[00:00] ScottK: gotcha (i'm still using bzr, but the branch hasn't been updated yet, which is why i asked). i may not get to it until tomorrow [00:00] ScottK: err.. uhm.. there were a number of other changes too that somehow got missed there. [00:00] barry: dget https://launchpad.net/ubuntu/+archive/primary/+files/cheetah_2.4.3-0ubuntu1.dsc also works. [00:01] SpamapS: I just uploaded what barry gave me, so please get it sorted out with him. [00:01] We've leap-frogged the debian package [00:01] Which is OK since it gets us ~OK with 2.7. [00:01] SpamapS: we'll have to get this back into debian experimental [00:01] (modulo whatever blew up the current build) [00:02] ScottK: barry: since you're into python, what do you think about wxwidgets in main? python-numpy now uses python-matplotlib to generate the docs (So it FTBFS ATM), so matplotlib has several universe build deps, but wxwidgets accounts for most of them [00:02] Just minor stuff that was fixed in the debian 2.4.2.1 package [00:03] micahg: isn't there still some controversy about python-numpy support especially w.r.t. py2.7 [00:03] er, python-numpy version [00:03] barry: http://kitterman.com/kubuntu/cheetah.debc is what my build looked like. [00:04] barry: re numpy/matplotlib - We ought to decide sooner rather than later if we'll split the package or promote all the depends. [00:05] SpamapS: re "minor stuff" work it out with barry and I'll upload the fixed version when you both have it ready. [00:05] barry: idk much about the package, just happened to be browsing through FTBFS and saw this [00:06] barry: I think 1.5.x has support though [00:06] barry: one way to get this right is to make 2.4.3-0ubuntu2 from the 2.4.2.1-1ubuntu1 that I merged on bug 663343 (sans the pyversions limitation) [00:06] Launchpad bug 663343 in cheetah (Ubuntu) "Please merge cheetah 2.4.2.1-1 (main) from Debian unstable (main)" [Undecided,Confirmed] https://launchpad.net/bugs/663343 [00:07] barry: and then send that back to debian for experimental. [00:07] SpamapS: i'll take a look at that tomorrow. unfortunately, gotta run [00:09] barry: np will throw the bug at you when its ready [00:13] barry and SpamapS: The debian/watch file also needs to be updated to point at http://pypi.python.org/pypi/Cheetah/2.4.3. [00:20] ScottK: IIRC, pypi watches don't need to point to a particular version [00:20] SpamapS: Yes. We don't want to point to just the particular version, that was meant to be exemplary and not precise. [00:23] ScottK: the shebangs in /usr/bin/* for cheetah come out as /usr/bin/python2.7 .. I doubt thats intentional. [00:23] Probably not. [00:23] especially since it breaks without python2.7 installed [00:24] Yep. [00:24] Depends: python (<< 2.7), python (>= 2.6), python-support (>= 0.90.0), libc6 (>= 2.2.5) [00:25] is that a bug in pysupport? [00:27] Yes. Or more correctly the change to support python2.7 as a supported version was probably incomplete. [00:29] I'm thinking that setup.py should be called in such a way where it sets the executable to /usr/bin/python, no? [00:30] There's some option for that -externally-managed or so. [00:31] --install-platlib=/usr/lib/python$buildver/site-packages/ --prefix=/usr --no-compile -O0 --single-version-externally-managed; \ === solid_liquid is now known as solid_liq [00:33] its being called that way.. but clearly didn't work. :( [00:33] Sounds right, except we want to install in dist-packages [00:33] Throw --install-layout=deb in there somewhere. [00:35] thats just from the build log.. :-/ [00:35] guessing cdbs is doing something naughty [00:35] actually, hmm.. [00:35] DEB_PYTHON_INSTALL_ARGS_ALL += --single-version-externally-managed [00:35] * SpamapS tries adding it right there.. [00:36] crap, its right there in the rules file [00:37] there's an attempt to fix it [00:37] which looks really hacky [00:41] ScottK: --install-layout=deb doesn't do it.. but I think I have a workaround. Like barry, I'm out of time too. :-P [00:42] OK. Tomorrow then. [02:00] hardy's ffmpeg is FTBFSing on a hardy host, with unintelligible C datatype mismatch errors. [02:00] Where would I find Ubuntu's equivalent of buildd logs, so I can check when *ubuntu* last successfully built hardy ffmpeg? [02:02] http://launchpad.net/ubuntu/+source/ffmpeg [02:02] Thanks. [02:02] build logs are linked in the hardy section when you expand it [02:03] (I'm trying to build it with DEB_BUILD_OPTS=risky for a customer who wants ffmpeg with lame, but I can't even get it to build normally.) === jussi01_ is now known as Guest48460 [02:57] has easycam been developed for meerkat? [02:58] I had no luck with an ordinary hardy VM, but I rolled a pbuilder chroot and that is building ffmpeg just fine [02:59] I speculate that my hardy VM had an optional build dep installed, and ffmpeg's ./configure was set to enable it if found, and it doesn't actually work with that version of the build dep. === Guest71947 is now known as kklimonda_ === soren_ is now known as soren === vila_ is now known as vila === Guest48460 is now known as jussi [08:09] Good morning [08:13] good morning! [08:14] hey dholbach [08:15] hey pitti === almaisan-away is now known as al-maisan === dholbach_ is now known as dholbach === debfx_ is now known as debfx [08:41] tumbleweed, what do you think about adding merge proposal info to the sponsoring overview (like in the last column of https://code.edge.launchpad.net/loco-directory/+activereviews for example)? [08:42] probably best to put that info in the status column [08:43] yeah, sounds good [08:43] I'll file a bug and see what I can do about it :) === delan is now known as Guest15172 === Guest15172 is now known as delan_ [08:47] dholbach: is the sponsoring overview page going to stay? As I hope to finally finish the splitting of code and template for it. [08:47] geser, I hope that a lot of it can go into Harvest instead [08:48] which uses Django [08:50] what's the process for getting a package accepted into the official main/universe if it's already packaged in a PPA? [08:50] 'i've had a look at the wiki, but that's more about the wishlist needs-packaging bugs (mine are already packaged) [08:50] delan_, There's lots of different ways. The basic rule is that you want to have two members of ~ubuntu-dev say it's packaged well (because we all make mistakes) [08:51] If the initial packager isn't already a member of ~ubuntu-dev, usually the second reviewer uploads it. [08:51] how can I contact the team? [08:52] There's no reliable way to contact everyone. What kind of package is it? Maybe one of the flavour teams would want to help. [08:52] one is mathtext, a simple c-based shell-like program that transforms text into various 'styles' using Unicode [08:53] the other is getbooru, a php-based downloader that can scrape images from a website once a profile of regexes is written [08:55] delan_, Hrm. Neither of those strikes me as being tightly integrated with any of the flavours. You might ask the MOTU if one of them would review it (ask in #ubuntu-motu). You could also upload to REVU, but I'm not sure how many people check that regularly. [08:56] please tell me, what is REVU? [08:57] ah, i see [08:57] https://wiki.ubuntu.com/MOTU/Packages/REVU [08:57] i'll probably talk to the motu first, as you said [08:57] thanks for that ;) [08:58] Be aware that a large number of MOTU don't really like more packages that don't have a maintainer, and may advise you to maintain it in Debian and let Ubuntu sync it. [08:59] my problem with that is (i'm not sure how to fix this, though it must be simple) [08:59] in the debian/changelog, I could enter 'unstable' and let it build on debian [08:59] but if I try to upload that to a PPA, it won't work [08:59] it will only work with an ubuntu version as the name, maverick [09:00] Oh, sure. The assumption seems to be that once you get it into normal distributions, you won't need it in the PPA anymore. [09:00] hey slomo [09:01] And I doubt anyone is going to fuss too much about the changelog entry for review: most folk consider that a trivial change. [09:01] hi dholbach :) [09:02] hm, okay then. [09:02] brb === Lutin_ is now known as Lutin [09:30] pitti: can you please moderate my voting announcement mail to ubuntu-devel-announce? Thanks [09:31] geser: done [09:35] tumbleweed, maybe you can review my ubuntu-sponsoring merge proposal? :) [09:38] lol, /me looks [09:38] geser: err I got a german ballot [09:39] geser, 2 nominees? [09:39] damn, why did CIVS sent out german ballots. I didn't pick any language. Or did it took by browser language. [09:39] dholbach: yes, only 2 :( [09:40] thanks geser for organising this [09:42] tumbleweed: but you still manage to vote? [09:42] geser: The site itself was English [09:42] geser: yes, I understand enough german to see what it was, and the subject line was english. [09:43] And the vote description was English. And there was only one link that could possibly have mattered [09:43] I feared the voting page was mostly in German too [09:44] geser, beware of seb128 - he marks all German mails as spam [09:44] an option for members with private e-mail addresses is to mail launchpadusername@ubuntu.com - AFAIK that should work for almost-everyone [09:46] dholbach: yay, looks good - I'll test it later today [09:47] tumbleweed, ok, just follow up on the merge proposal and I merge it when you're happy [09:48] tumbleweed, It doesn't: lots of folks have differences between LP username and @ubuntu.com addresses, for complicated reasons. [09:57] hmm, I thought those were the minority (and that they tended to have their lp username, too) - I also assume some of them will loop [10:01] tumbleweed, hard to say. The first few hundred @ubuntu.com addresses were given out before LP integration. [10:02] And lots of folks have changed LP names, but that doesn't automatically change @ubuntu.com alias (or didn't, for a very long time) === amitk is now known as amitk-afk [11:02] haha === tkamppeter__ is now known as tkamppeter === 52AACHNJT is now known as csurbhi-afk [11:40] pitti, slangasek, hi === amitk-afk is now known as amitk === jjohansen is now known as jj-afk === dendrobates is now known as dendro-afk [12:33] ajmitch: When you upload boost, would you also please make sure to add -py27 for boost-python? [12:39] pitti: I think clamav in lucid-proposed is ready to go now: https://bugs.launchpad.net/ubuntu/+source/clamav/+bug/653738/comments/6 [12:39] Launchpad bug 653738 in clamav (Ubuntu Lucid) "Microversion update SRU for clamav in Lucid" [High,Fix committed] [12:40] ScottK: ok, thanks === ivoks_ is now known as ivoks [13:33] yello [13:33] After giving up on #ubuntu [13:33] i came here to look for hhelp [13:33] anyone having probs with BCM43** drivers and ad hoc? [13:34] this isn't the place for support, sorry. [13:35] #ubuntu doesn't have real support, rather slackers with no idea whatsoever [13:36] You might try #ubuntu-cc (where cc is your country code) to ask in your LoCo [13:36] Or file a question on launchpad. [13:36] hmmm [13:36] Eximius, dissing people who spend a lot of time helping others won't increase your chances elsewhere [13:37] I'm developing a app to use BCM43** drivers for ad hoc [13:37] :D [13:37] I'm not dissing [13:37] That might be #ubuntu-app-devel, but I very much doubt anyone there can help with the problem, although they may be able to help with app creation. [13:37] well, "slackers with no idea whatsover" is dissing in my book [13:38] rite, so "no idea whatsoever" might be a little not true [13:38] ... [13:41] Is there some reason that people use cdbs now that dh 7 exists? Seems like it just confuses things. :-P [13:42] SpamapS: There are some reasons why people *have* to use cdbs [13:42] for example, in python packages [13:42] though dh7 can handle that [13:42] SpamapS, AFAIK at least in desktop land we have a couple of extra things that cdbs just sorts out (language-packs, etc.) [13:42] dholbach: yes, you're right [13:42] ah [13:42] err dh7 is much nicer for python packages [13:42] the gnome-*.mk files [13:42] for python modules its just confusing me [13:42] right, dh7 gets them right [13:43] although some people use it for python+autoconf packages (which is vaguely relevant) [13:43] hmm, I cannot say much about python [13:43] yeah, CDBS screws up shebangs (I must actually test my patch for that) [13:43] * bilalakhtar gets pinged the moment someone says 'cdbs' [13:43] but as for GNOME packages, there are some CDBS includes which help [13:44] tumbleweed: yeah I think thats a bug in python-distutils.mk [13:44] no? [13:44] SpamapS: yes and no. dh7 works around it by building with /usr/bin/python last (after all the versioned ones) [13:44] but these days setup.py can take a --executable option to force a specific shebang (in many cases) [13:44] tumbleweed, replied :) [13:45] I worked around it with DEB_PYTHON_BUILD_ARGS += --executable=/usr/bin/python [13:45] but now I'm getting leftover files in /usr/lib/python2.7/site-packages [13:45] its like dh_pysupport doesn't even see them. [13:45] coolbhavi is also facing similar problems [13:45] in his mail to ubuntu-motu [13:46] ugh.. I am loathe to subscribe to yet another mailing list [13:46] bilalakhtar: There aren't any reasons why people have to use CDBS. [13:46] ScottK: do you know what needs to be changed to fix FTBFS with python 2.7 and files being installed in site-packages? need the python packaging tools need to get updated for python2.7 or the package itself? [13:46] I'm on 37 already [13:46] geser: I don't, but barry was going to look into it today. [13:47] Does natty use python 2.7 by default rather than 2.6 ? [13:47] ok, then I wait on barry [13:47] bilalakhtar: not yet no [13:47] bilalakhtar: No. It's an additional supported version for modules and extensions, but not default. [13:47] bilalakhtar: but it's already part of python-dev package [13:47] :) [13:47] * ScottK hands geser an "all". [13:48] right [13:49] dholbach: replied [13:49] tumbleweed, super [13:50] tumbleweed, which vote summary page are you referring to? [13:50] (just making sure I don't misunderstand :-)) [13:52] dholbach: err, +activereviews and the summary table at the top of +merge [14:02] Ubuntu 4.10 was released 6 years ago today [14:02] on week 42! [14:02] that means that 42 was built into the Ubuntu computer since the beginning. [14:02] who maintain hall-of-fame ? [14:02] I hope the project doesn't get destroyed before it figures out the question. [14:03] ari-tczew, it's slightly unmaintained, but there's a session at UDS to revive it - what's your question about it? [14:04] dholbach: Interviews is not updated. Last interview was reffer to Rhonda. [14:04] ari-tczew, ah yes [14:06] pitti, slangasek, hi [14:06] hello tkamppeter [14:07] pitti, slangasek, it is about bug 494141. It seems that the upstartification of CUPS is not enough, but also the upstart config of Samba needs to be modified. [14:07] Launchpad bug 494141 in cups (Ubuntu) "CUPS starts after SAMBA; printers are not available (convert cups to upstart)" [Undecided,Fix released] https://launchpad.net/bugs/494141 [14:08] pitti, slangasek, and the fact that smbd re-reads the CUPS config every 750 sec seems to be wrong. [14:09] tkamppeter: but cups' upstart script already reloads samba [14:10] pitti, and how do the complaiunts of the users come, like comment #32 and #34? [14:10] geser: You have been uploading package multiboot for quite some time. Can I go ahead with a merge? [14:11] tkamppeter: then perhaps reloading samba isn't sufficient? I don't know [14:11] hey pitti - would you mind sponsoring a pango SRU for me please? [14:11] (for bug 655707) [14:11] Launchpad bug 655707 in pango1.0 (Ubuntu) "Firefox crashes opening pages that use webfonts " [High,Triaged] https://launchpad.net/bugs/655707 [14:12] bilalakhtar: I wouldn't count two uploads within 15 min "for