lamontmdz: amd64/i386 built, ppc still chunking along.12:18
lamontyou want log files?12:18
=== azeem [~mbanck@socks-out.lrz-muenchen.de] has joined #ubuntu-devel
mdzjustdave: at the point of the last message you see, does the system hang, power off, do anything?12:36
mdzlamont: logs are not necessary; just copy the debs to rookery so I can put them up for download12:36
lamontmdz: source and binaries (all 3 arch) at http://people.ubuntu.com/~lamont/mozilla - feel free to toss them somewhere else.12:39
mdzlamont: thanks12:40
justdavemdz: it just hangs01:12
=== azeem [~mbanck@socks-out.lrz-muenchen.de] has joined #ubuntu-devel
=== elmo_ [~james@george.kkhotels.co.uk] has joined #ubuntu-devel
=== elmo_ [~james@george.kkhotels.co.uk] has joined #ubuntu-devel
mdzelmo_: there seem to be packages in DesktopSeed which are lacking Task: ubuntu-desktop headers; is that normal?01:58
mdzseem to be all gstreamer stuff01:58
mdze.g, gstreamer0.8-alsa01:58
elmo_mdz: I just do it based on the 'desktop' output file of my germinate run01:59
mdzelmo_: I think that stuff was added on 09-02; whet was the last time that was updated?01:59
elmo_hmm, and gstreamer0.8-alsa is in there01:59
mdzthey show up in my germinate output02:00
mdzhmm, I think apt is just being weird02:00
mdzyeah, it's showing me the status file entry rather than the Packages file entry02:00
mdzwhich is a very weird thing for dumpavail to do02:01
mdzI guess it's playing its "the version numbers are the same so everything is cool" game02:02
=== jono [~jono@82-37-194-90.cable.ubr05.wolv.blueyonder.co.uk] has joined #ubuntu-devel
jonohi all02:06
jonoI am writing Linux Desktop Hacks for O'Reilly and we are looking for contributors to write a few hacks - would anyone be interested in getting involved?02:06
sivangjono : what are those Desktop Hacks? :)02:07
jonowell, we are taking suggestions02:07
jonowe are writing 80 hacks and 20 are available for contributors02:08
jonoany cool ideas, mail them to me at jono AT jonobacon D-O-T org02:08
elmo_heh, 'ubuntu-desktop' in Section: base02:11
justdavemdz: new lead.  added ramdisk_size=8192 to the yaboot boot parameters, and successfully booted non-smp02:11
justdavedropping a dmesg on the bug as soon as it finishes logging in02:11
mdzjustdave: how big is your initrd?02:12
elmo_mdz: good god, that is the Depends line of doom - are you sure that doesn't overflow one of apt's static buffers at least in woody?02:12
justdavemdz: the smp initrd is 4 MB even, the non-smp initrd is 3.8 MB02:14
mdzelmo_: no clue about woody02:14
mdzbut woody users won't get that package upgraded anyway02:14
mdzubuntu apt is fine with it afaik02:15
mdzjustdave: stinks of weird bootloader problems02:15
mdzelmo_: the stanza is only slightly larger than the d-i Sources one02:16
=== justdave reboots again to see if it's yet another fluke or if it's consistant this time
justdaveit's a fluke02:20
justdavejust booted without the ramdisk_size thing and succeeded02:20
mdzjustdave: have you upgraded since the last time you tried?02:20
justdavestill has devfs in there though.02:21
justdaveno.  I did update the kernel yesterday, but before my previous attempts.02:21
justdaveoh, and yaboot does let you pass kernel params from the boot prompt02:22
justdavetook all the garbage out of yaboot.conf, and experimenting from the boot prompt now that I proved that02:24
justdaveok, just got it to hang again02:26
justdaveonly changed from the previous boot was the "idebus=66" got removed.02:26
=== theantix [~ryan@80.185.novustelecom.net] has joined #ubuntu-devel
justdaveone more attempt, same params, just to make sure, still hangs02:27
justdaveadded idebus=66 back in, still hangs.02:28
justdaveso there's apparently no pattern to it02:28
justdaveadded video=ofonly in addition to the idebus=66 and it booted02:31
justdavebtw, the new startup sound is cool, I like it :)02:32
justdaveok, just successfully booted it with the default command line.02:35
justdaveso the command line options are another red herring02:35
justdaveappears to be just luck or timing or something02:36
mdzjustdave: so basically, it happens on UP and SMP systems, and regardless of the kernel command line02:39
Kamionsivang: ':syntax on'02:40
justdaveseems that way.02:40
