ok, I can just leave those in the queue for Riddell to review ;)00:27
Tonio_sladen: yup, that should concern kblogger and knemo, and skrooge is waiting for your or his approval...00:27
lamontslangasek: halley:~lamont/make-chroot.sh00:28
Riddellslangasek: yeah I'll get to them tomorrow00:29
slangaseklamont: so maybe the 'ssmtp' listed as a build-dep is the clue? :)00:30
slangaseklamont: also, did you mean to install git-core instead of git? :)00:30
lamontheh00:31
lamontdunno00:31
slangasekoh, you're grabbing the build-deps /of/ the packages listed in builddep... hmm00:31
lamontslangasek: it'll have git-core and a few other things (kernel builddeps) before I'm done with it00:31
slangasekanyway, smartmontools Recommends: mailx | mailutils, mailx Depends: m-t-a00:32
lamontah ha00:32
Tonio_Riddell: thanks :)00:32
lamontit was more one of "wtf??" without otherwise looking at it00:32
mrooneyHmm, anyone have any input on bug 336992, is that expected?01:37
ubottuLaunchpad bug 336992 in ubiquity "UUID of existing partitions are changed" [Undecided,New] https://launchpad.net/bugs/33699201:37
=== asac_ is now known as asac
greg-gpitti: what is the likelihood of getting an update of apport for hardy/intrepid to include the apport-collect feature? as it is a new feature I don't believe it qualifies for a SRU.02:07
mrooneygreg-g: that could be pretty awesome02:18
greg-gpitti: after testing it, I see it requires a new dependency (python-launchpadlib) so that probably decreases the chances even more.02:19
greg-gmrooney: yeah, it would be great02:19
calcdoko: are you in charge of libxalan2-java? :)02:24
roy_hobbsHey, if I'm commenting on a bug in launchpad, how do I reference a bug properly so that it shows up as a link to that bug?02:56
ScottKBug #nnnnnn is enough02:57
roy_hobbsthanks02:57
ScottKYW02:57
leftyfbwhere can I go to find the changes in the latest kernel/headers/restricted modules ?03:35
ebroderI know this is a little off-topic, but does anyone know of a package that will list the files that dpkg's database doesn't know about?03:48
TheMusoebroder: apt-file03:50
LaserJockTheMuso: will that list files that dpkg doesn't know about?03:51
ebroderNo, no - I'm trying to figure out where all of my data is so I know what to backup on this machine. I want to know the files that are currently on my system that dpkg didn't put there03:51
TheMusoLaserJock: it will list files from packages that dpkg doesn't yet know about03:51
TheMusoebroder: ah ok03:51
TheMusowell you could script something.03:52
TheMusoget a list f all files on your system, dump all package file lists of isntalled packages, and compare.03:52
ebroderYeah, sure03:52
ebroderThat was what I was going to do if someone hadn't done it already :)03:52
LaserJockI suspect that'll be very unhelpful though03:52
leftyfbAnyone know where can I go to find the changes in the latest kernel/headers/restricted modules ?03:53
LaserJockmaybe not "very", but there's a lot of stuff there that are created postinst, for instance03:53
TheMusoLaserJock: good point03:53
ebroderLaserJack: Maybe. I'm willing to see what it comes up with. It might be filterable by-hand03:53
LaserJockand you've got things like /var03:53
ebroderHmm...yeah, but a lot of the files I want this script to track down are in /var. Maybe I can come up with a relatively short blacklist03:54
LaserJockanyway, not saying you shouldn't do it, but you're like to get thousands of files and going through by hand might be a bit tedious03:54
LaserJock*likely03:54
foxbuntuleftyfb, look here: http://packages.debian.org/search?keywords=kernel&searchon=names&suite=stable&section=all03:58
leftyfbfoxbuntu: how does that help me?03:59
leftyfbthat's not ubuntu or the latest kernel03:59
foxbuntuleftyfb, the Ubuntu kernel comes from upstream Debian04:00
leftyfbI didn't think every update came from upstream. There is a kernel dev team at ubuntu04:00
LaserJockleftyfb: are you wanting changelog or patches?04:00
foxbuntuleftyfb, they submit upstream04:01
leftyfbhttps://bugs.launchpad.net/ubuntu/+source/linux/+bug/32287904:01
ubottuLaunchpad bug 322879 in linux "Verizon CDMA card no longer works" [High,Incomplete]04:01
LaserJockfoxbuntu: not everything comes from Debian04:01
dtchen_leftyfb: http://kernel.ubuntu.com/git?p=ubuntu/ubuntu-jaunty{,-lrm}.git;a=summary04:01
leftyfbi'm looking to see if that was fixed in the latest update that was pushed yesterday04:01
foxbuntuLaserJock, not saying it does04:01
dtchen_leftyfb: replace jaunty with the intended Ubuntu release04:01
leftyfbit seems to be working, but I'd like to confirm it was fixed and then submit that that bug be closed04:01
leftyfbdtchen_: 403 Forbidden - No such project04:02
dtchen_leftyfb: that's shorthand for http://kernel.ubuntu.com/git?p=ubuntu/ubuntu-jaunty.git;a=summary and http://kernel.ubuntu.com/git?p=ubuntu/ubuntu-jaunty-lrm.git;a=summary04:02
ebroderLaserJack: Apparently `diff -u <(sort -u /var/lib/dpkg/info/*.list) <(find / -xdev)` runs pretty damn fast :)04:04
leftyfbdtchen_: thanks ... might have found what I was looking for04:06
leftyfbdo you think "  * USB: gadget: cdc-acm deadlock fix" is what fixed the bug i posted previously?04:06
ScottKfoxbuntu: The kernel is one place we explicitly do not take from Debian.  Generally we are on a newer version than Debian and the kernel team works directly from upstream kernel sources.04:09
foxbuntuScottK, ah, well thanks for clearing that up for me04:10
TheMusoHowever the way the kernels are built is loosely derived from Debian.04:10
ScottKTrue.04:10
ScottKSo if your interested in kernel packaging methods, Debian is a good place to look.04:10
Ademandoes anyone know how I could run a process after a given amount of idle time? (as root) is that something i could do via upstart?06:34
jdong_haha what's "idle time" defined as?06:51
Ademanjdong_: user idle time I guess?  based on mouse/keyboard use.07:00
jdong_I don't think at this time that's an upstart answerable question07:00
jdong_I guess you can have a proxy in each user's X session that reports whether it's idle07:01
Ademanjdong_: :-/  any idea how gnome-screensaver does it?  I'm not sure I want to depend on the same mechanism, but if that's the best i've got i'm interested07:02
brycetjaalton_: hmm, on fdo #19574 it is said that our patch 156 should be upstream, however the change doesn't appear to be present in our xserver git tree07:02
Ademani suppose xscreensaver probably uses the same method, whatever it is,07:02
brycetjaalton_: this is our bug 311254 btw07:02
ubottuLaunchpad bug 311254 in xorg-server "Xorg crashed with SIGSEGV in CopyKeyClass()" [High,Fix released] https://launchpad.net/bugs/31125407:02
jdong_Ademan: yeah I would guess it's a *-screensaver API call on the GNOME side07:03
jdong_Ademan: also look at how gnome-power-manager grabs idle time for DPMS blanking07:03
jdong_because that also works on the KDE side07:03
Ademanthanks jdong i'll poke around07:04
jdong_which leads me to believe there's some DBUS-ish call for it07:04
jdong_or other freedesktop.org tpye call07:04
Ademanhrm, i didn't think kde 3.x used dbus at all... either way... freedesktop may have a spec for that, thanks07:04
=== tjaalton_ is now known as tjaalton
=== Trewas666 is now known as Trewas
pittiGood morning07:33
ion_Ditto07:33
bluefox_http://sourceware.org/ml/binutils/2009-03/msg00032.html  Relevant to anyone who likes playing with the toolchain07:33
pitticalc: ah, thanks for the heads-up07:33
bluefox_and also because I'm crazy and need attention so I can fill an information gap (i.e. why the hell does/doesn't this work and why hasn't it been done this way if it actually works as I think)07:34
pittigreg-g: apport-collect> indeed; you can get it from http://people.ubuntu.com/~pitti/scripts/apport-collect and ask people to run it; it will point out that you need to install python-launchpadlib07:34
a|wencprov: can see you are a soyuz contact, i'm sorry for disturbing you this unorthodox way, but we are a little in a hurry, kde4.2.1 is due out today and we've reach the limit in the ppa for release packaging -> https://answers.launchpad.net/soyuz/+question/6292507:34
Hobbseemorning pitti07:36
Mithrand1ryo, Hobbsee07:37
* Hobbsee stomps on Mithrand1r's feet07:37
Hobbseehello, imposter!07:37
=== Mithrand1r is now known as Mithrandir
* ion_ returns Mithrandir’s i, thanks for the loan.07:37
Mithrandir08:37 -NickServ(NickServ@services.)- 31 failed logins since last login.07:38
Mithrandirhmm07:38
StevenK31 imposters?07:38
Mithrandiror just one who's trying a fair bit.07:38
Hobbseeunless it was yours :P07:39
Mithrandirlast time I checked, I did not irc off a host in Germany.07:40
