/srv/irclogs.ubuntu.com/2012/05/17/#ubuntu-devel.txt

zulcan someone review python-jsonschema its been sitting in new for a while now and its needed for glance00:11
geofftSomeone mentioned at UDS there's a way to tag patches as forwarded (or not or notneeded) to Debian, separately from forwarded upstream. I don't see this immediately in the packaging guide or in DEP-3, what's the syntax?00:37
tumbleweedgeofft: Bug-Debian: ...00:38
geofftI see the Bug-Debian etc. field. I guess that's usable for the same thing00:38
tumbleweedthe debian bug is the appropriate place to forward the patch, so if you are mentioning the bug, you shuold forward the patch there too.00:39
geofftHrm. I would have interpreted Bug-Debian as "this fixes this Debian bug", not as "the pull request for this patch is this Debian bug", but there's not a huge difference00:39
tumbleweedDebian doesn't have pull requests00:40
geoffti.e. if you have a wishlist feature request that doesn't solve an existing bug. I guess you should go file a wishlist bug, yeah00:40
tumbleweed(unless the packaging is in a non-alioth-hosted VCS somewhere that does)00:40
geofftThe one thing that Bug-Debian doesn't provide is a "not needed" option00:40
tumbleweedthen you are going to mention that it's ubuntu-specific in the Description and have a Fowarded: not-needed, anyway00:41
tumbleweedthe way you forward a patch to debian is by opening a bug, with a patch...00:41
geofftoh, and I see that the implicit value for Forwarded is conditional on the Bug field.00:41
geofftokay, I can see that way of thinking00:42
geofftwell, you could have Ubuntu-specific patches that should get forwarded upstream, e.g. if Ubuntu has some random person's patchset and Debian doesn't00:42
tumbleweedfair enough, but that's non-ideal in the first place :)00:43
geofftfirst example that comes to mind is a patch to overlayfs; Debian doesn't need it forwarded but upstream does00:43
geofft(the kernel is special, admittedly, but you can imagine it happening elsewhere)00:43
infinityIf it's suitable for upstream, it's suitable for Debian (except in the case of software that just isn't *in* Debian, which renders the above conversation moot).00:44
infinityIf it's not suitable for other downstreams, it's not suitable for upstream.00:44
geofftI think the other suboptimality might be that you could use Bug-Debian to say "this fixes this Debian bug, but I didn't actually make a comment on the Debian bug saying as much"00:45
geofftbut then you're just being a jerk.00:45
broderwe mostly enforce that socially00:47
geofftyeah, fair enough, I'm convinced that Bug-Debian is what I actually care about and the exceptions are sufficiently rare00:49
=== jalcine is now known as Jacky
=== pp7 is now known as malkauns
tumbleweedbarry: yay (re launchpadlib py3k porting INPROGRESS)02:39
broderand we're getting queue control!02:45
tumbleweedorly? I missed that02:47
cjwatsonbroder: I'm only working on upload permissions so far02:47
cjwatsonBut it's a start02:48
cjwatsonHaven't quite figured out how it works when people have queue admin for only part of a queue; no doubt I'll get there ...02:48
=== scarneiro_ is now known as scarneiro
brodertumbleweed: i just saw cjwatson mark the relevant bug as in progress :)03:06
cjwatsonbroder: bug 648611 (kind of mistitled) is the bug for the queue admin side of it03:07
ubottuLaunchpad bug 648611 in Launchpad itself "ubuntu-sru either have too much or too little permission as queue admins" [High,Triaged] https://launchpad.net/bugs/64861103:07
broderoh, hmm03:07
broderi wonder why i decided that bug #914779 was exciting03:07
ubottuLaunchpad bug 914779 in Launchpad itself "Pocket maintainers cannot always upload to their pocket" [Low,In progress] https://launchpad.net/bugs/91477903:07
broderi guess that's for ubuntu-backporters should be able to upload to all of -backports?03:08
broderkind of stopped caring once i got core-dev :-P03:08
cjwatsonYe03:08
cjwatsons03:08
cjwatsonAnd other such things; ultimately it should let us get rid of the hardcoding of ubuntu-security in LP03:09
cjwatsonThough that requires a bit more considerably more complicated work03:09
cjwatsonAnyway, the DB column is the same for upload and queue admin, I just didn't want to try to do both in the same branch03:10
psusicjwatson, is there *any* part of ubuntu you don't work on? ;)03:16
superm1siretart: oh i must have missed that further in the log, i'll re-read the rest.  as for the internal copy of ffmpeg, they have some code they have forked for now that their delta doesn't make sense upstream but they need internally when used with DVB of sorts04:08
superm1so while it is possible to link to a system ffmpeg it will certainly break04:08
elkyStevenK, woot04:09
janimocjwatson, hi, is there a clean way in grub-mkconfig snippets to modify an existing line in an already generated part of the config file - to remove a boot cmdline argument to be precise?07:42
janimoto be even more precise the vthandoff=7 part should be deactivated in a custom build and currently a grub-mkconfig hook is sedding the grub.cfg.new file but that does not always work07:44
Kalidarndoes anyone know if a feature like https://www.youtube.com/watch?feature=player_detailpage&v=TRIyqLrwXVY#t=4s is planned?08:09
Kalidarn(ie dragging to the corner to get the window to lock to 1/4th of the screen)08:10
htorqueKalidarn: that's already doable with compiz (grid plugin)08:12
Kalidarnoh awesome, last time i used ubuntu you could only do 1/2 a screen like in windows 7.08:13
htorqueKalidarn: download the compizconfig-settings-manager and change the resize actions in the grid plugin: http://img.xrmb2.net/images/271735.png08:14
Kalidarnah, thankyou very much.08:15
htorquenp :)08:15
SpamapSKalidarn: be *careful*08:19
SpamapSKalidarn: ccsm can really break unity08:20
Kalidarnfair enough, i'll know what caused it to break if it does.08:21
Kalidarnmm, i wonder hwo 11/close08:45
Kalidarnoops08:45
=== Mez_ is now known as Mez
* apw just did an install and at the end it says 'downloading langpacks' and later 'installing langpacks' but if i look its actually also installing libreoffice en-toto, is that expected10:53