mdzjustdave: try unplugging all your USB devices when you boot02:40
mdzand 139402:40
Kamionjustdave: yaboot lets you pass kernel parameters from the boot prompt as long as they aren't initrd= (well, you can pass it, but it's not useful), which is kind of annoying02:41
sivangKamion : you're replying my vim question? :)02:41
Kamionsivang: yes02:41
justdaveno 1394 on it right now.02:41
sivangKamion : boy that was a while, thanks!02:41
justdaveI'll have to change the default kernel in yaboot to do that. :)  my keyboard is USB :)02:41
mdzKamion: any idea about #1379?02:41
mdzjustdave: you can plug it in after boot if it works02:42
mdzjustdave: just try it and see if it makes a difference02:42
mdzoh, you mean you need to change yaboot.conf02:42
Kamionsivang: it's IRC, it often has a big delay built-in02:42
justdaveyep, done.02:42
sivangKamion : yes, I've got used to it by now ;-)02:43
Kamionmdz: I haven't had any idea about that since the start02:43
justdave(well, powered up -- still waiting for yaboot prompt to timeout)02:43
justdaveok, all USB and 1394 removed, it hung02:43
mdzKamion: I've searched around a bit and haven't found anything in Debian land02:43
mdzbesides that bug report you pointed out02:43
mdzwhich doesn't seem to have any leads02:44
mdzthere goes my last idea02:44
mdzthat was from the YDL website02:44
mdzHang after install: If after install your computer hangs on re-boot, unplug any USB hubs you may have. An intermittent problem causes the hub to interact poorly with the CUPS printer software. 02:44
mdzthat sounded fairly promising, since the CD boot was working and not the HDD boot02:44
mdzof course, this is hanging way before CUPS starts02:45
mdzwhat bootloader does the CD use on ppc?02:46
Kamionthere's no other sensible option02:47
justdaveplugged the USB stuff back in and booted, and it worked.02:51
Kamionbah, linux-kernel-di-{i386,powerpc}-2.6 failed to build02:57
Kamion... and it was because I'm dim02:58
=== Kamion fixes
=== carlos was able to downgrade debian sid into debian warty without major problems
carlosdo we want a kind of howto about it?03:04
mdzcarlos: only with BIG BOLD BLINKING warnings :-)03:05
carlossure :-)03:05
carloswiki or website ?03:05
mdzwiki, I suppose03:05
carlosok, I will do it tomorrow. Good night03:06
sivangwhy would someone like to downgrade from sid to warty anyways? :-) 03:18
=== sivang hides form mdz
sivangwow, just noticed the girl on the artwork....the best default livecd background I've yet to see...03:26
sivangmdz : would it be reasonable to push in stuff to yelp before the release (no further than the 17th) as this does not fall even under special circumstances? although yelp might not be effecting too much I suppose.03:31
mdzsivang: what kind of stuff?03:33
sivangmdz : I am not sure it will be ready for this, but just in case it will. It this listed under "roadmap" in http://wiki.ubuntulinux.org/DocumentationTeam.03:39
sivangmdz : if you have any comments, suggestions I'd be glad if you could mail me in , or on -devel. I'm hitting bed now, this time _for_real :)03:43
sivangnighters everybody03:43
=== x4m [~max@198.155-200-80.adsl.skynet.be] has joined #ubuntu-devel
danielsha ha ha, oh man06:01
fabbioneoh god06:30
fabbionemdz: LATER is not even documented06:31
fabbioneand imho it means that it will be fixed later06:31
fabbionethat also means that after release all LATER bugs will be reopened06:31
=== justdave hates LATER
justdavethat should be a status, not a resolution, if we have it at all06:35
danielsRESOLVED/LATER is interesting06:35
danielsbecause then the bug just falls out of our BTS06:36
danielsi think setting the target milestone to Hoary is far better, IMO06:36
justdavethe only reason that's not gone from Bugzilla upstream already is because customizable statuses was coming down the pike, and we didn't want to screw over people that were actually using it before they had an easy way to put it back.06:42
justdaveexcept the guy who was working on the customizable statuses dropped off the face of the planet06:42
=== hazmat [~hazmat@c-24-15-10-12.client.comcast.net] has joined #ubuntu-devel
fabbionedaniels: 2153 is your ;)06:48
danielsfabbione: yeah06:48
=== daniels heads off to vote.