some time" :) but sure, you can merge it. [14:12] thanks === 92AABEL1U is now known as ximion [14:16] pitti, perhaps "kill -HUP" has not the desired effect on smbd. Perhaps it needs to be like a "restart smbd" or so. [14:16] chrisccoulson_: the attached branch is already in natty, and I don't see a debdiff? [14:17] pitti - yeah, after i pushed my change to bzr, robert did a natty upload [14:17] pitti - it's rev 7 that needs to go in to maverick - do you want me to branch off that? [14:17] chrisccoulson_: if you could prepare a debdiff or a branch, with the necessary changelog bug ref and all that, that'd be great [14:17] chrisccoulson_: (sorry, working a bit "under the gun" today) [14:17] pitti - ok, no problem :) [14:18] i'll do that in a bit [14:18] . o O { and why isn't pango1.0 in the desktop set? } [14:18] pitti - it might be worth backporting this specific fix also to lucid [14:18] not sure why it's not in the desktop set ;) [14:34] lool: hi! now that python2.7 is a supported version in natty, python-support needs to be aware of that (some FTBFS have been reported already). Do you plan to look at it? === bilalakhtar_ is now known as bilalakhtar [14:37] DktrKranz: IIRC barry was looking at it. [14:38] ScottK: ah, good. Thanks [14:40] ScottK: Would you think we should implement the changes I committed in Debian for Ubuntu? [14:40] lool: I haven't checked the details to have an opinion. [14:41] * ScottK is really hoping barry is up for worrying about it. [14:42] barry: I committed some changes to the python-support SVN to read the supported and old versions from debian_defaults; POX and Np237 weren't too sure about that; I personally think that's ok, but it might be that it causes this list to be odd at some times [14:59] cjwatson: I'm not sure, but I'm afraid, that MoM is not updated anymore. [15:16] ari-tczew: yeah, it appears to be locked since 18th [15:22] so i was looking at this message: https://lists.ubuntu.com/archives/ubuntu-motu/2010-October/006931.html and james_w and others suggested to install pkgbinarymangler to see the failure reproduced locally. yesterday i was successfully building cheetah locally but the buildd failed. turns out to be the exact same failures. but the meta issue is: why isn't pkgbinarymangler a builddep for python packages since its sanity check seems [15:22] critical (and somewhat magical)? [15:24] barry, we don't want to modify every package for that [15:24] we could e.g. make python depend on it, but then it would put the package in the default install, which we don't want [15:24] we could have pbuilder and other tools automatically install it in the chroot [15:24] ari-tczew, cjwatson: I killed the hang process, mom should recover with the next cron run [15:25] james_w: those are downsides, for sure. we'll probably be touching most python packages anyway to transition to dh_python2. i suppose we could do it then, *if* that was a direction we wanted to go in [15:26] but having pbuilder and other tools install it might be a better way to go [15:26] pkgbinarymangler is ubuntu-only - so it can't be a bulid-dep for python packages [15:26] tumbleweed: i see [15:27] yeah, it should probably be installed when variant=buildd (although I know nothing aobut debootstrap) [15:27] another thought: that particular check is appropriate for both debian and ubuntu, so maybe that check should live somewhere else? dh_python2 or related? [15:27] barry, that would be good [15:28] barry: sounds like it should be in dh_python2, yes. In fact dh_python2 should just put them in the right place - that's what it does for /usr/local/... AFAIK [15:28] tumbleweed: yes, dh_python2 will (or should :) just put them in the right place [15:28] so maybe it's a moot point after that transition is complete [15:41] barry: There are other reasons pkgbinarymangler can cause a build to fail, so it's generally a best practice to have it installed in your build chroot (I'll confess I forgot to