cjwatsonapw: suspect you might find that's an upgrade11:04
cjwatsonjanimo: in general, no - the text has already been generated but there's no guarantee that it's been flushed11:05
cjwatsonjanimo: if you don't mind changing gfxpayload as well, you could just emit 'set linux_gfx_mode=text' perhaps11:06
cjwatsonjquery-ui-themes wins the prize for today's silliest binary package names.11:14
cjwatsonI: jquery-ui-themes -> libjs-jquery-ui-theme-hot-sneaks_1.8.18+dfsg-1.11:14
cjwatsonI: jquery-ui-themes -> libjs-jquery-ui-theme-le-frog_1.8.18+dfsg-1.11:14
cjwatsonI: jquery-ui-themes -> libjs-jquery-ui-theme-swanky-purse_1.8.18+dfsg-1.11:14
janimocjwatson, would set linux_gfx_mode=text imply a non-graphics splash?11:23
cjwatsonjanimo: No, just a text-mode handover from GRUB to the kernel11:30
janimocjwatson, would this make the change to vthandoff  unnecessary?11:33
janimoI think the issue with vthandoff was it caused black screen with a poulsbo-like driver11:33
janimotseliot, ^ ?11:33
cjwatsonjanimo: 'set linux_gfx_mode=text' has two effects: it causes 'set gfxpayload=text' rather than 'set gfxpayload=keep', and it drops vt.handoff=711:35
=== yofel_ is now known as yofel
cjwatsonIt'd be great if somebody in the MIR team could review jbigkit (bug 993304); it's blocking a few things now.11:40
ubottuLaunchpad bug 993304 in jbigkit (Ubuntu) "[MIR] jbigkit" [Undecided,New] https://launchpad.net/bugs/99330411:40
apwcyphermox, hey ... i have a pile of machine on my local lan which keep disconnnecting and reconnecting to the network (its an ipv4/ipv6 dual stack), the first thing i see is NM saying device state change to failed due to ip-config-unavailable; am wondering if you have any advice as to how to debug12:21
tseliotjanimo, cjwatson: it's the gfxpayload=text part that I'm trying to avoid12:40
cjwatsontseliot: so just emitting 'set linux_gfx_mode=text' is probably what you want to do12:44
tseliotcjwatson: well I would need 'set linux_gfx_mode=keep' but how would you suggest that I do it?12:47
cjwatsonOh, sorry, misunderstood12:48
cjwatson=keep will leave you with vt.handoff=7 by default12:48
cjwatsonI assume you're trying really really hard to avoid modifying /etc/grub.d/10_linux directly, since that's easier than this hackery12:49
cjwatsonBut you *could* perhaps redefine the gfxmode function in a later grub-mkconfig hook12:49
janimoyes, probably that's the easiest12:49
cjwatsonTo just something like12:49
cjwatsonfunction gfxmode {12:49
cjwatson        set gfxpayload="$1"12:49
cjwatson}12:49
cjwatsonpitti: Am I right in thinking that ~lp_queue/manual-queue/ was only ever used for langpack uploads, and that we no longer do them that way?12:50
cjwatsonThere was a rejected set of langpack uploads in there from something like 2007, which I removed, and nothing else in evidence12:51
geser"/usr/include/i386-linux-gnu/bits/string3.h:103:1: error: inlining failed in call to always_inline 'strcpy': function not inlinable" is this an error in the package itself? or somewhere else? and how to fix it?12:56
geserhttps://launchpadlibrarian.net/104397135/buildlog_ubuntu-quantal-i386.libcitadel_8.05-1_FAILEDTOBUILD.txt.gz is the build log12:56
cyphermoxapw: i'll have a patch for you to test in a few minutes13:08
tseliotcjwatson: where would I call that function from?13:20
tseliotgfxmode13:20
cjwatsontseliot: You don't need to, the fragments generated by 10_linux do so13:23
tseliotcjwatson: but wouldn't I still get vt.handoff=7 this way?13:24
cjwatsontseliot: Not if you redefined gfxmode to not set vt_handoff to anything13:25
tseliotcjwatson: ok, I get it now. Thanks13:26
janimocjwatson, but redefine it in a hook prior to 10_linux so it takes precendence?13:28
=== rsalveti` is now known as rsalveti
=== tkamppeter_ is now known as tkamppeter
roaksoaxis merges.ubuntu.com being fully deprecated now?14:09
cjwatsonroaksoax: Not at all!14:10
cjwatsonroaksoax: It's just having hardware trouble which I've been trying to work through with IS.  It looks as though it'll need a replacement machine.14:11
roaksoaxcjwatson: ah! I see now. I was just wondering why the lists were outdated! But thanks for the clarification! :)14:11
geserroaksoax: not that I know of, I know that is has (had?) hardware issues14:12
gesercjwatson: would it be possible to add a note about it on the MoM frontpage till it got fixed?14:13
slangasekcjwatson: would it be more straightforward to move merges to the cloud somehow?14:21
xnoxslangasek: we need persistent storage for the results, but the cloud can do the merge attempts.14:22
slangasekright14:22
* xnox cloud tends to have holes in the floor, which make valuable files drop off the cloud.14:24
mdeslaurslangasek, cjwatson: merges needs to be a trusted machine, since some people rely on the tarball that is generated. Else, we need to throw the tarball away and just keep the report.14:24
slangasekmdeslaur: I didn't mean /that/ cloud ;)14:25
xnoxif rossetta is emailing me about errors in translation templates from a recent package I uploaded.... do i have to deal with them?14:26
slangasekxnox: generally no - they're usually upstream issues that you don't need to do anything about14:26
mdeslaurslangasek: just making sure :)14:26
slangasekxnox: if you *touched* the translations and are now getting errors, you should worry about it :)14:27
cjwatsongeser: good idea; done14:28
xnoxslangasek: On 2012-05-17 07:33z (2 hours 55 minutes ago), you uploaded a file with14:29
xnoxCatalan (ca) translations for grub in Ubuntu Precise package "grub2" to14:29
xnoxLaunchpad14:29
LaneyDoes mom download all sources from D + U?14:29
cjwatsonslangasek: the main difficulty, aside from sheer awkward size, is that merges needs to keep a lot of source packages around which are no longer in any current archive to serve as base revisions; not to mention archived patches and the like14:29
* xnox somehow thinks catalan speaking users just might use sudo over the next 5 years.14:29
cjwatsonxnox: I think that one is ignorable14:29
slangasekxnox: yeah... maybe that's cjwatson's problem? :) (grub2)14:29
slangasekcjwatson: right14:29