mdzfabbione: yeah, I am not so sure why we would use that rather than leaving the bug open and setting its milestone07:05
fabbionemdz: same reason as i posted in the comment :-)))07:05
fabbioneit is not something that we can fix for sure for hoary07:06
mdzfabbione: I don't want those bugs to vanish from the list of open bugs though07:06
fabbionemdz: it's not vanished...07:06
fabbioneit's still in bugzilla07:06
fabbioneand i did plan to reopen tham afterwards07:06
mdzto me, 'resolved" means it has been dealt with07:06
mdzI know, I understand07:06
fabbionewell bugzilla is wrong07:06
fabbioneLATER is a "tag" not a status07:06
mdzit is best if we all use the same semantics for handling bugs07:07
mdzif you are using LATER and someone else is using a milestone, that's unnecessarily confusing for everyone07:08
fabbionemdz: i don't disagree, what i wrote in the bug is: LATER because we don't know if it can be fixed for hoary07:10
fabbionemdz: that's kinda different from a milestone07:10
justdavemaybe add a milestone called "Future"07:10
mdzfabbione: "hoary" is our name for all of the time between the warty release and when the world crumbles into dust :-)07:10
mdzanything that is not now or the past is "hoary"07:11
mdzour list of feature goals is long enough to fill that time07:11
fabbionei will reopen the bugs and mark them hoary07:12
fabbionemdz: 2119 -> for me it's invalid07:15
mdzfabbione: I didn't read the logs; what is the deal?07:16
fabbionemdz: it's 24 to 16 bit and no DRI07:18
fabbione+ one obscure option that for sure i am never going to set by default07:18
mdzfabbione: wait, I am rereading it and he is saying Ubuntu is faster07:20
mdzand his problem is with userlinux?07:20
fabbioneno the other way around07:20
mdzUbuntu is07:21
mdzmarginally faster07:21
mdzthan UserLinux -- moving a card in freecell takes 5-8 seconds07:21
mdzinstead of 10-15 seconds.07:21
mdzmoving a card in freecell in UserLinux takes 10-15 seconds.  07:21
mdzI think he has reported his bug to the wrong BTS07:21
chrisaSomeone is using freecell as a benchmark? That terrifies me07:22
fabbioneeither he wrote to the wrong BTS or the wrong order of distros07:22
fabbioneboth cases -> INVALID07:22
mdzI did not see that before for some reason07:22
fabbionewe can check in the logs07:23
fabbioneXFree86 Version (Ubuntu 4.3.0.dfsg.1-6ubuntu23 20041004140652 root@terranova.warthogs.hbd.com)07:23
fabbionehe is talking about ubuntu07:23
mdzfabbione: both the fast and slow logs say Ubuntu07:25
hazmatis it possible to have packages not install themselves into the rc.d directories by default.. ie make an admin explicitly enable them.. i just got myself locked out of my ubuntu system after installing openldap, and watching slapd hang the system during startup.. had to fix things with the install cd.07:25
mdzfabbione: so either UserLinux is borrowing Ubuntu packages, or he is confused07:25
fabbionemdz: exactly07:30
fabbionemdz: do you where the userlinux archive is?07:34
fabbionei can't find the pool07:34
jdubfabbione: userlinux == sarge07:34
jdubfabbione: apart from a few meta packages07:34
fabbioneIt also doesn't configure the national language keyboard for X. It certainly shouldn't install any such thing without asking first. The country the box happens to sit in may determine it's time zone, but that's about it. Certainly not the language (there are often several in the same country anyway, and chances are none of them is English, but the admin still wants his computer in English)07:35
fabbioneThat's from their wiki07:35
fabbioneit looks sooooo much as the bugs we get07:35
fabbionejdub: they released an update the 17th of Sept.07:35
jdubfabbione: sarge is updated all the time :)07:36
fabbionejdub: our code is not in Debian07:39
fabbioneat least not the autodetection stuff07:39
jdubfabbione: it seems, from the bug, they're saying userlinux but using ubuntu. simplest explanation.07:40
jdubso, a couple of people have mentioned that they're not getting any sound at all07:44
jduband i think i've hit a similar situation07:44
jdubsnd-intel8x0 driver, nforce2 hardware07:44
fabbionemdz: your announce is on distrowatch07:44
jdubno amount of alsamixer tweaking has helped07:44
mdzfabbione: I sent that Tuesday, and didn't realize it was waiting in the moderation queue :-(07:45
jduboof, really?07:45
jdubi thought it had gone through07:45
jdubdid you send it to deve or user too?07:45
mdzjdub: mako just pushed it out today07:45
mdzno, I think only -announce07:46
mdzsome folks were listening, though -> 50 torrents07:46
jdubsorry about that07:46
mdz(more than the install CD right now)07:46
mdzjdub: for some reason I don't think I got a notification saying it was held for moderation07:46
mdzor I would have just taken care of it07:46
jdubmight've been the double-auth mailman thing i'm getting07:46
jduboh wait07:47
jdubi'm not even on -announce moderation07:47
mdzI thought the sender was notified too?07:49
jdubshould've been, yeah07:49
jdubbut you're also moderator07:50
jdubi think07:50
jduboh, that's just ubuntu-security-announce07:50
=== jdub is stuck on console atm ;)
jdubcould sharing an interrupt with ehci be an issue?07:51
mdzdepends on what with07:51
mdzif it's a parallel  port or something, yeah :->07:51
jdubthe sound card07:51
=== jdub is talking about the sound issue
mdzis this the mixer-volume-is-up-and-it-looks-like-it's-working-but-no-sound issue?07:52
mdzhaven't seen that one myself07:52
mdzcould be IRQ stuff, yeah07:52
jduba few locals are asking about it07:52
jdubonly *just* started happening on this machine07:52
jdubpreviously, i was enjoying the just works sound/volume foo ;)07:52
jdubstuff under /proc/asound looks fine07:53
mdzjdub: pci=noacpi?07:59
mdzthose are my stock answers to everything these days :-)07:59
mdzheadache? noapic07:59
jdubhaven't tried them... will do now07:59
mdztoothache? pci=noacpi07:59
mdzpeople look at me funny08:00
danielsmdz: no more jellybeans08:59
=== hazmat [~hazmat@c-24-15-10-12.client.comcast.net] has joined #ubuntu-devel
danielsthom: ping09:17
=== doko [doko@dsl-084-057-024-146.arcor-ip.net] has joined #ubuntu-devel
=== mojo [~mojo@220-244-212-78-vic.tpgi.com.au] has joined #ubuntu-devel
mojohi all fellow developers!10:02
mojoI'm getting stuck with stupid Bug #155210:02
mojotoo shit I am, can't find other way around to fix this, hope someone here give me a hand, let SSH to my box and fix it with me10:03
=== Maskie [~maskie@196-30-110-210.uudial.uunet.co.za] has joined #ubuntu-devel
=== evilstoy [HydraIRC@cmung3567.cmu.carnet.hr] has joined #ubuntu-devel
=== evilstoy [HydraIRC@cmung3567.cmu.carnet.hr] has left #ubuntu-devel []
=== doko [doko@dsl-084-057-059-003.arcor-ip.net] has joined #ubuntu-devel
=== carlos [~carlos@69.Red-80-33-181.pooles.rima-tde.net] has joined #ubuntu-devel
carlosjdub: should we report missing icons from ubuntu-artwork?11:24
thomdaniels: ack?11:28
danielsthom: ack-ack11:28
=== daniels tries to remember what he wanted two and a quarter hours ago.