StevenKPerhaps you need to check again07:41
Hobbseeblame pitti.07:41
lifeless.com?07:41
Hobbseeit must be pitti's fault.07:41
pittisheesh!07:44
StevenKpitti: architecture-mismatches.txt looks tasty. I'm happy to sort it out, but should I be demoting or promoting?07:53
lifelessugh, I'm lost where on cdimage.ubuntu.com/releases/intrepid/release is a plain ol 32-bit live CD07:55
StevenKlifeless: It's on releases.u.c07:55
StevenKhttp://releases.ubuntu.com/intrepid/07:55
lifeless*thats* not confusing at all. No really.07:55
lifeless(thanks btw)07:58
lifelessI could rant more if you like07:58
lifelessif hal loses the plot08:12
lifelesshow do I get it to release the .hal-mtab lock and start over08:12
pittiStevenK: promote; it's recommended by libgpod408:46
pittiStevenK: that is, gpod-common; python-gpod doesn't have main rdepends, so that should be demoted08:47
StevenKpitti: Okay, gpod-common gets a promote, and python-gpod gets demoted. I'll sort it out.08:47
pittiStevenK: great, thanks08:47
seb128pitti, StevenK: if somebody of you does some NEW could you give a quick review to nautilus-sendto-universe and evolution-mapi? I did work with the contributors on those so would be nice to have somebody else checking too09:01
=== simira_ is now known as Simira
loolmvo: You might recall I had lost my keybindings in compiz after an upgrade?  this was on my desktop and it hit my laptop yesterday (hadn't updated it for a while)09:30
seb128lool: is the gnome compat option enabled?09:41
loolseb128: I can't check on the laptop right now, but it was enabled on my desktop09:54
mvo_lool: I suspect the gnome compat module, I'm not 100% sure what the best way to fix it is, but I have it on my radar09:55
loolseb128, mvo_: Was disabled indeed10:00
loolmvo_: Ok if you know about it then I'm fine10:00
=== Tonio__ is now known as Tonio_
mok01Keybuk: ping10:40
Keybukmok01: hello10:40
mok01Keybuk: hi10:40
mok01Keybuk: have you seen I've proposed a branch for merge with m-o-m10:40
mok01?10:40
Keybukno10:41
mok01Keybuk:  https://code.edge.launchpad.net/~ubuntu-core-dev/merge-o-matic/trunk/+merges10:41
mok01Keybuk: it's a couple of very simple things10:41
Keybukcould you e-mail me, and I'll take a look at some point10:43
mok01Keybuk: sure10:43
Keybuksomeone else is working on a redesign of MoM's pages (RainCT), you might want to co-ordinate with him10:43
mok01Keybuk: it's rainct10:43
mok01Keybuk: will do, ok! Thanks!10:44
=== directhe` is now known as directhex
Keybukpitti: gnargh11:07
KeybukI have to update all my slides11:07
KeybukDeviceKit is no more11:07
pittiKeybuk: huh?11:07
TheMusowooa if thats true11:08
pittiKeybuk: 003 was just released yesterday11:08
pittihttp://hal.freedesktop.org/releases/11:08
TheMusoas in my goodness11:08
pittiKeybuk: is it DavidKit now?11:08
Keybukdid you read the release notes? :)11:08
pittiKeybuk: no11:08
pittiwhat, back to hal?11:08
Keybukah, read them ;)11:08
pittiKeybuk: url?11:08
Keybukno11:08
Keybukbasically the current DK will become simply the D-Bus API to udev11:09
Keybukand the D-Bus name will be just org.kernel.udev11:09
Keybukyou'll still have DeviceKit-power and stuff11:09
pittiaah11:10
pittiKeybuk: I didn't see release notes on http://cgit.freedesktop.org/DeviceKit/DeviceKit/tree/11:10
TheMusoare there still plans for device kit sound, as that seems pointless now, at least to me11:10
KeybukTheMuso: pointless how?11:10
pittiKeybuk: nice if it just gets integrated into udev11:10
Keybukhttp://lists.freedesktop.org/archives/devkit-devel/2009-March/000123.html11:11
TheMusoKeybuk: well most sound related stuff goes through alsa-lib11:11
TheMusoif not all of it11:11
KeybukI didn't know there were plans for a DK-sound11:12
Keybuksince HAL doesn't really have any kind of sound API right now11:12
TheMusoKeybuk: I thought there were, but I don't know where i heard that11:12
lifelessis there an offline bug reader for lp ?11:27
pittiemail?11:38
Keybukrandom Q of the day11:41
lifelesspitti: bah11:41
Keybukdoes anyone have a machine that doesn't have ACPI anymore?11:41
lifelesspitti: that doesn't go back in time11:41
lifelessKeybuk: is that new or old? I have a pentium that I still can't boot a current ubuntu kernel on ...11:42
Keybuklifeless: when was the last time you could boot it?11:42
lifelessKeybuk: couple of releases back, its my firewall :P11:42
Keybukdid it work on hardy?11:42
geserlifeless: try leonov, iirc it has some caching so you can use it offline11:43
lifelessKeybuk: don't think so, I thik thats where I started complaining to ben :P11:43
lifelessKeybuk: anyhow, if you want before-acpi, that may qualifyt11:45
KeybukI'm asking because we don't have any bugs that I can see that non-ACPI machines aren't working anymore11:46
Keybukand we think they've been broken for a couple of releases now11:46
Keybukand since nobody's complained11:46
Keybukand we hadn't actually noticed they were broken11:46
apwlifeless, i have an old clunker which does work even though acpi says "noooooo"11:46
Keybukthat maybe we should just note that we don't support them anymore :p11:46
lifelessKeybuk: how can I tell its non-acpi ?11:46
apwlifeless, what is the symptoms?11:46
apwif you can't boot it it is hard11:47
lifelessapw: my one, is hard-lock on boot11:47
apwlast thing it says?11:47
lifelessapw: no monitor :P11:47
cjwatsonI don't understand this "nobody's complained" mentality11:47
apwheh11:47
lifelessapw: and I last tried about 9 months back11:47
cjwatsondoesn't it mean "I haven't seen any complaints"?11:47
lifelessapw: because its such a high hurdle to test11:47
cjwatsonunless you read all bug reports and all forums posts :-)11:47
lifelessbest way to tell if a machine (thats running) is acpi ?11:48
apwcjwatson, yeah it means the machine is blowing up and noone has brought it to the attention of <someone who cares>11:48
cjwatsonI suspect it may just be that developers are biased towards newer machines, and so there is a higher chance that people with older machines will report problems in the "wrong place"11:48
cjwatsonI'm not sure that translates accurately into "nobody complained"11:48
apwcjwatson, i think that is very likely11:48
lifeless2.6.15-23-386,11:48
apwwe are demand dirven though as we have nothing to test on11:48
lifelessmodel name      : Pentium 75 - 20011:49
apwit took me half a day to get this machien with a floppy to even work11:49
cjwatsonno, I understand that, I just dislike the "nobody managed to file bug reports, let's bin it" approach11:49
lifelesscjwatson: me too :P11:49
lifelessKeybuk: its running 8.0411:49
apwcjwatson, indeed, but if we lost something and didn't notice and need to maek a kernel change to fix it but cannot as we have none, its very hard to do anything about11:50
lifelessapw: do you know, how I can tell if this is acpi or not ?11:50
apwthere are acpi_available and apm_available iirc11:50
apwwhich exit 0 if there is some of that11:50
apwapw@dm$ acpi_available ; echo $?11:51
apw011:51
apwapw@dm$ apm_available ; echo $?11:51
apw111:51
lifelessapw: neither command exists11:51
apwgurgle11:51
lifeless:)11:51
Keybuk[ -d /proc/acpi ]11:52
Keybuk:p11:52
apwcat /proc/apm11:52
apwls -l /proc/acpi11:52
lifeless$ [ -d /proc/acpi ]11:52
lifelessrobertc@lifelessgwy:~$ echo $?11:52
lifeless111:52
apwso no acpi11:52
apwapm ?11:52
lifelesscat /proc/apm11:52
lifelesscat: /proc/apm: No such file or directory11:52
Keybukapw: apm only covers power, not device discovery/enablement11:52
apwright, but having apm tells us you don't have acpi11:53
lifelessKeybuk: so, I can't until next weekend, but then I can, get a newer kernel and see if its bootable and report back11:53
lifelessKeybuk: and if it boots answer any other queries while I upgrade from 8.0411:53
apwlifeless, sounds good to me11:53
Keybukapw: not necessarily11:53
KeybukI have a laptop that has both11:53
apwdouble gurble11:54
* apw climbs back into his nice shiney acpi only world, la la la, can't hear you11:54
lifelessKeybuk: will that help you?11:55
Keybukcjwatson: I'd claim the opposite11:56
Keybukthost people who are trying to run Linux on older machines are more likely to be tech savvy and trying to keep the older machines running, so more likely to complain in the right place11:56
Keybuk(or at least complain somewhere audible)11:56
Keybuklifeless: I'd be interested to see the results11:56
lifelessKeybuk:11:56
lifelessbah.11:57
apwKeybuk, a resonable conjecture11:57
lifelessok, will fiddle around to get a newer kernel and a monitor and give it a spin11:57
lifelessif it doesn't work, I may bite the bullet and replace it, and shit it to you. Its about 1/8th of a cubic metre.11:57