cjwatsonthere are a few things that Rosetta more or less legitimately whines about that I haven't bothered fixing because it involves liaising with upstream translators and it didn't look that important; and I think possibly also some cases where it was just being pedantic14:30
* xnox notes down "if in daubt, `maybe that's cjwatson's problem?`"14:30
cjwatsonLaney: only given series in each, but yes14:30
cjwatsonxnox: ...14:31
* xnox rips the paper apart.14:31
* Laney saw something about block storage somewhere …14:31
cjwatsonyeah, it might be doable now that canonistack does block storage, but I'm not about to move an important production service to it until others have shaken out the bugs first :)14:32
* slangasek nods :)14:32
LaneyI was thinking as an interim until the hardware is acquired14:32
xnoxmsgfmt didn't give any error...14:33
cjwatsonLaney: given its apparent I/O problems, I'm not sure I could transfer the current state to the cloud without IS help anyway14:33
* xnox is ignoring that translation file14:33
cjwatsonMoM isn't desperately cloudy right now; it could be, but if it were actually going to take advantage of the cloud properly it would need to be refactored rather a lot14:33
cjwatsonnot to say that wouldn't be a good idea14:33
Laneycjwatson: Can it not recreate its own state? Or do you think that would be too network heavy?14:34
* Laney would be willing to give the current code a spin14:34
cjwatsonIt needs to have its pool available for base revisions14:34
cjwatsonIn general it cannot recreate its own state entirely from scratch14:35
Laneyah, so it can't use snapshots / LP to download old stuff14:35
Bluefoxicyso I was thinking about hibernation14:35
BluefoxicyI'm not sure how the kernel does it.  :|14:35
cjwatsonNot at the moment, and it's possible not everything is available on either; I've certainly seen cases in the past where casey was the only thing I could find that had managed to retain something14:35
LaneyI can believe that it wouldn't get 100% coverage14:36
cjwatsonI don't know exactly what the problem is, but doing a full run causes it to reboot, so I have no confidence that I could rsync off the data; but I gather the problem is more likely to be RAM or RAID controller or something, not the disks themselves14:36
cjwatsonSo I'd rather wait for IS and continue to hassle them :)14:36
Bluefoxicyfor a rough outline my thinking is:  1)  Prepare the system (flush all buffers, finish fetching I/O from hardware, etc); 2) Stop scheduling all user applications; 3) Flush all buffers and page cache; 4) Remount read-only everything except the target for /hiberfile; 5) Flush all application RAM to /hiberfile (through LZO or LZ4); 6) extend /hiberfile to hold kernel RAM; 7) RO-mount the file system; 8) Flush kernel memory to /hiberfile14:38
Bluefoxicythat's probably completely different from what the kernel actually does :|14:39
Bluefoxicyalthough I think being able to flush to a dynamically created hibernation file and (OPTIONALLY--good for 32GB RAM and slow hard disk, not so good for SSD or SCSI-RAID etc) compressing the data as it's written out would be useful14:40
* Bluefoxicy rambles about nothing interesting14:40
xnoxBluefoxicy: what are you trying to achive?14:41
Bluefoxicyxnox:  I think it'd be nice to be able to hibernate without a swap partition, or with a swap partition that currently has too much swapped onto it to take on all of RAM14:41
highvoltageuwin 4114:42
Bluefoxicyxnox:  This is already possible--you can hibernate to a swap file; but more interesting would be swapping to a dynamically created file (i.e. not a swap file), and the potential applications for compressing the hibernation file may be interesting (compression may be slower than just writing to disk; then again disk may be too small to hold 5-10 gigs of hibernation data)14:43
BluefoxicyI think on MY system compressing would be slower than writing straight, because I have an SSD with a SandForce chipset that compresses data on-the-fly to improve throughput anyway--compressing stuff before writing it to disk actually slows my disk access14:44
xnoxBluefoxicy: i always have swap as twice my RAM, exactly because of failing to swap. Usual and rude answer is "you should be using LVM and just extend your swap partition".14:44
xnoxI don't have fancy drives.14:44
xnoxI do have full disk encryption with LUKS14:44
xnoxgood luck, sounds like a reasonable cause.14:44
Bluefoxicyxnox:  I just use zram14:44
Bluefoxicy(by the way, don't ever load the zcache module:  it doesn't actually do anything--as I understand it should work compiled it, but does not work as a module--and it completely destroys the kernel's ability to manage page cache)14:45
xnoxnice ! =)14:46
* xnox goes to blog 'how to load zcache module to make your ubuntu run faster'14:46
Bluefoxicyhttp://article.gmane.org/gmane.linux.ubuntu.devel.discuss/13766 <-- this is me14:46
* xnox wait a second!14:46
Bluefoxicyhaha14:46
xnoxmaybe i should read -discuss....14:47
Bluefoxicyseriously, when I had zcache loaded my kernel kept purging cache down to like 150MB and then struggling with all the extra disk access, with a gig of free RAM sitting around :P14:47
Bluefoxicythings got very slow14:47
slangasektremolux: hi, are you guys by chance able to marshall some SRU testing resources for this software-center update?15:23
slangasektremolux: it's a very large delta with a lot of different bugs, and per the SRU process we should be trying to get validation for each of them; I fear this SRU may stall without a concerted effort to validate, which would be unfortunate given that it fixes a lot of top crashers15:25
barrycjwatson: \o/15:34
cjwatsonwe'll see if it works for other people ;)15:34
barry:)15:34
evcyphermox: once we have this connectivity check thing in place, what are your thoughts on me backporting it to 12.04? GNetworkMonitor is causing me lots of pain.15:54
cyphermoxev: this fills me with fear15:55
evhaha15:55
evoh dear15:55
cyphermoxev: connectivity checking has already been included functionally, it's just not enabled by default15:55
cyphermoxespecially for something that will go poll a website, I much rather doing this only in quantal15:56
evokay15:56