jdubcarlos: nah, they're known11:35
thomdaniels: heh11:54
thomyou're too young to be going senile11:54
danielsthom: and you don't have that problem? :)11:55
thomno, but i work around this problem by asking at the time :P11:56
danielsthom: heh, fair cop11:56
danielsoh, right.11:56
danielssuspending on the X40.11:56
danielseven when removing ehci, can't get it going with echo 3 > /proc/acpi/sleep, then coming back with echo 0 > /proc/acpi/sleep11:57
danielsit never comes back, it just sits there with both the battery and sleep lights on11:57
danielsany known issues?11:57
danielsi.e. is this your acpi-support bong gone wrong? :)11:57
thomno, acpi-support does nothing with suspend11:57
thomhit the power button when it's suspended, what happens?11:58
danielssuspended with ecoh 3 > /proc/acpi/sleep?11:58
daniels(the /proc/acpi/sleep stuff being my additions to lid.sh)11:58
thomdaniels: grab http://www.planetarytramp.net/acpi-x40.tgz12:00
carlosjdub: ok12:07
danielsthom: do I need the tpbutton stuff or whatever?12:07
danielsthom: (fwiw, it froze, and I needed to reboot)12:08
thomdaniels: shouldn't do12:10
danielswell, it works fine with x4012:10
danielsbut with mangling /proc/acpi/sleep, it freezes when resuming12:11
danielswith your x40 stuff it doesn't actually suspend, however12:11
danielsit just chvts12:11
thomit'll only suspend if you're on battery12:11
danielsso I did it on battery, and it hung :)12:17
thomi've not tried with the current kernel tree, i have to admit12:17
danielsthis is just with
thomi (or rather mjg59) had a custom 2.6.7 where it worked12:20
danielsany known gotchas?12:20
danielswhat were the patches -- acpi.sf.net?12:20
thomthat and one other i don't remember12:21
danielsmjg59: ping?12:25
daniels(own reference to avoid senility-related troubles: x40 acpi)12:25
=== seb128 [~seb128@ABoulogne-110-1-2-101.w80-13.abo.wanadoo.fr] has joined #ubuntu-devel
=== elmo_ [~james@] has joined #ubuntu-devel
=== sivang [~pooh@] has joined #ubuntu-devel
sivangmorning/afternoon all01:15
=== x4m [~max@198.155-200-80.adsl.skynet.be] has joined #ubuntu-devel
=== sjoerd [~sjoerd@fire.ipv6.luon.net] has joined #ubuntu-devel
=== sivang [~pooh@] has joined #ubuntu-devel
thomdaniels: btw, you really need to push freedrtools, since otherwise we're gonna wind up with about 50 forks01:49
=== cc [~byte@drbyte.fedora] has joined #ubuntu-devel
dokoanybody around who can/wants to review #2193?02:19
Kamiondoko: patch looks good to me, I'd been hoping somebody would investigate that02:26
dokook, so I'll upload that.02:29
dokowhile you are here: did a fresh powerpc install, sound works, keyboard not.02:31
dokokamion: which package writes /etc/environment on installation?02:31
Kamiondoko: prebaseconfig02:46
Kamioncombined with base-config02:46
Kamionwhich reminds me ...02:46
Kamiondoko: keyboard doesn't work in what way?02:47
Kamionanyone object to this patch to base-config?02:49
Kamion-                       grep -v "^LANG=" /etc/environmment > /etc/environmment.new || true02:49
Kamion-                       mv /etc/environmment.new /etc/environmment02:49
carlosmdz: ping02:49
Kamion+                       grep -v "^LANG=" /etc/environment > /etc/environment.new || true02:49
Kamion+                       mv /etc/environment.new /etc/environment02:49
dokokamion: no, don't object. could you have a look at #219502:57
Kamionso my patch fixes the first half at least02:58
Kamionand almost certainly the second; what breaks exactly?02:59
Kamion"The file /etc/enviroment (without typo)"02:59
dokohow do I force dpkg-reconfigure console-common to ask for the language?03:00
dokothe grep in line 47 prints two lines.03:00
Kamion        for var in LANG LC_ALL LC_CTYPE; do03:00
Kamion                value=$(egrep "^[^#] *${var}=" /etc/environment | cut -d= -f2)03:00
Kamion                eval $var=$value03:01
Kamion        done03:01
Kamionright, ok03:01
Kamionso you mean /etc/init.d/keymap.sh I'm guessing03:01
dokocorrect, thanks for thinking for me :)03:01
Kamionthat's such a crap script; why doesn't it just source the damn file?03:01
Mithrandirit should use the interface provided by PAM, shouldn't it?03:02
Kamionoh, yeah, duh, /etc/environment not a shell script03:02
Kamionyes, it should use pam_getenv then03:03
Kamionanyway, fixed03:04
dokook, keyboard works, after I manually reconfigure console-data to use the german keyboard.03:07
dokowhy isn't this done correctly by the installer? wondering ...03:07
Kamiondoko: hm? it should be, isn't it just that base-config bug?03:09
dokowhich one?03:10
Kamionconsole-data is very bizarre, but the installer should set up the keymap03:10
Kamion(even if it doesn't look like it has when you reconfigure console-data)03:10