cjwatsonKeybuk: neither of our opinions is supported by data, I suppose :-)11:58
Keybuklifeless: hey, I didn't say I was going to _fix_ it :p11:58
apwKeybuk, that reminds me, if you are shipping an empty modprobe.d how are we handling blacklisting11:58
KeybukI'm somewhat concious that the amount of work needed to put code into the kernel to deal with the non-ACPI cases is so large and error-prone that it may simply not be worth it11:58
cjwatsonI keep reading stuff at the moment about how people are more likely to be recycling old machines in the depths of a recession, rather than buying new ones11:58
Keybukapw: they're probably going in /lib/modules/$(uname -r)/modules.conf11:59
apwso we will be cslurping them up basically11:59
Keybukcjwatson: "old" in this case does mean "1999 and before though", no?11:59
apwyeah i was amazed to find there was a floppy on a machine that worked any more11:59
apwfor osmeone to complain about it11:59
cjwatsonI've learned never to make that kind of generalisation :-)11:59
cjwatsonI was trying to get Ubuntu to work last night with a monitor that doesn't support DDC12:00
Keybukheh12:00
cjwatsonI'd say the monitor is at least 15 years old, at a guess12:00
apwKeybuk, so are you handling the integration with the kernel or does someone here need to start thinking about it?12:00
Keybuk"Writing X Mode lines for dummies"12:00
Keybukcjwatson: I think that's the point though - we don't actually support those monitors either right now, do we?12:00
TheMusocjwatson: Interesting you should say that, some cheap KVM switches out there don't support DDC either.12:00
cjwatsonI gave up, but I'd rather argue that I shouldn't have to12:00
cjwatsonbecause if I weren't an Ubuntu developer, I'd be installing some other operating system on it right now12:00
Keybukapw: I'll handle it ;)12:00
apwKeybuk, ok cool.  /me scuttles off to his tunnels12:01
Keybukcjwatson: I'd be intrigued to know whether the current version of any other operating system works either <g>12:01
KeybukI'd wager lots of money that neither Vista nor Mac OS X would support that monitor12:01
cjwatsonI could, for example, reinstall the Windows 2000 that was on there when I got it12:01
KeybukWin 2000 is EOL12:01
Keybukyou could install warty on it12:01
* apw wonders if hardy would work12:01
cjwatsonso what? (from the point of view of somebody who isn't an OS person)12:01
cjwatsonworks > current12:01
KeybukI guess I don't see "the current version doesn't work -> use an older one" as a problem12:02
* sistpoty|work wonders wether qemu than does work OOTB with a daily, since iirc it doesn't support ddc either12:02
apwwe probabally don't recommend it loudly enough12:02
Keybuksince that's generally the solution12:02
cjwatsonthe operating system that was already there will get a free pass in a lot of people's minds, whether it's old or not12:02
apw"if you have pre-2000 h/w" you might want to start with this version12:02
lifelessKeybuk: security fixes12:02
cjwatsonsome *other* old operating system will not get that same free pass12:02
Keybuklifeless: Win 2000 doesn't have security fixes from MS anymore12:02
lifelessKeybuk: I don't use win2000, so thats irrelevant :)12:03
Keybukcjwatson: and this is the end of the world?12:03
cjwatsonKeybuk: oh FFS12:03
cjwatson"this is a bug, I'm trying to get you to acknowledge that" != "end of the world"12:03
cjwatsonstop turning everything into the worst possible extreme12:03
KeybukI'm not saying it's not a bug12:03
KeybukI'm saying its a bug I'm inclined to Won't Fix12:03
cjwatsonor even "we should fix this" != "end of the world"12:03
* TheMuso has jaunty running on a dual celeron 466 box, and the board was from 1999. Works fine, appart from the broken mga driver for the matrox card in there last I checked. :)12:04
apwTheMuso, woh thats old!12:04
apwi should introduce it to my thinkpad 570e, they are the same age12:04
cjwatsonsistpoty|work: qemu support some modeline that X tries by default; this monitor apparently doesn't12:05
Keybukcjwatson: maybe that's where we're disagreeing here12:05
TheMusoapw: heh, its an abit BP6. I keep it around because it has ISA slots, which allows me to muck around with some ISA hardware speech synthesizers i have.12:05
apwcjwatson, the right thing to do is fedex the thing to bryce ...12:05
Keybukcjwatson: I'm not disagreeing that things don't work (they clearly don't)12:05
cjwatsonKeybuk: I have consistently been given direction that we are to support the long tail of hardware in Ubuntu12:05
KeybukI'm claiming that the effort to fix those things is probably too large compared to the benefit12:05
sistpoty|workcjwatson: ah, makes sense12:05
Keybukcjwatson: and at some point, the tail has to end12:06
Keybukwe can't support all hardware ever created12:06
Keybukmaybe we'd like to12:06
Keybukbut that's an awful lot of work12:06
cjwatsonI acknowledge that; I think you perhaps overestimate how frequently people upgrade hardware, though12:06
evandI don't pretend to know what I'm talking about here, but just an idea, what if we only rip out support for non-ACPI machines in the desktop kernel.  That would allow people who really cared about getting Ubuntu running on their P2 to still do so through the server CD.12:06
cjwatsonI think we're cutting off a bigger tail than you think12:06
Keybukand you'd get into the strange situation that the exotic hardware you're supporting only exists in your labs, because you purchased all available models of that hardware to test it on ;)12:06
cjwatsonevand: this is about the work required to reintroduce support dropped due to a bug12:07
evandoh, my mistake then12:07
cjwatsonevand: it's not about the space required to keep support12:07
apwi would say peoples first experience with ubuntu is as often as not installed on their 'old' laptop after an upgrade12:07
Keybukit's not necessarily "because of a bug"12:07
cjwatsonKeybuk: that's a gross exaggeration, and unhelpful in this case12:07
cjwatsonpre-ACPI hardware is not as rare as all that12:07
Keybukit's not a "oh, look, a one line patch"12:07
Keybukor a "this little patch broke non-ACPI machines"12:07
cjwatson"due to a bug"> OK, that was an excessive abbreviation12:08
Keybukit's "the kernel really assumes ACPI for a lot of things now, and there's no equivalent non-ACPI code"12:08
apwcjwatson, i wonder if machines booted acpi=off count12:08
Keybuk(and there probably wasn't any non-ACPI code before - since the kernel didn't bother with such niceties in those days)12:08
Keybukie. we never had non-ACPI hotplug12:08
apwit cirtainly assumed synchronised userspace12:09
apwwe did have userspace hotplug though12:09
apwand its our removal if that which broke things12:09
Keybukwe used to run isapnp and write config files, and then use those12:09
cjwatsonso the appropriate way to do this would be to post somewhere very public saying "we've discovered that we don't appear to support pre-ACPI machines any more; do you care? if so, please write to us"12:09
Keybukapw: nah, that was just how we _found_ this underlying problem12:09
apwthat is hotplug in userspace12:10
cjwatsonand assess the responses objectively12:10
Keybukcjwatson: which was my suggestion12:10
* TheMuso still has an ISApnp sound card, that actually still works.12:10
=== azeem_ is now known as azeem
KeybukTheMuso: it should work on an ACPI-machine12:10
cjwatsonKeybuk: I didn't see that suggestion - where?12:10
TheMusoKeybuk: It does.12:10
cjwatsonyou said12:10
cjwatson11:46 <Keybuk> that maybe we should just note that we don't support them anymore :p12:10
cjwatsonwhich sounded like a release note "hi guys, unsupported, you lose" :-)12:10
TheMusoanyway, bed time for me.12:10
Keybukcjwatson: somewhere, this conversation has crossed three channels and several bugs now ;)12:11
cjwatsonheh12:11
KeybukI started by asking lifeless ;P12:11
Keybukand I did start here by asking <g>12:11
Keybuk(generally)12:11
cjwatsonright, and several people said that they had machines with issues12:11
cjwatsondidn't they?12:11
Keybukjust lifeless so far12:11
cjwatsonand apw12:12
* apw has a machine whcih is affected too for the record12:12
apwthough i prefer not to admit its existance12:12
apwKeybuk, i wonder if some combination of my fixes are appropriate12:13
Keybukwhoo12:13
apwto generate pnp event IFF acpi is not working12:13
KeybukI get a full-on kernel panic trying to boot with acpi=off12:13
mneptokjames_w: meep.12:13
james_wmneptok: ahoy12:13
apwKeybuk, what from?12:13
Keybukapw: the problem there is you have to then patch all the pnp modules to also support the other subsystem12:13
Keybukapw: somewhere insire parport_pc_init12:13