evI might drop it anyway. libcurl will still fail, so we wont actually lose the report.15:59
tremoluxslangasek: hey! yes, I was going to give it another day or two to wait for other folks to verify, then I was going to go through any remaining ones myself so that we don't delay past the initial window16:07
tremoluxslangasek: I think we can marshall the esteemed davmor2 to help as well16:08
slangasektremolux: hurrah :)16:08
tremoluxslangasek: :D16:08
* tremolux is dying for USC to no longer have the top crasher on errors.ubuntu.com ;)16:09
micahgmdeslaur: did lintian build for you yesterday?16:32
cjwatsonmdz,pitti,kees,stgraber,soren: just found mail in the TB moderation queue about a catch-up with the CC in their meeting starting in 25 minutes ... can any of you attend?  I might be able to be at part of it but possibly not all, so it would depend where we are in the agenda16:34
cjwatsonalso the TB has been fairly quiet lately so not sure I have much to say :)16:35
stgrabercjwatson: I won't be able to make it, at least not the first half hour or so16:37
Bluefoxicyanyone know where the script is for the liveCD boot system?16:40
BluefoxicyI want a peek at it16:40
Bluefoxicyparticularly the part that mounts the /cow16:40
cjwatsoncasper16:40
Bluefoxicy(which right now is just tmpfs)16:40
cjwatsonspecifically you probably want scripts/casper there16:40
Bluefoxicyk16:41
mdzcjwatson, I can16:41
* Bluefoxicy apt-get src casper16:41
Bluefoxicycjwatson:  trivia #2:  why 'casper'?16:42
cjwatsonmdz: thanks - I'll try to sit in if I can16:42
cjwatsonBluefoxicy: I blame mdz :)16:43
Bluefoxicyit's not really analagous to Norton Ghost or anything16:43
cjwatsonit was a friendly ghost reference, because it did weird magic16:43
cjwatsonI think16:43
Bluefoxicyah16:43
mdzmore or less16:43
Bluefoxicyand Sardra would have been less appropriate because Sardra is an a-hole16:43
Bluefoxicy(A wizard did it)16:44
mdzcjwatson, the fridge calendar disagrees with the email :-/16:45
cjwatsonis that a WoW reference?  in that case, not to mention that ~nobody had heard of Sardra when casper was written16:45
cjwatsonmdz: imagine my surprise16:45
Bluefoxicycjwatson:  8 bit theater, a running gag is that a certain small boy is repeatedly subjected to horrible, horrible things as consequence of everything the main characters do16:46
mdzperhaps the CC meeting would be a good place to propose abolishing all forms of scheduling except one16:46
Bluefoxicyso he becomes the most powerful wizard in existence, returns to the beginning of the universe, and attempts to become god by commanding it and remaking the universe WITHOUT those guys16:46
Bluefoxicythis fails so he just takes to doing horrible things to them16:46
barrycjwatson: btw, you should post to u-d@ about the ubiquity port :)16:46
mdeslaurmicahg: lintian?16:47
cjwatsonbarry: oh yeah16:47
barrycjwatson: feel free to make it a veiled call for volunteers <wink>16:47
infinitycjwatson: Yes, please, include a diffstat and make sure we all feel appropriately inadequate! ;)16:47
micahgmdeslaur: yes, you sponsored it :)16:47
mdeslaurmicahg: oh, is that one of the syncs? no idea16:48
* broder perks up at mention of lintian16:48
micahgmdeslaur: umm, it didn't build :)16:48
mdeslaurmicahg: oh well16:48
infinitymicahg: Did you happen to retry it already?16:48
micahginfinity: failed locally as well :)16:49
infinitymicahg: Same ENOSPC failures?16:49
infinityOh, wait.16:49
infinity*cough*16:49
mdeslaurmicahg: would be nice if we'd get build failure emails16:49
micahgmdeslaur: there's a bug for that (syncs don't send build failure e-mails)16:50
mdeslaurmicahg: how's it failing?16:50
micahgBug #90211416:51
ubottuLaunchpad bug 902114 in Launchpad itself "When sponsoring using copyPackage, the sponsored person is not sent email from Soyuz" [Low,Triaged] https://launchpad.net/bugs/90211416:51
micahgmdeslaur: one of the tests is failing16:52
LaneyI think you mean bug #86225116:52
ubottuLaunchpad bug 862251 in Launchpad itself "Sync requester doesn't receive build failure emails" [High,Triaged] https://launchpad.net/bugs/86225116:52
mdeslaurmicahg: that bug is for the sponsored person to get the email16:52
infinityActually looked more like fakeroot failing.16:53
* infinity waits for it to fail again.16:53
Laneydoes the sponsor get the FTBFS mail for normal uploads?16:53
micahgLaney: yes, that's the bug I was looking for16:53
micahginfinity: yeah, I think you're right16:53
micahgmdeslaur: just a reminder to test build before sponsoring :)16:56
mdeslaurmicahg: yeah, I'll definitely keep away from sponsoring next time :)16:56
infinityActually, there are a ton of failures in the test suite...16:57
infinity*sigh*16:57
micahgthis seems to be a common recurrence with lintian every release16:57
infinityWell, if I'm reading the changelog correctly, this one might have dropped Ubuntu support and forced us to modify it anyway, unless the maintainer was kind enough to include an Ubuntu-specific DB for his new generic derivatives handling.16:58
* infinity hasn't grabbed the source yet to see.16:59
broderwait, huh?16:59
broderthe stuff niels did was ubuntu friendly16:59
mdeslaurmicahg, infinity: sorry about that, I didn't usually test build syncs before, but I'll be more careful16:59
broder(i.e. the purpose of that code was so that we no longer threw tags about original-maintainer)16:59
infinity    + [NT] Remove Ubuntu specific handling of distribution names.17:00
infinity      Instead replace it with a more generalized one that derivatives17:00
infinity      can reuse by extending vendor specific data files.17:00
broderright17:00
broderand the data files are in there17:00
infinitybroder: I'm not implying that those changes are unfriendly, just that they require ubuntu-specific files to be there, and I didn't check to see if they are, or if we're expected to tack them on.17:00
broderlintian.uw.o is currently running with something ~2 commits older than the release17:00
infinitybroder: Ahh, kay.17:00
broderso lintian itself works, even if the test suite doesnt'17:00
infinityThe hardening failures in the testsuite seem pretty obvious (the baseline expectation is just plain wrong for Ubuntu compilers).17:01