seb128doko: not sure than #2194 is a bug03:10
Kamiondoko: /etc/environmment for /etc/environment03:10
Kamionif that screws over /etc/init.d/keymap.sh then it's maybe not too surprising that the keymap's broken?03:11
dokoseb128: believe me, it is. if you use two nomens in the singular form, you have to use the verb in the plural form.03:11
seb128doko: on an another chan I've 2 people saying that "ist" is correct03:12
Kamiondoko: isn't it like English where if you use "or" the verb is singular, whereas if you use "and" the verb is plural?03:12
seb128Kamion: I think so03:12
dokoah, ok. I'll review that with the next installer.03:13
seb128that's a "or" in this case .. so "ist" seems to be right03:13
dokohmm, I'll review that in the "Duden", but it sounds wrong.03:13
Kamionalthough I do remember a German class where we were told there was some weird grammar rule, but I don't remember exactly which way it went, hmm03:13
KamionI could believe either03:13
dokoseb128: which channel?03:14
seb128doko: a gnome, but they are not translators afaik03:14
seb128I'll try to ping a i18n guy03:15
carlosseb128: we are missing some spanish translations for our gnome-panel changes, could you give me an updated .po so I could finish it?03:18
azeemdoko: use "ist/sind fehlerhaft" :)03:18
seb128you (i18n guys) need to set a system up03:18
seb128I can't keep to synchronize all sort of translations in all the packages in that way03:19
KamionThe i386 daily CD image I just built should have support in the installer for ipw2100 and ipw2200 cards.03:19
Kamionallegedly acx_pci too, but that didn't work for me03:19
seb128carlos: that's not manageable03:19
KamionI guess I have a bunch of d-i translations lying around in bugs somewhere to integrate, too.03:19
Kamionmdz: if you could test the current i386 CD on your ipw* box, that'd be good03:20
carlosseb128: O:-)03:21
carlosseb128: hoary time03:21
seb128carlos: no seriously, people keep reporting translation updates/changes on different GNOME pieces03:21
seb128carlos: merging the changes are a pain each time03:21
carlosseb128: btw, changing the talk subject... the trash applet does not sees my iPod trash directory03:21
carlosseb128: how do you handle that under Debian?03:21
Kamionseb128: in fairness this *is* what carlos is working on full-time :-)03:22
seb128carlos: we don't, that's upstream job03:22
carlosKamion: yes, that's the idea :-P03:23
seb128carlos: BTW for the trash, please check with jamesh, he made the changes for monitor extra drives03:23
carlosseb128: ok03:23
=== carlos did his first kernel patch today :-D
carlosand I'm able to use my iPod under Ubuntu as a normal user, finally !!03:24
dokoseb128, kamion: as usual: both things are ok, altough they say the singular finitum is more common.03:27
seb128carlos: I've updated #941 to include the list of strings, if you do a translation include the full file with this format (no information about lines or so, it's problematic for updates when code change)03:29
seb128doko: ok, so NOTABUG, ok ?03:29
carlosseb128: ok, thanks03:37
dokoseb128: yes03:45
=== sivang [~pooh@] has joined #ubuntu-devel
=== Mithrandir waves with his nvidia-glx package for amd64
elmo_hmm, lshw on amd64 is interesting...04:40
elmo_it seems to be peeking through every single byte of memory04:41
Mithrandirhow so?04:41
Mithrandirapart from it listing a fair bunch of hw, it seems fine here.04:42
elmo_read(3, 0xb2e41811, 1428514031)         = -1 EFAULT (Bad address)04:43
elmo_read(3, 0xb2e41810, 1428514032)         = -1 EFAULT (Bad address)04:43
elmo_read(3, 0xb2e4180f, 1428514033)         = -1 EFAULT (Bad address)04:43
elmo_read(3, 0xb2e4180e, 1428514034)         = -1 EFAULT (Bad address)04:43
elmo_root      4560 32.1  0.0  2692  860 pts/48   R+   15:24   6:01  |           \_ lshw04:43
Mithrandirweird, doesn't do that on my box.04:44
elmo_heh, 'dpkg: warning 'amd64' not in the remapping table'04:44
=== pitti [~martin@] has joined #ubuntu-devel
carlospitti: here :-P04:55
pitticarlos: yes, I remember04:55
pitticarlos: the "I have to unload the module first" problem?04:55
carlosno, that's not the bug I'm talking about04:56
carlospitti: the one that does not let me use the iPod after mount04:56
carlosbecause the owner is root04:56
carlosI did a patch (https://bugzilla.ubuntu.com/show_bug.cgi?id=2197)04:56
pittiall files should be owned by you, but only the mount point itself is wrong?04:56
carlosno, all files are owned by root04:57
carloswell, in fact the problem is that all files have their original values, like when you mount an ext2/ext3 filesystem04:57
pittihmm, can you add the patch to the bug as attachment?04:58