Keybukand pnp_register_driver12:14
mneptokjames_w: are you responsible for kurts_self_image.deb? if so, you're doing a terrible job as package maintainer. allow me to send you a 14 page e-mail with details ....12:14
StevenKMy firewall is too early for the cutoff too.12:14
apwKeybuk, well yes we'd need some new aliases12:14
apwbut other than that probabally not12:14
Keybukapw: dunno, am reading through the code of many of them12:14
james_wmneptok: oh, I thought we were using a different image of you now? mnempolo.deb?12:15
* StevenK twitches12:15
StevenKjames_w: I did *not* need a reminder of that.12:15
mneptok:D12:16
mneptokStevenK: install Kees' usplash ... you know you want to ....12:16
StevenKmneptok: I'd rather burn out my optic nerves12:17
mneptoklike i said. install the PPA.12:17
Mithrandira mneptok!12:18
apwKeybuk, ok at least one of my machines boots acpi=off12:18
mneptokMithrandir! Mithrandir!12:19
* mneptok feels like a juvenile hobbit about to see fireworks12:19
apwKeybuk, on the module aliases for isa those actually are already in place in a lot fo cases12:19
apwas i got some broken modules loaded as a result12:19
Mithrandirmneptok: how's the new job?12:20
apwwhen i started generating pnp:dPNP700 style ones12:20
mneptokMithrandir: delightfully busy and stressful :)12:23
Keybukapw: err, well, there's about 6 modules with them12:24
Keybukand those modules don't look like they've been touched in years12:24
Keybukso those aliases are probably only in there by accident <g>12:24
Keybukhmm12:25
Keybukinterestingly there is a struct pnp_device_id12:25
apwyep, there is a full infrastructre to get them to modules.modaliases12:25
apwbut nothing generates them right now, other than my patch of course12:25
Keybukwhich is odd12:25
Keybuklook at parport_pc for an example12:26
Keybukit only has the pnp_device_id DEVICE_TABLE12:26
Keybukbut it ends up with both acpi: and pnp: aliases12:26
apwthe device is also listed in the acpi tables independantly12:26
Keybukis it?12:26
apwfor it to have an acpi thing i would say yes12:27
KeybukI can't find _that_ code12:27
apwi think udev fakes the events12:27
Keybuknope12:28
apwthere are modalias files in /sys for somethings12:28
apwi thought udev had to scan there for things that appeared before it did12:28
apwand i had been assuming it did so using the modalias files in /sys12:28
Keybukudev doesn't touch those12:29
apwhow does it find old things, things that preceeded its first invokation12:29
Keybukit just writes "add" to all the uevent files12:30
Keybukwhich makes the kernel resend the netlink messages12:30
apwwhich is probabally why the kernel keeps them in modalias12:30
Keybukthat's mostly just for debugging12:31
apwKeybuk, if you have a machine with a floppy with acpi could you let me have your udev log12:34
apwwant to see how it would interact should i introduce a modalias on the pnp events too12:35
Keybukhttp://people.ubuntu.com/~scott/udev12:35
Keybukyou'd get two different MODALIAS strings from two different devices on two different subsystems12:35
Keybukwhich would be bad12:35
Keybuk(both resolving to the same module)12:35
apwwhy would that be bad12:35
apwyou would get two events anyhow12:35
apwone with each12:35
Keybukthe pnp devices aren't "under" the acpi ones12:36
Keybuklikewise the acpi ones aren't "under" the pnp ones12:36
Keybukit'd be interesting to find out why the kernel did it this way in the first place12:37
apwactually you don't get the pnp events at all on yours12:37
Keybukooooh12:37
KeybukI just had a thought12:37
apwthe events in my log with the APW: on them i didn't add those events they were there12:37
Keybukcould we make the MODALIAS for the pnp thing only show up if pnpacpi isn't working?12:37
apwyeah we could indeed though i suspect you arn't getting them anyhow already12:38
Keybukshould be12:38
KeybukI have /devices/pnp0 and stuff12:38
apwwhats in your /proc/bus/pnp12:38
KeybukENOENT12:39
Keybukquest scott% ls /sys/bus/pnp/devices12:39
Keybuk00:00@  00:02@  00:04@  00:06@  00:08@  00:0a@12:39
Keybuk00:01@  00:03@  00:05@  00:07@  00:09@  00:0b@12:39
Keybukfor example12:39
Keybukyou have12:39
KeybukUEVENT[1236073926.075992] add      /devices/pnp0/00:11 (pnp)12:39
KeybukUDEV_LOG=312:39
KeybukACTION=add12:39
KeybukDEVPATH=/devices/pnp0/00:1112:39
KeybukSUBSYSTEM=pnp12:39
KeybukMODALIAS=APW:dPNP070012:39
KeybukSEQNUM=97912:39
apwright ... i had those events even before i added APW: to them12:39
apwjust without alias12:39
KeybukI have them too12:40
KeybukUEVENT[1235981705.266498] add      /devices/pnp0/00:09 (pnp)12:40
KeybukACTION=add12:40
KeybukDEVPATH=/devices/pnp0/00:0912:40
KeybukSUBSYSTEM=pnp12:40
KeybukSEQNUM=212912:40
Keybuk(that id file contains PNP0401)12:40
apwso i don't see why it would be an issue to have the pnp: style aliases on there12:40
Keybukthe trouble is, I also have12:40
apw(i missed them somehow)12:40
KeybukUEVENT[1235981705.260729] add      /devices/LNXSYSTM:00/device:00/PNP0A08:00/PNP0401:00 (acpi)12:40
KeybukACTION=add12:40
KeybukDEVPATH=/devices/LNXSYSTM:00/device:00/PNP0A08:00/PNP0401:0012:40
KeybukSUBSYSTEM=acpi12:40
KeybukMODALIAS=acpi:PNP0401:12:40
KeybukSEQNUM=196712:40
apwbut thats not an issue12:40
apwloading the modules twice isn't an issue12:40
Keybukthe issue is that why does pnpacpi exist at all?12:40
KeybukI'm trying to understand why the kernel doesn't already do it the way you describe12:41
apwwe simply don't have the aliases on there as far as i can see12:41
Keybukright, but when upstream fixed that problem12:41
Keybukthey did it completely differently12:41
apwand as you have found we have some destinations12:41
Keybukit would have been far simpler to do your patch12:41
apwwell if our userspace was anything to go by12:41
Keybukbut instead they wrote this entire new pnpacpi subsystem12:41
apwthey were allowing udev to generate them12:42
Keybukand, most notably12:42
apw_ahh_12:42
Keybukyour patch makes pnpacpi irrelevant12:42
Keybukat least, apparently so12:42
apwthe one problem i found is there is no way to generate mroe than one12:42
apwevent per device12:42
Keybuksure there is, just attach multiple kobjects to it12:42
apwnote the proggy in the 80-* patch does a while <> { }12:42
apwwell not in the interface as it is12:42
Keybukie. why isn't it /devices/pnp0/00:09/PNP040112:42
Keybukapw: note the Author in the patch you're reading <g>12:43
apwright, so there is no way at the bus level which is where they are generated12:43
apwto produce all the aliases there12:43
Keybukso how does pnpacpi do it?12:43
apwit cheats right12:43
apwand produces acpi:id1:id2:id312:43
apwnot sure if that is right, if it is that is why we have to load12:44
apwall matching modules not just the first12:44
apwMODALIAS=acpi:PNP0A08:PNP0A03:12:44
apwthat is carrying two things12:44
apwthough i have yet to understand why we would not just make the format12:44
apwMODALIAS=acpi:PNP1 acpi:PNP212:45
Keybukyou can't have spaces in a modalias12:45
apwand do the right thing with those in udev12:45
apwthere isn't a space, there are two aliases12:45
apwand modprobe would do the right thing12:45
apwit might need an option i can't remember but nothing insoluable12:45
apwbut thats getting hugely far ahead of ourselves12:46
=== AstralJa1a is now known as AstralJava
Keybuksimplicity of interface, I guess12:46
KeybukMODALIAS being a single string you pass to modprobe has its elegance12:46
apwbut i think the line in udev would be the same12:46
Keybukand it actually lets you do things like have modules that say12:47
apwwe we do pass it in such a way if it has spaces its N arguments12:47
Keybuk  acpi*:PNP1:PNP2:*12:47
Keybukie. a module that handles devices that export two pnp ids12:47
apwyeah, though those two id's repreent separat functionalities12:47
apwso that would be strange12:47
apwand you could just as well say12:47
apwMODALIAS(acpi*:PNP1*)12:47
apwMODALIAS(acpi*:PNP2*)12:48
apwbut hey12:48
Keybukthen you'd be loaded for both12:48
Keybuknot for once device that implements both12:48
Keybuk*shrug*12:48
apwbut i can't really see any reason not to produce pnp: style aliases at the PNP level as well12:48
KeybukI figured out where the pnp:/acpi: aliases come from12:48
apwother than as i have discovered new modules _do_ get loaded, and some of them PANIC12:48
Keybukhmm12:49
Keybukif a device is currently using acpi12:49
Keybukyou only want it loaded on the acpi: device12:49
Keybuksince you need the acpi subsystem to have been init'd12:49
Keybukif a device is using pnp: you can load it on either the pnp: device or the acpi: device12:49
Keybukthat's the problem12:49