infinityOther failures looked a bit more confusing.17:01
brodercan you pastebin a log?17:01
infinityThe build's still running.17:01
infinityWhen it's done, it'll be in the "pastebin" called the "launchpad librarian", referenced from https://launchpad.net/ubuntu/+source/lintian/2.5.7/+build/3493637 :P17:02
broderheh17:02
micahgmdeslaur: thanks, at this point of the cycle, I think it's more about encouraging good practices than the actual failure as we'd probably have gotten it next week some time17:03
cjwatsonmdz: well, now I'm completely confused, since there's been no activity in -meeting either at the fridge time (which also == the usual obvious UK DST misinterpretation of the mail) or the mailed time17:06
mdeslaurmicahg: I didn't think building locally when syncing was a good practice. Would help if this was documented somewhere.17:06
mdzcjwatson, indeed. I just sent an email17:06
mdzI wonder if they were expecting a response and canceled it when they didn't receive one17:07
micahgmdeslaur: hmm, okay, /me wonders where syncing is documented at this point17:07
cjwatsonSyncingAndMerging hopefully17:07
cjwatsonHm, maybe not, memory like a thing with holes in17:08
Bluefoxicymkfs.ext2 is valid from casper?17:09
Bluefoxicyor is that not in $PATH at that time?17:09
melodiehi17:15
mdeslaurmicahg: ah, here's the docs: https://wiki.ubuntu.com/SyncRequestProcess17:18
zulCan an archive admin review  python-jsonschema please?17:18
mdeslaurmicahg: I'll add info about a local build to the wiki17:18
micahgmdeslaur: it's already implied, but a test build of some sort should really be done, thanks17:19
mdeslaurmicahg: implied by what?17:19
* micahg is reluctant to say required and isn't sure why17:19
micahgSometimes the sponsor requires a build log of the Debian package compiled in the Ubuntu release as proof that the new Debian version still compiles in Ubuntu.17:19
micahgthat implies a requirement of still building in ubuntu to be sync'd17:20
cjwatsonBluefoxicy: easy test: boot with break=casper-bottom, and you get a shell so you can look17:20
* Laney thought it was accepted best practice to test build all uploads (sponsored or not)17:20
Bluefoxicycjwatson:  nice17:20
mdeslaurLaney: well, I do test build uploads, but I wasn't test building syncs17:20
micahgLaney: only for certain people :)17:20
Laneynot all people follow best practice :P17:21
micahgtrue17:21
* infinity admits that he doesn't always test-build syncs, but does watch to make sure they don't fail.17:21
gesermdeslaur: any special reason for this? as syncs are similar to uploads17:21
Bluefoxicyhttp://pastebin.com/dfabCsAU17:21
broderman, i had forgotten how intense the lintian test suite is17:22
Bluefoxicycjwatson:  any way I can break into the shell BEFORE casper is executed so I can replace the script and see what happens?17:22
mdeslaurgeser: no special reason, it just didn't occur to me that syncs were supposed to be handled differently than autosyncs17:22
Bluefoxicyor do I need to build a livecd and boot it in VirtualBox to find out?17:22
Laneymy, Lintian's test suite takes rather a long time to ru17:23
Laneyn17:23
Bluefoxicy(the diff above is my first, untested attempt to tell casper to use a zram device instead of tmpfs for the /cow and copy-to-RAM, with ext2 being the filesystem of choice)17:23
Bluefoxicy(the advantage is that these are compressed and so take up less RAM)17:24
gesermdeslaur: for me syncs are a special kind of uploading, so I test-build them like I do before uploading a package17:24
mdeslaurgeser: yes, I agree I should have been doing that, and will from now on...It just didn't occur to me before17:25
cjwatsonBluefoxicy: you could just  zcat | cpio -it  the initrd and see what it contains17:27
cjwatsonBluefoxicy: oh, misread - you can break=mount IIRC17:27
* micahg hugs mdeslaur17:27
cjwatsonyeah17:28
cjwatsonthough you don't have many tools, so sometimes rebuilding is easier17:28
* mdeslaur hugs micahg back17:28
Bluefoxicyhahavvv17:29
Bluefoxicythere's no editor17:29
Bluefoxicywell that's not a problem17:30
* Bluefoxicy mounts the squashfs from the medium, uses the tools there >:D17:31
=== Ursinha` is now known as Ursinha
infinitymicahg, broder: Alright, lintian failed again on the buildds, this time without crazy fakeroot issues (that must have been some other transient oddity).17:32
mdeslaurinfinity, broder: is one of you already taking a look, or shall I poke at it?17:35
broderi am not currently taking a look17:36
cjwatsonBluefoxicy: there is sed, which is occasionally useful with a bit of creativity; but yeah, no full-screen editor17:36
broderi can take a look later if you'd like17:36
infinitymdeslaur: I was just rebuilding to see if a previous issue was reproducible, I'll leave it to you to fix the current issues. :P17:36
micahgmdeslaur: I'd suggest pushing back on AnAnt to look at it17:36
brodermight be worth pinging nthykier in #debian-qa on oftc to see if he can hel[p17:36
infinity(The hardening failures are obvious, some of the others look slightly less obvious at first glance)17:37
mdeslauryeah, I'm trying to figure out the apache2-modules failure17:37
mdeslaurhah, lintian is FTBFS on debian too17:40
mdeslaurhttp://bugs.debian.org/cgi-bin/bugreport.cgi?bug=67319817:40
ubottuDebian bug 673198 in lintian "lintian: The 2.5.7 source package is missing files (causing FTBFS)" [Serious,Open]17:40
xnoxstgraber: http://debblog.philkern.de/2012/05/lazyweb-question-how-to-avoid-leaking.html lxc ?17:40
debfxmdeslaur: <nthykier> Alternatively touch t/tests/apache2-modules-general/skip will make Lintian skip the test, which should be fine until the next upload17:40
infinitymdeslaur: Oh, that's comforting.  One more reason why Debian needs to switch to source uploads.17:41
mdeslaurdebfx: awesome, thanks17:41
micahgthat's the problem with Debian and arch:all packages, no clean build env17:41
infinitymicahg: Yeah, see, speaking of people's "best practices" for syncs above, one of the first things I do is check buildd.debian.org logs.  Which is useless for arch:all.17:42
micahginfinity: yeah, I like to do that as well17:42
micahghelps on non-x86 archs17:44