carlosit's in the upstream bugzilla, but yes, I could do it04:58
pittiI think I cannot decide that, but mdz should. Maybe you should assign the bug to Herbert and CC mdz for approval?04:58
pittiWould be great if that finally worked04:58
pittiiPods are quite common04:58
carlospitti: no, that's not why I want to ask you :-P04:59
carlosafter reading some code, seems like my fix is wrong04:59
chrisaThe number of ipods I see on campus frightens me04:59
carlosthe hfsplus code implements the uid/gid options like a fallback when they don't have any posix information for a file05:00
pittiugh, that's ugly05:00
carlosthat information exists with iPods so the driver just ignore it05:00
pittiI thought it should just map any root to the given uid05:00
carlosme too :-)05:00
carlosthe problem with my patch is that if you mount it as root without uid/gid arguments, all files are owned by root05:01
carlosthat's why I'm not sure if my fix is valid05:01
carlosno matter if you have a file owned by uid=500, next mount will show it as owned by root05:02
pittiwhatever you modified, can't you just check if uid != 0 before changing anything?05:03
carlosso, the question is... How do you think it should be fixed?05:03
carlosso if the current uid == 0 I don't change anything?05:03
carlosyes, it's possible05:04
pittiwell, this is just my first thought, I've never seen the code05:04
carlosdon't worry about the code.05:04
carlosthink about ext2/ext305:04
carlosit does not have any feature that lets you map all uid/gid to one concrete uid/gid05:05
carlossame as hfsplus05:05
elmo_hmm, after upgrading the buildds to, I've had two in a row hang on apt-extracttemplates - strace shows it stuck in a futex.. anyone else seen this/know about it?05:05
carlosI just need a suggestion (under your point of view) about a way to implement it :-P05:05
pittithis uid= option is certainly supposed to be a hack anyway :-)05:05
pittibut if some fs support it, all others should, too05:06
=== pitti looks at the patch
pittihmm, just by looking at the patch I cannot see how it works05:06
pittiso inode->i_uid is the uid that the file really appears under?05:07
Mithrandirfabbione: around?05:07
pitticarlos: and HFSPLUS_SB(inode->i_sb).uid is the uid that is saved on the iPod?05:07
pitticarlos: in the patch I don't see the value of the uid=XXX mount option05:08
carlospitti: that's a "general" uid value that it's set on startup to the current uid or to the uid argument05:08
pitticarlos: where is that stored=05:08
elmo_nm, it's #185405:08
pitticarlos: s/=/?/05:08
carlospitti: inside that structure05:08
carlosHFSPLUS_SB(inode->i_sb) is a kind of superblock struct05:09
carlospitti: it's initialized as:05:10
carlosopts->uid = current->uid;05:10
carlosopts->gid = current->gid;05:10
carlosopts is HFSPLUS_SB(inode->i_sb)05:10
carlosand then, it parses the options05:10
carlosand updates those values if they are present05:10
pitticarlos: phone, brb05:10
carlosno problem05:11
pitticarlos: back05:15
carlosI'm looking at MacOSX for the way they handle it05:15
pitticarlos: as you describe it, it should be fine to only update opts->uid to the override value if the latter is != 005:17
pitticarlos: since uid=0 makes not really sense, it can be regarded as "not set"05:18
carloslatter == uid option or current->uid?05:18
pitticarlos: the override value, so the uid= option05:18
pitticarlos: but AFAICS you should not delete the lines in the patch then05:18
pitticarlos: just override them afterwards if uid= != 005:18
pitticarlos: I did not see the whole code, but does that make sense?05:19
carlospitti: well the check should be if uid argument is given or not05:19
pitticarlos: if you can decide that direcly, so much the better05:19
carlosthe problem comes when you don't use the argument :-)05:19
pitticarlos: I thought that 0 (or -1) or whatever could be a default value if no uid= option is givne05:20
pitticarlos: the code should work correctly if uid== is not given, right?05:20
carlosif there is no uid given, the uid that is used is the process uid05:21
carlosso perhaps I could do that check05:21
pittibut that sounds wrong as well05:21
carlosif current->uid ==  HFSPLUS_SB(inode->i_sb).uid, then I do nothing05:21
pittiif no uid is given, it should use the uid that is saved on the ipod?05:21
carlospitti: yes, it does it that way, the problem comes with my patch :-)05:22
carlosseems like hfsplus let you store files without posix permissions information05:22
pitticarlos: ah, I see.05:22
carlosand they are using the uid and gid options to give a default value for that case05:22
pitticarlos: hmm, but if it does not have POSIX permissions, it should default to root, to be sure05:22