add it to mine after I made it for natty). [15:43] ScottK, it's the second time I see this come up today... would it make sense to have pkgbinarymangler added to extra packages that pbuilder should install to a chroot (though that would only work for ubuntu?) [15:43] ari-tczew, fixed [15:43] cyphermox: It only applies to Ubuntu buildds, so not necessarily. === bjf[afk] is now known as bjf [15:53] shtylman: whoah, fast grapevine! [15:56] Keybuk: your own doing... shouldn't use twitter :p [15:56] I haven't actually said anything specific on Twitter yet [15:56] hey Keybuk, how are you? [15:56] pitti: hey, good thanks - you? [15:56] Keybuk: fine -- wrapping up my last week @ oem [15:57] ScottK: do you agree that the python check it does will not be necessary once everything's transitioned to dh_python2? if so, then i will just add that as a recommendation to the wiki page i use to build out the chroots [15:57] pitti: looking forwards to getting back to the distro or will you miss oem? [15:57] he misses distro surely! [15:57] or, iow, it would be okay not to install it by default [15:57] Keybuk: little bit of both, but I'm definitively looking forward to some natty work [15:57] hi kees. would appreciate your help with bug 663294 if you get some free time :) [15:57] Launchpad bug 663294 in gcc-4.5 (Ubuntu) "Firefox built with gcc-4.5 is a non-starter on i386" [Undecided,Confirmed] https://launchpad.net/bugs/663294 [15:57] Keybuk: lets say I have good analytical skills [15:57] and working on stuff that *I* actually use :) [15:57] Keybuk, question is will *you* get to the distro at some point :P [15:57] ogra_ac: hmm? [15:57] Keybuk, youre such a rare guest recently [15:58] heh, to IRC? I'm usually on Jabber : [15:58] ah [15:58] IRC is full of users who want to rant at me for things I write in changelogs and stuff [15:58] secret communication channels ;) [15:59] but mostly been coding, so IRC is kinda distracting for that [15:59] Keybuk: weren't you referring to yourself in that changelog? [15:59] yip [15:59] I thought it was just some clever wordplay. [16:00] jcastro: I don't remember - it's quite probably [16:01] barry: In what century do you think the transition will finish? [16:01] I'd just add it to the wiki page though. [16:01] ScottK: the same one we remove python-support and python-central in [16:02] I agree with that. [16:02] Although century was probably overly pessimistic. I should have said decade. [16:03] ScottK: given that this decade technically ends in about 2.5 months, yeah :) [16:03] It won't be this one. [16:03] no, it won't [16:03] but i'm mildly optimistic it will be the next one :) [16:04] Right, but you're the optimist in the room. [16:04] it's a dirty job but someone's gotta do it === plars` is now known as plars === dendro-afk is now known as dendrobates === diwic is now known as diwic_afk [16:38] ajmitch: Transitioning packages off of boost1.40 is now blocked on boost-python1.42 having -py27 support ... [16:42] cjwatson, slangasek: do you have a minute to review postgresql-common in hardy/lucid? [16:58] ogra_ac: please reupload alsa-utils SRU with -v to include previous SRU [16:58] that didn't go to -updates yet === sebner_ is now known as sebner [17:00] pitti, oh, i need to do that for every update ? ok, no prob [17:00] ogra_ac: only if you put another upload on top of a -proposed one [17:00] k [17:00] ogra_ac: once it goes to -updates, it's not required any more [17:00] i. e. we always need -v [17:00] oki [17:00] i never did multiple uploads to proposed before ... [17:01] thanks [17:01] pitti: I think the current gnome-web-photo SRU should be accepted [17:01] pitti: we won't have a better fix for a couple months [17:01] micahg: so that one will work then? [17:02] the bug trail sounded like there were some doubts [17:02] but sure, I can accept it [17:02] pitti: yes, it'll work, chrisccoulson_ was just saying that they app itself should connect differently, but that type of