stgraberxnox: yeah, something like http://www.stgraber.org/2010/12/12/having-fun-with-containers/ (back when arkose was called sandbox). The only issue is that it needs root privileges...17:45
stgraberxnox: you could also do something similar by directly using lxc-unshare -s PID <cmd>17:45
stgraberxnox: but that'll all be a lot more secure once we have the user namespace in the kernel and these no longer require root access17:45
geserjust to realise that the build log for the arch you're interesting it doesn't exist because that's the arch the DD uploaded17:47
cjwatsonI vaguely recall hearing once that there was an interface whereby a DD could post their own build log, but of course hardly anyone does17:49
infinityYes, it's a tricky and high tech interface called "email".17:49
infinityThough, it's been so long since I was a buildd admin, I now forget where the emails get sent. :P17:50
cjwatsonYeah, I also find that I just send e-mails addressed to Santa Claus and they get where I meant them to go17:51
infinity;)17:52
cjwatsonAlternatively, "nobody remembers the right address" :-)17:52
cjwatson- [cjwatson] Port ubiquity: DONE17:53
cjwatson+ [cjwatson] Port ubiquity: INPROGRESS17:53
cjwatsonbarry: hmm?17:53
cjwatsonbarry: (also, sent that mail now)17:53
=== charles_ is now known as charles
* Bluefoxicy fixes a few bugs.18:01
Bluefoxicywell it ALMOST worked (besides that I had to cp mkfs.ext2 out of the squashfs)18:02
Bluefoxicyit's notably possible to mount the squashfs early and run /filesystem.squashfs/sbin/mkfs.ext2 ... all the libraries it needs are in /lib on the initrd18:02
cjwatsonstill best to use LD_LIBRARY_PATH when running binaries from a different fs18:02
cjwatsoneven if it happens to work at the moment - we learned that lesson in d-i a while bac18:03
cjwatsonk18:03
mdeslaurnot to self: never touch lintian again18:06
Bluefoxicycjwatson: i think it doesn't matter anyway ... tmpfs swaps, this is silly.  Just using zram as swap would accomplish the same goal methinks18:12
barrycjwatson: maybe the work items got out of sync in our browsers?18:12
barrycjwatson: thanks for the email18:13
Bluefoxicyi installed chromium, brought the universe repository down ... there's about 400MB in /cow now and it's taking up 66% of its original size (the actual file data takes up 56%)18:13
Bluefoxicyso this isn't great anyway18:13
* Bluefoxicy scraps this as wasted effort.18:13
cjwatsonbarry: ah well, fixed now18:14
* barry remembers to refresh the page before editing it18:14
dobeyhmm, multiarch breaks a symlink installed by libqt4-dev it seems18:14
cjwatsonshame there's no collision detection18:14
dobeyno idea why the symlink is there at all though18:14
barryyeah18:15
bryceh@pilot in18:26
=== udevbot changed the topic of #ubuntu-devel to: Archive open | Dev' of Ubuntu (not support or app devel) | build failures -> http://bit.ly/HaWdtw | #ubuntu for support and general discussion for hardy -> precise | #ubuntu-app-devel for app development on Ubuntu http://wiki.ubuntu.com/UbuntuDevelopment | See #ubuntu-bugs for http://bit.ly/lv8soi | Patch Pilots: bryceh
brycehslangasek, lp #1000541 is assigned to you (and you fixed it for precise-proposed) but it's in the sponsoring queue with a patch from someone else that looks ok; shall I go ahead and upload it for quantal or would that mess you up?18:27
ubottuLaunchpad bug 1000541 in ia32-libs (Ubuntu Quantal) "ia32-libs-multiarch depends on gstreamer0.10-fluendo-mp3, causing problems when installing packages from partner" [Medium,Triaged] https://launchpad.net/bugs/100054118:27
=== skaet_ is now known as skaet
brycehslangasek, I'll go ahead and send the fix for quantal18:48
slangasekbryceh: ia32-libs was going to be pocket-copied from precise-proposed to quantal18:52
slangasekI don't think there's any point in fixing separately for the two, since it's a metapackage18:52
brycehslangasek, ah, too late18:53
slangasekpfft18:53
slangasekbryceh: did you increment the version number in the changelog?  Otherwise it'll be rejected and I still win ;)18:54
brycehslangasek, 20090808ubuntu3618:55
brycehdidn't change it from the supplied patch18:55
slangasekright, that's the version number I already used for -proposed because I knew it was going to quantal18:55
slangasekso I'll pocket-copy now18:56
slangasek(done)18:56
brycehyou and your silly archive admin powers18:56
slangasek;)18:56
brycehthere's a surprisingly large amount of not valid sponsor items in the queue today :-/18:57
macoi hope the one i pre-reviewed before telling the person to submit it isn't one of them18:57
micahgwell, with 79  items, I'd hope some are valid :)18:58
bryceh0 for 4 so far19:00
stgraberbryceh: if there are some you want to see off the queue, ping me with the URL and the status you want (Work in progres / Rejected / Merged)19:00
macooh good, her package did get sponsored and is in -proposed19:00
farkerhaikuHi, I was wondering for the correct value to insert into the gtk settings.ini file for gtk-shell-shows-app-menu and gtk-shell-shows-menubar.  For gtk-shell-shows-app-menu, the documentation says to use something that evaluates to TRUE, but we've tried "true", "TRUE", and 1 with an exception in my xsession-errors (asking here due to the canonical gtk fork)19:01
brycehstgraber, yeah https://code.launchpad.net/~jeffreytremaine/ubuntu/precise/balazarbrothers/fix-829819/+merge/106110 needs off the queue; already got taken by debian and we'll just sync it in (just a typo fix)19:01
brycehstgraber, thanks19:01
stgraberbryceh: actually, just mark them as disapproved and I'll do a mass cleanup tomorrow morning during my shift, will be easier that way I guess19:03
brycehstgraber, ok sounds good19:04
farkerhaikuThe error we get is "Key file contains key 'gtk-shell-shows-menubar' which has a value that cannot be interpreted"19:05
slangasekcjwatson: hmm, seb128 pinged bug #900526 about stalled SRUs... did we intend to have any further SRU verification for !lucid?  Not sure if we should kick those back out as "not worth it", push them without further verification, or rustle up the testing resources19:06
ubottuLaunchpad bug 900526 in debian-installer-utils (Ubuntu Hardy) "d-i fails to divert initctl when upgrading packages during install" [High,Fix committed] https://launchpad.net/bugs/90052619:06