pitti(which is then mapped to uid= of coruse)05:22
pittiwhich is essentially the same, as I see now05:23
pitticarlos: now I understood what they use the uid and gid options for currently05:23
carloswell, the way they do it let a normal user mount it and all files with that information will be owned by that user05:24
pittiso my proposal would be: set uninitialized uid/gid of files to root:root05:24
carlosif I change it to 0 then I will break that case05:24
pittiand later, when you evaluate the mount options, change all root files to uid=XXX05:24
carlosmakes sense, but there is still the problem when you mount it as a normal user from a fstab entry with the "user" option05:25
pittidoesn't that imply uid=getuid() and gid=getgid()?05:26
pittiit should05:26
pitti^^^ (as mount options)05:26
pittijdub: here05:26
carlosdon't know05:26
carlospitti: is that possible inside fstab?05:26
carlos uid=getuid(),gid=getgid()05:27
pitticarlos: no, not in fstab05:27
pitticarlos: it should be in the option parsing05:27
pitticarlos: if no uid= option is set, but 'user' is set, it should imply uid=getuid()05:28
carlosso if there is a "user" option, then I should assume that?05:28
carloslet me check...05:28
pittijust as VFAT05:28
pittiwell, I had to try that out05:28
pittino, please forget that05:29
pittiuser does not imply uid=getuid(), sorry05:30
carlostrue I don't see that option inside the fat driver05:30
pittithis is not done with VFAT, I just mixed that up, pmount explicitly gives uid=getuid() options05:30
pittiif you use pmount, the 'user' option is not important anyway05:31
pittiI think we don't need to consider the case where the iPod is in fstab05:31
pittiit should work with pmount, so uid= and gid= has to work correclty05:31
carloswell, it's not iPod but any hfsplus drive05:31
pitti(more correct than my spelling, that is :-) )05:31
carlosapple has the uid and gid options like vfat has05:32
carlosso I suppose it's a good argument to use it that way05:32
pittiso conclusively, you revert the patch and instead set the effective inode uid to the mount option only if the mount option != 0?05:32
=== carlos looks for the hfsplus documentation for a reference about what happens when there is no posix information...
pittiand maybe implement defaulting to root:root if no posix info is available for a file?05:33
carlospitti: yes, something like that05:33
pitticarlos: well, anyting other than root makes no sense05:33
carlosI will read the documentation first to be sure I'm not breaking anything05:33
pitticarlos: anything other than root makes no sense05:35
pitticarlos: of course it should be overridden by uid= afterwards05:35
carlospitti: funny:05:37
carlos    The Mac OS X user ID of the owner of the file or folder. Mac OS X versions prior to 10.3 treats user ID 99 as if it was the user ID of the user currently logged in to the console. If no user is logged in to the console, user ID 99 is treated as user ID 0 (root). Mac OS X version 10.3 treats user ID 99 as if it was the user ID of the process making the call (in effect, making it owned by everyone simultaneously). These substi05:37
carlostutions happen at run time. The actual user ID on disk is not changed.05:37
pittiphone again... sorry ...05:38
carlosdon't worry05:38
pittiparty this evening...05:38
=== pitti cries out
pittiwho invented this crap?05:39
carlospitti: Apple :-P05:40
pittithe actual user ids on the disk should not be changed anyway05:40
carlos    The Mac OS X group ID of the group associated with the file or folder. Mac OS X typically maps group ID 99 to the group named "unknown." There is no run time substitution of group IDs in Mac OS X.05:40
pittiso they have "root stays root" and "99 becomes the user in uid="?05:40
carlospitti: no05:41
carlosroot stays root, 99 becomes the user that is logged in05:41
pittioh no, sorry05:41
carlosand with latest MacOSX 10.305:41
carlos99 becomes getuid()05:41
pittidynamically for every process then05:42
carlosand then, they have the uid, gid options to change it at mount time05:42
pittiso if both you and me are accessing the same file, both of us would think that the file belonged to us, right?05:42
carlospitti: no, dynamically for the mount command05:42
carlosperhaps you are right...05:42
pittiwell, what is "the call"?05:42
carloslet me check it :-)05:42
pittifile access or mount?05:42
pittibut they say "making it owned by everyone simultaneously"05:43
carloswe will know it soon, let me create a new user in my imac05:43
pittibut anyway, if that were already implemented like this in linux, the files should belong to you no matter whether it is mount or file access05:43
Mithrandircould somebody review http://err.no/patches/linux-restricted-modules- ?05:44