apwi assume the acpi: ones have more information in them, such as the device id's12:49
Keybukif you generate pnp: modaliases for all pnp devices, including acpi: ones12:50
Keybukudev will load the modules12:50
Keybukpossibly before the pnpacpi stuff has even been initialised12:50
Keybukor while it's being initialised12:50
apwif the device is found using acpi it will have to necessarily have ACPI up by then12:50
Keybukahh12:50
Keybukno you're misordering things12:50
Keybukthese devices can be found by either pnp *or* acpi12:50
Keybukthey are in fact found by *both*12:50
Keybukhowever some drivers now assume acpi12:50
Keybukwhile older drivers assume pnp12:51
apwright but actually they are only found way after that12:51
Keybukfloppy happens to assume pnp12:51
apwcause udev isn't there to see the event order12:51
Keybukbattery for example assumes acpi12:51
Keybuknot necessarily12:51
apwfind me a way you can get to userspace before acpi is complete12:51
Keybukconfig PNPACPI=m12:52
Keybukmeh, that's a bool :p12:52
apwi think it would have to be12:53
apwi just don't see why it is bad that way12:53
apwudev has been doing it that way before12:53
apweven if we did fake up the rules12:53
KeybukI'm trying to forsee why your patch will be rejected upstream12:53
apwcompletly reasonable12:53
Keybukand right now, it will be rejected because "we already have these modaliases, they are generated by pnpacpi"12:53
apwi would be suggesting ripping out all the aliases as part of it12:53
apw(the pnp: ones12:54
Keybukwhy the pnp: ones?12:54
geserdoko (or any other main sponsor): have you time to review the debdiff for bug #336871 and sponsor it? (python 2.6 transition)12:54
apwor generating them in a differennt form12:54
ubottuLaunchpad bug 336871 in urlgrabber "Deprecated modules with python 2.6" [Undecided,New] https://launchpad.net/bugs/33687112:54
apwas they ahve risk, as i discovered, if you have the events12:54
apwthen the modules which are marked get loaded, and that blows up my machine12:54
apwas the modules are crap12:54
dokogeser: yes, will do12:54
Keybukisn't it just that the modules are assuming ACPI?12:54
apwpossibly dunno yet12:55
* apw thinks12:55
Keybukthe logic line I'm thinking is12:55
apwperhaps there does need to be something thats used !acpi12:56
Keybuk - we need MODALIASes from the pnp subsystem when pnpacpi doesn't generate them12:56
Keybuk - if we generate them from pnp subsystem, we don't need the ones pnpacpi generates12:56
Keybuk - pnpacpi exists for a reason12:56
Keybuk - so we're missing something :p12:56
apwi am suspicious that your 3 there is false and thats your 412:57
apwi will go spend half an hour trying to find out its origins12:57
KeybukI wonder whether this all fundamentally comes down to pnp configuration12:57
Keybukie. loading device configuration12:58
apwwell first thing, acpipnp is very very old12:58
Keybuk~200412:58
Keybukthat's about the midpoint of driver-core development12:59
Keybuk(assuming you mean the code)12:59
ScottKSorry to come late to the party (just finished the backscroll).  I've got a server running that's pre-2000 and so pre-ACPI.  It's running Intrepid apparently successfully.13:01
ScottKOr is it?13:01
* ScottK checks13:01
Keybuksome things seem to work, some don't13:02
apwKeybuk, where did you find the acpi:'s being gnerated13:03
Keybukapw: which bit?13:03
Keybukthere's two types of "generated"13:03
ScottKStill hardy13:03
KeybukMODALIAS for a hardware device13:03
apwthe kernel generating the uevents13:03
Keybukor alias line inside a module13:03
apwthe MODALIAS=foo part13:04
Keybukapw: drivers/acpi/scan.c13:04
Keybukwith the actual device ids created by drivers/pnp/pnpacpi/core.c13:04
* Keybuk looks at pnpbios with interest13:04
* ScottK considers continuing to forget to upgrade that one.13:05
apwthere is some mad shit still in there13:05
Keybukthis is quite, quite, evil13:05
Keybukbtw, can I see your patch?13:06
ScottKSo if you need something tested, I can do it on that box.  It's not actually busy with any real tasks currently.13:07
apwKeybuk, how does udev decide on ordering of its requests when catching up in the beginning?13:08
apwas i note your acpi: events come out before you pnp events13:08
Keybukapw: tree order13:09
Keybuk /a/b comes after /a13:09
Keybukwhich was the original problem I was alluding to13:09
apwhmmmm ...13:09
apwso they ahppen to be in the right order13:09
Keybukthe /devices/pnp0/* and /devices/LNXSYSTM:00/device:00/PNP*:00/PNP*:00 are unrelated in the tree13:10
Keybukso there are no ordering guarantees13:10
apwcreation order to some degree i guess13:10
Keybukno13:10
Keybukjust sunspot order13:10
apwi bet they will be in creation order in each directory, but no guarenttes over time13:11
KeybukI'm guessing you added a .uevent = pnp_bus_uevent line to pnp_bus_type13:11
Keybukand in that function added MODALIAS to the uevent?13:11
apwyes exactly, nothing magic13:11
apwi suspect that is an error for this13:12
KeybukI'm trying to work out what you need to put in the MODALIAS13:12
apwi suspect there must be an enumerator which is not the acpipnp which i shoul dhave hooked13:12
Keybukbased on pnp_bus_match13:12
Keybukyeah, I suspect that's probably more technically correct13:13
Keybukfigure out what's adding the pnp devices in the first place13:13
apwin the absence of acpi to do it13:13
apwi think i have some isa busses on there which are not there else13:13
Keybukhow do you mean?13:13
apwi'll let you know waht i find13:13
Keybukyou have things in /sys/bus/pnp/devices that don't show up as ACPI platform devices?13:13
Keybukbtw. interestingly, pnpacpi sets pnp_platform_devices = 113:14
apwthis is a non acpi machine13:14
apwand i have stuff in pnp13:14
apwmachine is booting13:14
apwmachine is still booting13:14
apwKeybuk, can't you speed this boot up or something :-P13:15
Keybuksilly question13:15
Keybukwhat calls pnp_add_device () if it's not pnpacpi?13:15
apwpnpbios13:15
apwand pnpacpi13:15
apwi suspect i need to hook the latter13:16
Keybukpnpbios isn't enabled in our kernels is it? :P13:16
Keybukat least, it doesn't show up in /boot/config-*13:16
apwits enabled13:16
apwdebian/config/i386/config:CONFIG_PNPBIOS=y13:16
Keybuksorry13:16
Keybuksomeone must have renamed the option13:16
Keybukit shows up in jaunty13:16
Keybukjust not intrepid13:16
apwahh, there is also an isapnp directory here13:17
Keybukdmesg | grep PnPBIOS13:17
Keybuk?13:17
* apw waits13:18
Keybuk(I had to look at /var/log/dmesg because my dmesg is full of TKIP)13:18
apwPnPBIOS: Scanning ...13:18
apwand finding 20 nodes13:18
Keybukaha!13:18
KeybukI see13:18
apwalso isapnp speaks13:18
KeybukPnPBIOS: Disabled by ACPI PNP13:18
apwso thats my likely target13:18
Keybukyeah13:18
Keybukit sounds like we just need PnPBIOS to add MODALIASES in almost the same form as acpi: but different13:19
apwpnpbios:dNNNNN13:19
apwor something13:19
Keybukthen we can have file2modalias spit out pnp: acpi: and pnpbios: MODALIASes for anything with a pnp_device_id13:19
KeybukI'd go with pnpbios:PNPXXXX:PNPYYYY: just like acpi13:19
apwyeah13:19
Keybuksince then you can probably just steal the code ;)13:19
Keybukanything using acpi_device_id will only work if you have ACPI13:19
apwwho me i would never, 10yyp13:19
Keybukbut those drivers probably only work if you have ACPI anyway13:20
apwagreed13:20
Keybukanything using pnpbios directly wouldn't work if you have ACPI13:20
Keybukbut I don't think there are such a thing13:20
Keybukand ancient legacy drivers (like floppy!) would work on ACPI or PnPBIOS13:20
Keybukand I'll rework my floppy patch to use pnp_device_id13:20
apwKeybuk, i would leave your patch alone13:21
apwas that is much more suitable for intrepid13:21
Keybukapw: actually, it's arguably wrong if I use acpi_device_id13:21
apwmaybe jaunty too depending on upstream13:21
apwperhaps os13:21
Keybukpnp_device_id would be correct13:21
Keybukmatch the other drivers nearby13:21
apwok then13:21
Keybukand make it work on intrepid and jaunty13:21
apwack13:21
Keybuk(pnp_device_id exists already - this would not depend on your patch)13:21
Keybukand the kernel already turns pnp_device_id lines into acpi*:PNPxxxx: matches for modules.alias13:22
apwoh fine then13:22
Keybuklots of things will still be broken for non-ACPI machines13:22
Keybukbut their floppy drive will work!13:22
apwheh13:22
Keybukand probably their soundblaster too ;)13:23
ScottKdoko: Would you please look at the xdelta3 package.  I've rebuilt it for Python 2.6 with layout=deb and all and while it provides python2.5/2.6-xdelta3 is manages still to depend on python2.4, python2.5 which is clearly wrong and I have no idea why.13:36