geserdoes quantal auto-sync from testing or unstable?19:16
geserbryceh: can you do a quick sponsoring of "syncpackage libmemcached"? libmemcached 1.0.6-1 fixes the FTBFS in quantal. (or do you prefer a bug for it?)19:21
farkerhaikuIf there's a better place to ask this sort of question, I'll be happy to ask there19:22
brycehgeser, a bug would be helpful yes19:28
brycehgeser, for some reason syncpackage fails to run for me.  Trying to get it sorted, but if I don't, the bug will be there for someone else to do19:29
macofarkerhaiku: maybe True19:29
macofarkerhaiku: since python likes that capitalization...19:29
farkerhaikumaco: I'll give that a shot19:30
maco(i'm just guessing)19:30
infinitybryceh: I'll sync it.19:30
infinitygeser: Don't bother with a bug.19:30
geserinfinity: thx19:30
geserinfinity: do you know if quantal auto-sync from testing or unstable?19:31
infinitygeser: Currently testing.19:31
infinityOh, hrm.  I hope you didn't want credit for that, I forgot -s19:32
infinity(synced, though)19:33
Laneybryceh: I'd appreciate it if you could process my emacs23 MP at some point during your shift too; emacs being uninstallable causes me great pain.19:34
geserno problem, no credit needed (and I doubt one sponsored upload more (or less) would make a difference during an application)19:34
stgraberLaney: I hear that vim works fine in quantal :P19:35
LaneyDon't look at the size of the diff. Launchpad knows nothing.19:35
brycehLaney, sure thing19:35
infinitystgraber: ^519:35
LaneyI actually use both, so you won't get anthing out of me :P19:35
geserinfinity: "currently"? will it switch at some point to unstable or it's not decided yet? or will it stay at testing for whole quantal?19:35
Laneyit will change19:35
infinitygeser: The plan is to switch to unstable, we're not entirely sure what will break when we do. ;)19:36
geseronly the whole universe and multiverse :)19:36
Laneyno change there19:37
infinitygeser: On, I didn't mean in the archive, but rather various launchpad UIs and bits going haywire because you can't really change derivation without fallout.19:38
infinityIt'll be a fun experiment.19:38
=== kirkland` is now known as kirkland
=== nxvl_ is now known as nxvl
brycehLaney, heh your merge broke bzr20:33
Laneyheh20:34
Laneymaybe I can rustle up a debdiff20:34
brycehLaney, awesome20:34
LaneyUDD: the future.20:35
ajmitchLaney: gold star for you!20:37
Laneybryceh: http://people.ubuntu.com/~laney/emacs23_23.4+1-3ubuntu1.debdiff.xz20:42
LaneyYes, I had to xz it.20:42
brycehwows20:43
LaneyThe ubuntu-restore-non-dfsg thing is pretty crazy20:43
slangasektwitch20:43
micahgLaney: that's against the new Debian revision?20:43
LaneyI'm sure it could be done better with multiple-orig20:43
Laneyyeah.20:43
LaneyOr moving the non-dfsg thing to main or something.20:43
micahgeek, /me adds to personal /ignore list20:44
Laneyslangasek: since you're here and we're talking about emacs, could you process bug #998460 please? :-)20:44
ubottuLaunchpad bug 998460 in emacs23-non-dfsg (Ubuntu) "Remove from Quantal and blacklist" [Undecided,Confirmed] https://launchpad.net/bugs/99846020:44
slangasekhmm20:45
slangasekpoooosssibly20:45
LaneyWell, unless you think we could move it into main20:45
slangasekis the blacklist still where it was?20:45
Laneyfor now, yeah20:45
slangasekmain, or universe?20:46
Laneyhmm20:46
Laneywhat relationship does it have?20:46
slangasekit sounds like it might be easier to just shuffle the pockets, and keep the packages in sync with Debian20:46
infinityLaney: If... Err, what vorlon said. :P20:46
infinityIf emacs23-non-dfsg really is just the same stuff that we're adding back in our diff, that seems like a silly amount of effort for us.20:47
slangasekthe only binary package it builds is emacs23-common-non-dfsg... should that be a dependency of emacs?20:47
slangasekhonestly, either way, I think that may be the simpler route20:48
slangaseki.e. I don't see any barrier to having it in main if that's where it belongs20:48
slangasekin fact since it's just a binary package split, it shouldn't need a MIR20:48
Laneyif we're OK having it in main as a Depends, that should get us to ~ where we are now20:48
Laneyand let us drop a bunch of delta20:48
infinitySounds reasonable to me.20:49
brycehdropping delta sounds nice20:49
slangasekit's just the docs, after all20:49
slangasekLaney: I'm in favor of just making it a binary dep from emacs23 and avoiding the horror deltas :)20:50
slangasekrather, from emacs23-common20:50
Laneyrock20:50
Laneydeleted debian/patches/ubuntu-restore-nondfsg-files.diff20:51
rbasakThere's a tool to filter various apt list type files, but I can't remember what it's called and I can't find it. I think it's named after grep. Does anyone know what I'm talkinga bout?21:00
LaneyProbably grep-dctrl21:01
rbasakthat's it - thanks|!21:03
brycehLaney, do you want to shoot me another .debdiff for whatever remains?21:05
Laneybryceh: yeah, 5 mins21:05
brycehLaney, great, thanks21:05
Laneylong emacs build is long21:37
=== SQLStud is now known as UrAznMaster
bdmurrayIf I'm planning on tag a bunch of bugs should I email ubuntu-devel or someone? or just have at it21:58
bdmurrayslangasek: what do you think⸮ should I email some group before tagging a ton of bugs in Launchpad22:15
Laneybryceh: http://people.ubuntu.com/~laney/emacs23_23.4+1-3ubuntu1.debdiff22:17
Laneyoff now, I'll pick up any issues in the morning22:18
Laneyo/22:18
brycehLaney, thanks cya!22:18
bryceh@pilot out22:18
=== udevbot changed the topic of #ubuntu-devel to: Archive open | Dev' of Ubuntu (not support or app devel) | build failures -> http://bit.ly/HaWdtw | #ubuntu for support and general discussion for hardy -> precise | #ubuntu-app-devel for app development on Ubuntu http://wiki.ubuntu.com/UbuntuDevelopment | See #ubuntu-bugs for http://bit.ly/lv8soi | Patch Pilots:
brycehbdmurray, 355 dupes against libx11, oh my22:25
bdmurrayand the one after that looks similar22:26
brycehbdmurray, of course the bug won't open now22:27
bdmurraybryceh: use /+text22:27
brycehyeah I know22:27
bdmurrayshould write a greasemonkey script to mark up +text22:28