change seems more risky for an SRU [17:02] okay [17:02] *the [17:02] pitti: if we have changes that involve the debian directory and we're dealing with a package that uses 3.0 (quilt) format, shall we modify the files in debian/ directly or shall we create a patch that does the same thing? [17:03] tseliot: directly, please [17:03] pitti: ah, this could be the reason why quilt is giving me so many problems. Thanks === dendrobates is now known as dendro-afk [17:15] barry: good morning.. are you around? I've been working on trying to get cheetah to build and work w/ python 2.7 properly... [17:19] * geser guesses that every IRC ping barry gets today it about python 2.7 related FTBFS :) [17:20] haha [17:20] you reap what you sow [17:23] cheetah at least builds, but it borks things up quite a bit. [17:23] can somebody moderate my mail on u-d-a? === jj-afk is now known as jjohansen [17:26] SpamapS: that's a common error in the build logs, I guess barry will fix the tools to cope with python2.7 too. I guess you just need to wait for that fix. [17:26] geser: Yeah, I'm hoping I can offer some help in some way. [17:27] * SpamapS is on an SRU/merge/backports mission in between reading blueprints. === deryck is now known as deryck[lunch] [17:28] chrisccoulson_: sure, I'll take a look [17:29] kees - thanks [17:30] i'm just building mozilla-central with -fstack-protector now [17:32] chrisccoulson_: looks like a glibc bug [17:32] * kees continues to read [17:32] kees, OOI, what makes you think glibc? [17:32] mvo, around/ [17:33] chrisccoulson_: hadn't gotten far enough, just saw the top level pthread stuff. [17:34] kees - ah, yeah, there's a bit more analysis below that ;) [17:34] does firefox overload malloc somehow? [17:34] anyway, I'll shut up until I read all the way through it :) [17:34] kees - yeah, it provides it's own malloc implementation [17:34] the stock glibc malloc suffers too much from fragmentation === beuno is now known as beuno-lunch === dendro-afk is now known as dendrobates [17:49] chrisccoulson_: huh? [17:49] Keybuk, was that in response to my last comment? [17:50] yes [17:50] I'm curious as to why this *matters* [17:52] Keybuk, when mozilla did a lot of work to optimise the memory footprint of firefox before 3.0, they found a lot of memory was being wasted due to fragmentation, which ends up with memory not really being released back to the system [17:53] sure, but it doesn't matter [17:53] Linux assumes no process releases its memory [17:53] firefox manages to consume so much it actually hits swap [17:53] which will be unrelated to that fragmentation [17:54] sounds more like they're blaming something else for their own failure [17:56] I do find it scary that it uses its own malloc; glibc's is actually pretty safe. [18:00] anyone used the python-apt ? i was hoping to use it to download and inspect archives from a different arch than the running arch. [18:00] but I didn't see a way to do so. === deryck[lunch] is now known as deryck [18:10] smoser: I always use the LP API to do archive validation [18:12] kees, well, this particular itch was converting binary package -> source package [18:12] to which i was told launchpad was not capable === dendrobates is now known as dendro-afk [18:12] cjwatson, whats the normal way for enabling bootstrap chroots? I have a patch to enable natty in Lucid debootstrap. Shall I upload it? [18:12] smoser: well, that's just because multiple sources can produce a binary package [18:13] (other than the screen scraping that "works for me" right now) from https://launchpad.net/ubuntu/// [18:13] smoser: but yeah, it's a fair point. [18:13] open a bug! :) [18:13] I can't search by CVEs in the API either. ;) [18:13] well, i supposed that is true. but i have binary package, version, and know it came from one of main universe updates or security [18:14] and i can't get what i want. [18:14] when I did the seed analysis, I'd pull down the Packages files and parse