dokoScottK: your debdiff?13:37
ScottKdoko: It's in Jaunty already.  I didn't notice before I uploaded.13:38
ScottKdoko: Here's what I changed http://pastebin.com/fdbb8fc13:40
dokoScottK: a .dirs file in debian creates a python2.4 dir, and xdelta3-regtest.py uses 2.5 as versioned interpreter (dh_examples is called *after* fixing the interpreter names)13:43
geserScottK: look at the shebang line of the examples in the usr/share/doc/python-xdelta3/examples: one uses python2.4 and the other python2.5 as interpreter13:44
ScottKdoko: I missed that.  Thanks.13:44
ScottKgeser: Thanks.13:45
nxvlrobbiew: hi robbie, just looking at the Karmic Release Schedule, is this FeautreDefinitionFreeze a new thing?13:59
nxvlcjwatson: ^^14:01
Hobbseemvo_: ping?14:04
pittinxvl: it was formerly called "Specifications must be finished" or so14:04
* Keybuk gnarghs @ fuse14:06
Keybukevil install rule14:06
Keybukslangasek: around?14:11
nxvlpitti: yeah, i understand what i means, i'm just curious since i see a lot of new Freezes that weren't there before14:15
nxvlpitti: well, a lot of "Notes" have been moved to Freezes actually14:16
Keybukpitti: you've made some uncommited changes to acpi-support14:16
pittiKeybuk: indeed, we should get those in at some point14:17
vbgunzpart 2. suspending to ram seems to work flawlessly. 2 issues though. 1. I have no way to associate the power button with "suspend to ram", powerdevil doesn't work here. 2. selecting "suspend to ram" in kickoff successfully works. pressing the power button successfully wakes up *but* the screen is not locked :/ . what would be nice is, press the power button, sleep, press it again, wake up with the screen locked14:18
vbgunzI just installed kpowersave and configured it to do this and it actually works *but* I already have powerdevil and want to keep it. what can I do to help powerdevil actually work here?14:18
cjwatsonnxvl: the only additions, I believe, are feature definition freeze and new packages freeze (I'm not sure how well the latter will work; I queried that)14:18
Keybukpitti: do you know why you were asked not to just upload?14:19
cjwatsonnxvl: this cycle some of the specs landed very late, and I do think that's a problem we should address14:19
nxvlcjwatson: yup, i think the same, just cheking :D14:23
pittiKeybuk: can't remember any more, I'm afraid14:24
=== ssweeny_ is now known as ssweeny
RainCTcjwatson: that's funny.. I've just re-opened a bug of yours someone closed and pointed them to the rant on planet.ubuntu.com and now I see that's actually a post from you :)15:03
* cjwatson grins15:03
cjwatsonRainCT: BTW I don't want it to be an argument from authority ("Ubuntu developer told me to do it this way, so I must obey")15:06
seb128worth noticing that bug triaging practice changes between maintainers and teams15:07
cjwatsonsome practices are just wrong wherever they sit15:07
cjwatsonclosing bugs without having taken the time to understand them and determine where they best live is one of those practices15:08
seb128well, we do agressive bug closing on desktop bugs which I'm pretty sure other maintainers would disagree with, but we just have enough good quality bugs to be busy until end of time15:09
seb128so there is no need to keep tons of low quality bugs which might turn to something useful after hard work15:09
cjwatsonwas my bug low-quality?15:09
cjwatsonI'm genuinely asking - if it wasn't, I don't see how this point is remotely relevant15:09
cjwatsonthe problem is that some of the people closing bugs are *unable to tell the difference*15:10
cjwatsonand that's wrong no matter whether it's desktop or anywhere else15:10
seb128well, we do close crash bugs which are not sent using apport for example15:11
cjwatsonnot relevant to this bug15:11
seb128which I'm pretty sure you will disagree with15:11
seb128oh, I don't know of what specific bug you are speaking about15:11
cjwatsonbug 32848615:11
seb128I just read some comments about bug triagers work15:11
ubottuLaunchpad bug 328486 in apturl "automatically add keys when whitelisted for apturl" [Wishlist,New] https://launchpad.net/bugs/32848615:11
seb128I don't discuss specifically this one15:12
RainCTcjwatson: Heh. I think yours was fine :). I'm not familiar with the recent changes so mvo_ will have to tell, though :P15:12
cjwatsonif desktop practices could effectively be confined to desktop work, I would have much less issue with it15:12
cjwatsonpart of the problem is that people see desktop practices and think they are appropriate everywhere15:12
* RainCT writes down "do not file crash reports for desktop bugs" *g*15:12
cjwatsonseb128: crash / not apport> *shrug* I don't feel strongly about that, actually15:13
cjwatsonseb128: I wouldn't if it were mine, but I can understand why you do that and it doesn't seem totally unreasonable15:13
seb128we also close duplicate without sending those probably as duplicate some time which tend to make some people unhappy too15:13
seb128anyway dealing with lot of bugs is not easy15:13
seb128some time -> sometimes15:14
cjwatsonagain, I don't feel strongly about that; I think it would be *better* if you could duplicate them properly but I understand that it is a lot of work15:14
RainCTseb128: Yeah I understand that. I don't agree, but sometimes I'm tempted to do the same ^^15:14
cjwatsonthe problems I have with bug triage are not ones that I think developers would generally accept; they are usually closely related to bug triagers working without particular regard to what developers need15:14
cjwatsonif you work closely with bug triagers, then almost certainly they're generally doing things that fit well with your practices15:15
cjwatsonmy problems are when bug triagers do things that don't fit well with my practices as a developer, and don't put any effort into checking15:15
cjwatsonand I don't really see that as likely to be controversial among developers?15:16
cjwatsonthe point of bug triage is to improve the state of bugs so that it's easier for developers to fix them efficiently15:16
cjwatsonif it's making developers' lives more difficult, then it is failing15:16
cjwatsonand the point of bugs, of course, is to improve the software15:17
* broonie notes that he doesn't often report bugs to Ubuntu largely because of the triage.15:17
cjwatsonyes, I've heard the same sentiment from many people15:17
seb128right15:17
cjwatsonI am deeply disturbed by this state of affairs and want to get it fixed if at all possible15:18
seb128I just know that I tend to read argument on #ubuntu-bugs between agressive closing or letting lot of bugs which lack details open15:18
broonieIt's particularly disturbing in that there are humans implementing what looks entirely like a mindless automated system.15:18
cjwatsonseb128: I don't mind closing bugs that genuinely don't have enough information to fix them, and where the reporter isn't responsive15:19
seb128ie we often get people who argue that we shouldn't close duplicate without marking them as proper duplicate even if we know they are dups for sure15:19
cjwatsonseb128: I *do* mind us closing bugs when the triager simply doesn't understand the information that was given15:19
cjwatsonseb128: or doesn't understand what information the developer needs, and so asks for a bunch of completely useless stuff15:19
seb128right15:19
=== ScottK2 is now known as ScottK-desktop
seb128I agree with you on that15:20
cjwatsonseb128: this is a situation I see regularly, and is essentially what I'm complaining about. I'm not complaining about the desktop team's practices15:20
cjwatsonand I don't think that, to date, the Ubuntu development team has really stood up and said "look, this isn't what we want, and it isn't what we need" in an organised way15:20
cjwatsonthat's sort of what I was trying to kick-start15:21
seb128cjwatson: right, I didn't take your post as something again desktop bug triaging, my comment was rather a follow up to some other comments I read on IRC since yesterday15:21
seb128ideally people who don't understand what they are doing should not be in bugsquad and allow to close bug to start15:21
cjwatson(BTW, I ran it past heno before posting, since I didn't want to be taken as having a go at the people who *are* doing a good job of Ubuntu QA)15:22
seb128allow -> allowed15:22
Keybukquest linux-jaunty% git rebase --onto gnargh c09024e129f515dd87e0abafc7ff3a4791216838 topic/modprobe15:38
Keybuk*sometimes*15:38
Keybukjust15:38
Keybuk*sometimes*15:38
KeybukI like git15:38
pittiwohoo :)15:55
pittiKeybuk: that extracts a patch and applies it to a different branch/file?15:56
Keybukit reapplies the branch of patches since the revid I gave to my current branch15:56