brycehbdmurray, note that it says it's _asserting_ in libx11.  That (usually) means that the client is making x11 calls incorrectly22:28
brycehheh that'd be awesome22:28
bdmurrayright, I think we talked about an update-manager bug too22:29
brycehyeah so 507062 sounds like it might be a bug in cairo causing it22:31
slangasekbdmurray: hmm, I don't see any need for emailing ubuntu-devel before tagging a batch of bugs.  will you be doing the tagging as a bot account?  That might be best, so people can filter out that kind of thing22:37
bdmurrayslangasek: yes a bot.  Right, I'm not sure the email is really necessary either, I just recall doing that years ago.22:38
Bluefoxicyhmm, so here's a good question then22:38
slangasekbdmurray: personally, I wouldn't worry either way22:39
slangasekanyone who has time to worry about mass-tagging clearly has too much time to read bug mail ;)22:39
Bluefoxicytmpfs gets swapped, so there's no reason to put it on a zram device22:39
BluefoxicyMaking an ext2 on a /dev/zram2 etc does work and will get you compression, but it's pointless22:39
BluefoxicyBut what's the equivalent for block devices?22:39
bdmurrayslangasek: and doing it now it will get mixed in with other noise22:40
Bluefoxicyto my knowledge there's no dm-compress, so the prospectus for compressing a COW on a persistent device is minimal22:40
psusiBluefoxicy, btrfs supports compression... there were some patches for ext compression but they were never merged22:53
psusiBluefoxicy, it can't be done at the block layer22:53
Bluefoxicypsusi:  good point.  I'm trying to figure what's good for a persistent root.22:53
Bluefoxicyer, cow22:53
psusiBluefoxicy, wait, what's your goal?  cow or compression?22:54
* Bluefoxicy patches casper to mount the cow at /moo instead22:54
Bluefoxicypsusi:  I was just thinking like if you did a persistent root on a USB stick, it'd be huge22:54
Bluefoxicypsusi:  especially if you upgraded packages etc22:54
melodieBluefoxicy, you could remove the packages and the docs once the update done22:55
Bluefoxicybut also if you install to USB, you wind up with a situation where a 700MB CD becomes 2 gigabytes of crap22:55
melodieI have persistant "changes" directory for one distro and I manage it not too bad22:56
Bluefoxicyor... you copy filesystem.squashfs to the USB stick under a partition, use a COW partition22:56
Bluefoxicymelodie:  that's true22:56
psusiBluefoxicy, try installing to btrfs with compression enabled?22:56
melodieI also use this mode to make the changes in the usb stick before installing with the changes to hard drive22:56
melodieBluefoxicy, and usb sticks tend to be cheaper and cheaper. a 4 gb here now can be found for less than 5 euros.22:57
Bluefoxicymelodie:  true as well, but does that have a dual-channel controller and get 60MB/s transfer across the bus?22:59
melodiehow can I check it ?22:59
Bluefoxicymind you my SSD gets about 384MB/s (it's on SATA2) and is capable of about 500MB/s if I get it on a SATA3 controller22:59
melodieSSD is faster than usb 2 right ?22:59
Bluefoxicyhdparm -t I guess23:00
BluefoxicySSD is much faster :P23:00
Bluefoxicyso is HDD23:00
melodiewhat is your goal exactly ? perform a usb install for yourself or add a feature for all in a program ?23:00
Bluefoxicythe other part of this argument is that reading 5MB of compressed libraries and expanding it to 15-25MB is likely faster than reading 25MB outright from a slow USB stick that can get 9MB/s23:01
melodieI am very much interested in persistant modes on usb sticks to tell you all.23:01
Bluefoxicyalso I'm just asking questions right now23:01
Bluefoxicyanyway psusi has a point23:01
melodievery probably.23:01
Bluefoxicybtrfs now supports LZ4, which is excellent:  unlimited parallel compression/decompression, the algorithm is fast, and it has pretty good compression ratios on par with LZO23:02
Bluefoxicyparallel execution is fantastic with all these 4 core Intel i5 systems and the likely 16-32 core ARM systems we'll see in the future23:02
melodiethe interesting part for persistant, apart from the nomadic which is obvious, is the possibility to install a fully up to date and personalized (on the fly) version... at any time after a given distribution version has been made available.23:03
melodiehow much data can you put in a 700 MiB compressed iso, if using btrfs with LZ4 ?23:05
melodieLet's say current usual programs, nothing special as sounds and videos.23:05
psusinot as much as you can with squashfs23:09
psusithe problem with compression is that it is always a tradeoff between compression ratio, and random access23:09
psusithe more data you compress at once, the better compression you get, but to access some random data in the middle, you have to decompress the whole block23:10
melodiewhich in clear means safe access to data ?23:10
melodieor fast access...23:10
mdawkinsjust means more memory is used to decompress23:10
melodieok23:11
psusithat and you have to read data you don't necessarily need.23:11
melodieok. :)23:11
psusithat's why gzip gets better compression that pkzip.... pkzip compresses 32k at a time... but if you want to extrat a file from a .tar.gz that is near the end, you have to spend a long time reading and decompressing the whole thing, but zip can skip to the end and start decompressing there23:12
mdawkinsis there any discussion about the unioning fs?23:17
melodiepsusi, what you said about squashfs does not concern unionfs right ? I think they are two different things ? (aufs vs unionfs maybe ?)23:20
mdawkinsmelodie: the unioning fs is what makes the livecd interesting23:20
mdawkinsotherwise you cant do much with it except for boot and install23:21
psusimelodie, unionfs just takes two different filesystems ( or more ) and combines them into one... it has nothing to do with compression23:21
psusiaufs is the same thing pretty much23:22
mdawkinsi have been playing with overlayfs23:22
melodiewhich one is used by default at Ubuntu ?23:22
mdawkinsi thot they were moving to overlayfs after speaking to dev23:23
mdawkinsbut that was awhile ago23:23
melodiegood night23:40

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