pittiso the current implementation neither fulfils the apple specs nor the usual uid=/gid= semantics05:44
carlospitti: true05:45
pitticarlos: I think we should not use the uid 99 semantics05:45
pitticarlos: because it could be a regular system user on somebody's computer05:46
pitticarlos: and this could even be a security hole (because any user could then access files which are meant only for the user with id 99)05:46
carlospitti: it's owned by the process that access the device not the mount command05:46
pitticarlos: so it's really the 'I belong to everybody' sematics?05:46
Mithrandirpitti: 99 is statically allocated and in the range owned by base-passwd, though.05:47
pittiMithrandir: I know, but you know, there is always one person.. never mind05:47
carlospitti: yes05:47
pittiIt's not a big deal, but  I learned to be paranoid :-)05:47
Mithrandirunix gives you enough blackpowder and rope to kill yourself in various ways.05:48
Mithrandirthat's a feature. :)05:48
carlospitti: if I'm root, I'm the owner, if I'm carlos, I'm also the owner :-)05:48
pittiMithrandir: yes, but if I shoot myself in the foot, then I should know what I'm doing05:48
pitticarlos: probably it's not too bad to just implement Apple's semantics05:48
pitticarlos: although I don't really like it05:49
carlospitti: the fact is that it's a hfsplus volume and the designer did it that way...05:49
pitticarlos: I would not want to enable other users to overwrite my hfsplus files05:49
carlospitti: yes, that's my point05:49
carlospitti: btw, that will not happen if you use the uid/gid options05:49
pitticarlos: OTOH I'm not really the one who can overrule Apple :-)05:49
pitticarlos: oh, not?05:49
carlosno, because it will not be 99 :-)05:49
carlosit will be the uid you say :-D05:50
pittiMithrandir: I meant that I would not expect my files go world-writeable if I create an user with id=9905:50
pitticarlos: but the file on disk has still 9905:50
carlospitti: but it's the hfsplus driver which changes it05:50
carlosI will prepare a patch to follow Apple rules05:51
=== pitti slowly gets confused about so many different uids
Mithrandirpitti: well, true, but you shouldn't create a user with uid = 99. :)05:51
carlospitti: when it's ready i will show you it, perhaps it's easier to understand that way...05:51
pittiMithrandir: now I know that05:51
pittiMithrandir: :-)05:51
pitticarlos: agreed. "Use the Source, Luke"05:51
pitticarlos: I think I now understood the problem05:52
pitticarlos: BTW, you should put this Apple spec into the bug report05:52
pitticarlos: then the patch can be verified against it05:52
pittiguys, I've gotta go. In an hour 10 friends will arrive and I have to prepare the meal05:53
pittiSee ya later!05:53
pitticarlos: Have fun implementing the stuff! :-)05:53
carlospitti: have fun05:55
=== Mithrand1r [~tfheen@vawad.err.no] has joined #ubuntu-devel
sivangam trying to build a package, and I get this :" found. Your intltool is too old.  You need intltool 0.29 or later.06:53
sivangcan anybody tell me why it does that? do we have a newer version of intltool ?06:54
sivangit's a gnome package btw06:54
seb128apt-get install intltool06:57
seb128warty has 0.31.106:57
sivangseb128 : ok, thanks06:58
sivangseb128 : it tells it's already the newest version 06:59
seb128ok, so fix the upstream soft06:59
|trey|sivang: apt-get update && apt-get install intltool07:19
sivang|trey| : thanks, already solved it. I had done something bad to my packages, so reinstalled from source and applied the changes back everything went to normal.07:20
elmo_gar, how did we end up shipping without ethtool07:32
=== sivang [~pooh@] has joined #ubuntu-devel
Mithrandirmdz: ping?07:45
danielsthom: yeah08:07
danielsthom: the red hat guys are getting a little antsy08:08
=== daniels sleeps.
=== pampa [~santiago@c68.112.163.226.dul.mn.charter.com] has joined #ubuntu-devel
=== jdub_ [~jdub@home.waugh.id.au] has joined #ubuntu-devel
mdzMithrandir: pong08:46
=== pampa_ [~santiago@c68.112.163.226.dul.mn.charter.com] has joined #ubuntu-devel
mdzwhat's all this gs-esp stuff about?09:00
dokomdz: see #2193, after a fresh install the installer wants to fetch tcl8.0, gs-gpl and gs, which are not on the CD. these patches resort the recommends/depends so that gs-esp (which is on the CD) satisfies the dependencies/recommendations.10:10
mdzdoko: did you test that it has the desired effect?10:20
dokowell, difficult to test. I hand edited the package list before the download started and no more packages were fetched from the net. no, I didn't build a new CD.10:23
=== sivang [~pooh@] has joined #ubuntu-devel

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