pittiseb128: ah, you already saw the weird amd64 retracer error and removed lock?16:04
seb128pitti: yes16:04
pittithanks16:05
seb128pitti: it seemed to be a launchpad timeout or something16:05
seb128np16:05
pittiseb128: as I said, might be that weird python-apt thing again16:05
seb128right16:05
pittibut well, at least it's so much better than a couple of weeks ago16:05
seb128it doesn't seem to be happened as often though16:05
pittiit wouldn't finish two bugs without a timeout back then16:05
seb128right16:05
pittiseb128: after the UIF rush I'll see to switch to launchpadlib, get backports for hardy/intrepid, and update the retracers to use it16:06
seb128cool16:06
pittibut that sounds like an entire day of work16:06
pittitseliot: there's a sponsoring bug 95444; do you think you can have a look at it in the next nvidia-180 upload?16:24
ubottuLaunchpad bug 95444 in nvidia-graphics-drivers-180 "No Screen Backlight Control; Notebooks (Vaio, Macbook, HP/Compaq, Samsung, Zepto et al.) with Nvidia Geforce8/Geforce9/Quadro series graphics" [Undecided,New] https://launchpad.net/bugs/9544416:25
tseliotpitti: sure16:25
=== mvo_ is now known as mvo
=== beuno_ is now known as beuno
=== ivoks_ is now known as ivoks
jdstrandseb128: does evolution-mapi have an FFe? was it discussed somewhere off-list?16:56
seb128jdstrand: expection granted, it's part of GNOME 2.26 official and I can grant universe desktop exceptions16:56
jdstrandseb128: ok cool16:57
jdstrandseb128: just out of curiosity, I guess JRiddell can do the same for KDE?16:59
* pitti args at Rosetta spam17:00
seb128jdstrand: not sure, MOTU sent an email with the delegation lists for teams, I would expect so yes17:01
jdstrandseb128: ok, I'll check that out. thanks :)17:01
seb128you're welcome17:01
seb128jdstrand: "Feature Freeze this Thursday" on u-d-a17:02
seb128jdstrand: this email has the detailed list17:02
jdstrandseb128: excellent17:02
Riddelljdstrand: yes I can17:03
asacslangasek: something funny: http://mail.gnome.org/archives/networkmanager-list/2009-March/msg00010.html17:04
jdstrandcool, got it17:04
asacslangasek: seems they try to achieve quite the oposite17:05
asacwhich might mean we need a different solution. at best getting a property into udev/hal17:06
asacfor virtual stuff17:06
=== fader is now known as fader|lunch
slangasekKeybuk: moo17:25
slangasekasac: hrm, surely hal should show a driver associated with those modules whether or not they're compiled in?17:26
Keybukslangasek: acpi-support17:26
slangaseksince "driver" != "module"17:26
Keybukwhy are we holding off uploading?17:26
slangasekKeybuk: mm, are we?17:26
Keybukslangasek: you'd asked pitti not to upload acpi-support17:27
Keybukand I'm pretty sure you asked me not to too17:27
slangasekKeybuk: that may have been an offer to do the upload that I failed to make good on17:27
Keybukmay I?17:28
slangasekyes, please :)17:28
Keybukdone17:29
asacslangasek: i think its a hal bug. yes17:29
asacslangasek: but i doubt its a fixable one for jaunty17:29
asac(guts feeling)17:29
Keybukyeah, this sounds one of those HAL being stupid problems17:29
slangasekasac: it surprises me if that's the case; seems to me that it should be straightforward to resolve via sysfs17:31
slangasekasac: except actually, the guy is talking about coLinux, so that may really be a virtual device and they have a different problem there17:31
Keybukhuh17:31
Keybukactually17:31
KeybukHAL resolves the last element17:31
Keybukso it should work with compiled-in17:31
Keybuk$ readlink /sys/bus/pci/devices/0000:00:1f.1/driver17:33
Keybuk../../../bus/pci/drivers/ata_piix17:33
Keybuk$ lshal | grep atea_piix17:33
=== maco_ is now known as maco
Keybuk  info.linux.driver = 'ata_piix'  (string)17:33
Keybuk(ignore spelling mistake typing from one screen to the other)17:33
Keybuk$ readlink /sys/bus/pci/drivers/ata_piix/module17:35
Keybuk$ echo $?17:35
Keybuk117:35
* slangasek nods17:35
asacslangasek: are we actually sure that virtual devices cannot be used like normal devices? e.g. dhlicent, get IP, and so on.17:38
slangasekasac: no - but there are definitely *some* set of virtual devices that should not be managed by NM, and the ones that can be should be identified and whitelisted17:39
slangasekand this patch restores the behavior from 8.1017:39
slangasekI think coLinux in particular is a case where you would want NM managing a virtual device.  You might have the same thing with e.g., VMware, depending how the network device looks from userspace17:41
jdstrandmathiaz: fyi, I deNEWd evolution-mapi source17:41
slangasekbut, those are special cases17:41
mathiazjdstrand: \o/ - thanks you!17:41
mathiazivoks: ^^17:41
ivoksmathiaz: jdstrand great17:42
asacslangasek: personally i would prefer if we could mark devices that cannot be managed than the other way around ... and only blacklist those that are marked like that17:42
ivoksmathiaz: i'll test it tomorrow17:42
slangasekasac: I think that's upside-down for the case of virtual devices; virtual devices that NM shouldn't touch are, I believe, the common case17:42
slangasekasac: I believe kees had this same problem on a different sort of virtual device17:43
asacslangasek: we will see. i will get this uploaded today i hope17:43
mathiazivoks: great - could you drop me an email with some testing instructions?17:43
asacslangasek: i know that there were complains from kees, you and soren ... and some bugs17:43
ivoksmathiaz: sure, once i try it my self17:43
asacslangasek: lets see how many complain when we dont do that anymore ;)17:43
mathiazivoks: so that I can write a call for testing and have other test the plugin too17:43
mathiazivoks: sure17:43
asacslangasek: i will try to get feedback from QA team and maybe soren to check VMs17:44
seb128jdstrand: thanks17:44
slangasekok :)17:44
asaci just fear that we introduce yet another upgrade regression by flipping the coin17:44
slangasekasac: it would be an upgrade regression confined to jaunty, and if someone complains, then we'll have information about what kinds of virtual devices it makes sense for NM to manage :)17:45
mathiazslangasek: openldap 2.4.15 seems to be an bug-fix only release. Should I request an FFe to merge from Debian?17:47
slangasekmathiaz: I haven't analyzed whether it's bugfix-only, I'll leave that to your judgement whether it needs an FFe. :)17:47
jdstrandmathiaz, ivoks, seb128: np17:50
=== fader|lunch is now known as fader
DavieyKeybuk: Is there much chance of getting a FFe for a conversion from init.d to upstart?  Once any potential upgrade regressions are tested?18:09
keesasac: I'm not sure my corner-cases of network interface madness really need to be considered until the NM core is more stable.18:13
=== maco_ is now known as maco
KeybukDaviey: not this release ;)18:22
slangasekkees: hey, a little support here, I'd like it if configuring openvpn didn't cause NM to screw up my default route ;)18:23
ograheh, upstart as FFe ?18:23
ografunny thought18:23
ograKeybuk, did we drop the hwclock initscript call on shutdown as well with your cleanup ?18:24
Keybukno18:25
asackees: right. i will decide whether we really fix the virtual-shoudlnt-be managed based on feedback i get during beta18:25
ogra:( ... sad that would have solved a lot of probs on ARM :)18:25
keesslangasek: heh, sorry.  :)  Yes, while I'd like NM to work sanely given my crackful configs, I'm not convinced it's possible in the short-term.18:25
asackees: also: the NM core is quite stable now18:25
asackees: at least architecture wise18:25
keesasac: I'm happy to start opening bugs if you think it's the right time for it.18:26
asackees: lets wait for 0.7.1 to be final ;)18:26
keesasac: between libvirt, my bridge, and VLAN interfaces, it gets really really upset at me.18:26
keesheh, okay18:26
asacshould happen in a few days ;)18:26
keescool18:26
DavieyKeybuk: forgive me, i mean just for one package.  Wasn't clear :)18:27
KeybukDaviey: what package?18:28
Davieymythtv-backend18:28
Keybukwhat's the rule?18:28
Keybukthe upstart job, sorry18:29
Davieyahh18:29
Davieycurrently just18:29
Davieyrespawn18:29
Davieyexec /usr/bin/mythbackend --daemon --logfile /var/log/mythtv/mythbackend.log18:29
Daviey(iirc)18:29
Daviey(and the "start on/stop on" ofc)18:30
Keybukthat job will not work with upstart 0.518:31
Keybukwhat's your upgrade plan?18:31
Keybukhow will you cope if people change the file between now and the next release?18:31
DavieyIt was the intention to tear out the old init.d using rm_conffile in packaging18:32
Keybukit's not the init script I'm concerned about18:32
Davieyofc, a warning might be prudent18:32
Keybukit's the migration between upstart versions18:32
Keybukthe next version of Upstart in Ubuntu does not have a compatible configuration file format18:32
Keybukthey're not even in the same directory18:32
Davieyahh, jobs.d ?18:33
Keybukno18:33
Keybuk(this is me saying I'd recommend not converting things to Upstart just yet, btw :p)18:33
Daviey(i guessed as such :) )18:34
Davieythanks18:34
Keybukyou'd only be creating more work for yourself18:34
Keybukwith no immediate benefit18:34
superm1well other than having the ability to automatically respawn the init script to cope with crashes18:35
superm1i think that was the driving feature for wanting to convert to upstart18:35
Davieyyep.. it's just that we've been umming and arrr'ing over switching for a couple of releases now :).  Currently we don't have a watcher, like mysql etc.  A crash goes unnoticed.. Ah well, good things will come from kinky koala18:36
DavieyKeybuk: ^^18:36
Davieythanks18:37
=== slangase` is now known as slangasek
liwhmm... https://bugs.launchpad.net/evolution/+bug/150963 shows the bug as confirmed/unknown in the Evolution project (that's upstream), and Fix released/wishlist on Ubuntu. because of the upstream status, it shows up in the list of bugs related to me.19:00
ubottuLaunchpad bug 150963 in evolution "evolution should allow me to configure it to not switch emails/folders when space is pressed at the end of a message" [Wishlist,Fix released]19:00
liwis there something I can to get it off the list?19:00
directhexask #launchpad imho19:01
LaserJockis there any place that says that spec wiki pages should use a particular namespace (such as Spec/ or Specs/)?19:02
LaserJockor a category like CategorySpec19:02
liwdirecthex, ack, good point19:03
jcastrodo we have an lp tag for bugs/patches that are major deltas from upstream or debian?19:09
slangasekliw: use https://bugs.launchpad.net/ubuntu/+bugs?field.subscriber=$me ?19:09
liwslangasek, I reported the bug, and I sort of feel I should stay subscribed, though19:10
liwthe status for the upstream project is wrong, that is19:10
slangasekusing that url certainly doesn't unsubscribe you, it just gives you a view that hides non-Ubuntu tasks19:10
slangasekoh19:11
liwslangasek, otoh, the url you gave is probably better for me anyway :)19:12
liwslangasek, incidentally, since you're awake -- do you have any19:13
liwer...19:13
liwslangasek, incidentally, since you're awake -- do you have any timetable for finalizing the debian/copyright spec in Debian? I expect it won't happen before jaunty releases19:14
slangasekfor finalizing it - yeah, I don't think that'll happen before jaunty19:14
slangasekI do want to still get a decision on harmonizing with Fedora's naming scheme done before then19:15
slangasek(with enough lead time to be able to deploy some of these in jaunty)19:15
directhexthe spec's going to be actually finalized? :o19:17
=== jdong_ is now known as jdong
bigondoes somebody know if I can get an upstream version freeze for dbus-glib with that debdiff? http://www.pastebin.be/16987 ?19:45
=== DrKranz is now known as DktrKranz
=== smb_tp_ is now known as smb_tp
ScottKpitti: Are you interested to be subscribed to bugs that affect gnome-stracciatella-session?20:11
=== maco_ is now known as maco
=== maxb is now known as Guest27135
rtgogra, lool: https://edge.launchpad.net/%7Etimg-tpi/+archive/ppa/+sourcepub/506110/+listing-archive-extra21:12
loolrtg: Ah it's great to see your efforts, thanks a lot; I'll try having a look tomorrow (late-ish here)21:25
loologra: Would love if you could have a look21:25
rtglool: yeah, now I need to make sure it all still works.21:25
loolrtg: I'm not sure how I'll test it yet, but I'm NCommander and ogra have use cases already21:29
lools/I'm/21:29
rtglool: I have an x86_64 test case, which is my primary interest21:30
loolrtg: I think NCommander was testing under qemu/armel21:31
rtglool: I'm pretty sure kexec itself will work. I've had mor issuers with initramfs hooks and such.21:31
mdkehi there. Is the removal of the log out and shut down buttons from the System menu in jaunty intentional?22:19
mdke(am asking for documentation purposes - that affects quite a few of our docs)22:19
Nafallomdke: they come back if you remove the FUSA-applet from the panel is my understanding.22:20
mdkeah, so they do22:20
slangasekIOW: yes, it's intentional22:20
superm1nifty; i didn't know there was a way to get them back22:20
mdkeok, so we'll have to try and explain that there are two possibilities in the documentation22:21
Nafallowell... they are not gone then ;-)22:21
Nafallojust hidden22:21
mdkeit's not very obvious how th get the FUSA applet back again :)22:23
mdkeit has a different icon in the Add to Panel dialogue22:23
mdkeso people who accidentally remove the applet may have some trouble turning off their computer...22:24
mdkeunless we document both ways22:24
mdkeso I guess the documents will have to explain both as alternatives22:24
=== Guest27135 is now known as maxb
=== maco_ is now known as maco
slangaseklamont: why does https://launchpad.net/ubuntu/+source/ia32-libs/2.7ubuntu3/+build/849008/+files/22:43
slangaseklamont: why does https://launchpad.net/ubuntu/+source/ia32-libs/2.7ubuntu3/+build/849008/+files/buildlog_ubuntu-jaunty-ia64.ia32-libs_2.7ubuntu3_FAILEDTOBUILD.txt.gz show ia32-libs FTBFS on ia64 when running a command that's not what it was told to run?22:43
slangaseklamont: nah, ignore me, reading comprehension fail22:44
slangasekdoko: do you know why lib32gcc1 on ia64 is being built from ia32-libs instead of from gcc-4.3?22:49
slangaseknet result is that ia32-libs is FTBFS on ia64 since intrepid, AFAICS22:50
dokoslangasek: ia64 doesn't have a biarch setup, would require a ia64->i386 cross compiler; I never did care much about doing this cross toolchain22:52
slangasekdoko: that's fair22:52
dokobut I can have a look at ia32-libs tomorrow22:52
slangasekdoko: ok - no hurry as far as I'm concerned, I just TIL but I don't particularly want to lose a whole day trying to wrangle the stupid thing22:53
directhexare the ia64 buildds itanic 1 or 2?22:57
mdkedidrocks: the latest update of yelp seems to have broken the adapt 04_new_ubuntu_layout.patch22:59
mdkedidrocks: scratch "adapt" from that, was a bad paste22:59
mdkedidrocks: yelp doesn't show the Ubuntu frontpage anymore, but rather the upstream TOC23:00
seb128mdke: he's sleeping now I think but I will sort that with him tomorrow23:03
mdkeseb128: ok, thanks very much. I can send an email or file a bug. It's not clear to me whether it's the change he made or an upstream change causing the problem23:04
seb128mdke: let me have a look23:04
mdkeseb128: no rush, thanks for looking into it23:04
seb128mdke: nothing obvious, the patch is still there, didn't change a lot and applied during the build23:09
mdkeyeah, I saw it applied ok23:09
seb128mdke: seems to be rather an upstream change, will sort that with him tomorrow23:10
mdkeseb128: thank a lot23:10
seb128you're welcome23:11
seb128there is not so many upstream changes, weird23:11
mdkeyeah, I was wondering if it was 2.24.123:11
mdkebut that's in intrepid too, right?23:11
mdkeoh, maybe not23:12
seb128no23:12
seb128no such version in the NEWS or changelog23:12
mdkeyeah23:13
mdkemysterious. I will try and get hold of DonS if we can't figure it out over the next few days23:13
ryanakcaOn the UDS sponsorship form, the ``crew'' does what? Setup tables / chairs / sound systems / errands / etc?23:14
seb128should be easy to revert the few commits between version and figure which one breaks it23:14
khemOn jaunty after applying updates from yesterday my wireless stopped working. I have a T61 before I investigate I thought to ask if someone else is seeing same23:18
maxbkhem: I have a Z61p, still working... probably too different to be relevant... anyway, probably #ubuntu+1 would be a better place?23:22
khemmaxb: thx23:23
maxbOops, I spoke too soon, after a reboot my wireless is gone too23:30
maxbOh no, it was just being slow to associate23:31
ebroderI'm...having a hard time debootstrapping a jaunty chroot - I'm getting an error about libstdc++.so.6 when I try to do an apt-get update within the newly created chroot. Did something about libc/libc++ change recently (or not so recently)?23:45
ebroder(I'm debootstrapping on a lenny machine. I've tried lenny's, unstable's, and jaunty's debootstrap package, and it errors out the same way each time)23:45
ebroder...oh wait. "W: Couldn't download package libstdc++6" is probably relevant23:47
maxb:-)23:53
ebroderIt looks like te mirror I was using sucks23:53

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