them. :( [18:14] yeah, which is what python-apt would do for me [18:14] but i dont see how i can tell it to pull from something other than arch=(uname -m) [18:15] just trying to re-invent as little as i can [18:16] mvo, ^^ do you know if i can tell python-apt to pull cache for a different arch than current ? === al-maisan is now known as almaisan-away [18:20] mvo: wrt bug #587186 I wanted to ask, what did you end up checking for? [18:20] Launchpad bug 587186 in update-manager (Ubuntu Natty) "libc6 upgrade fails: illegal instruction" [Medium,Fix committed] https://launchpad.net/bugs/587186 [18:33] pitti: still around? how do I get something onto the sync-from-debian-blacklist? [18:33] I don't want python-pyftpdlib coming in unless it clears its current hurdle of WAY too many CVEs open. :) [18:34] kees: need to run now, sorry; any archive admin can do that [18:35] tgardner: The usual way is just to backport debootstrap from the development release (or in this case from Maverick) [18:35] ScottK, I have the patch developed from a previous example. I just wanted to make sure I wasn't stepping on any toes [18:36] tgardner: It should be a no change backport that any Canonical archive-admin can just do. [18:36] jdstrand: why hello. :) got a moment for a blacklist addition? === beuno-lunch is now known as beuno [18:40] chrisccoulson_: what happens if you turn off the malloc replacement logic? [18:42] chrisccoulson_: you just did a test build with -fstack-protector or with -fno-stack-protector? [18:42] kees - i haven't tried turning off the internal malloc, but i guess it will work [18:43] i did a build with -fstack-protector, and it works fine [18:44] how were you building where it didn't get -fstack-protector automatically? [18:44] kees: whatcha need? [18:44] ok. so i'm not real familiar on the process for this, but currently natty python-paramiko depends on python-crypto (>= 2.1.0-2). but 2.0.1+dfsg1-4ubuntu2 is current. [18:44] kees: and hi! === johanbr_ is now known as johanbr [18:45] do i need to file a "sync from debian" request for python-crypto ? [18:45] jdstrand: hi! :) can you blacklist python-pyftpdlib so it doesn't enter natty from Debian? It has a stack of CVEs, and I want to make sure they're addressed before we get that into Ubuntu (it's very new in Debian) [18:45] oh. actually i see its already done: https://bugs.launchpad.net/ubuntu/+source/python-crypto/+bug/662883 [18:46] smoser: yes, set "apt.apt_pkg.config.set("Apt::Architecture","armel")" at the start (before you init your apt.Cache()" [18:46] Launchpad bug 662883 in python-crypto (Ubuntu) "Sync python-crypto 2.1.0-2 (main) from Debian unstable (main)" [Wishlist,New] [18:46] mvo, awesome. thank you. [18:46] kees - i'm just building a stock checkout from mozilla-central (make -f client.mk with a minimal set of build options in mozconfig) [18:47] Q-FUNK: hold on a sec, I can look for the commit diff [18:47] chrisccoulson_: but our compiler forces -fstack-protector ... [18:47] smoser: you know apt.Cache(rootdir="some-dir") too, right? [18:47] oh, ok. i didn't realise that [18:47] mvo, yeah, thats what i was using/abusing [18:47] chrisccoulson_: perhaps it's the -fPIE -pie bits from hardening-wrapper? [18:47] smoser: make creating multiple "apt-roots" really easy [18:47] smoser: oki [18:47] kees - yeah, possibly. will try with just those now [18:47] chrisccoulson_: so, just so I understand, which builds have worked, and which have failed? [18:48] chrisccoulson_: you updating to xulrunner2 ? [18:48] kees - building our ubuntu packaging branch with our gcc-4.5, and also debians gcc - all fail [18:48] building stock mozilla-central with both gcc versions - both ok [18:48] chrisccoulson_: okay, but a stock build worked? sounds like PIE [18:49] building our packaging without DEB_BUILD_HARDENING also works with both gcc versions [18:49] yup. [18:49] so that means it's either PIE or BIND_NOW [18:49] Keybuk: Are you serious on leaving Canonical? [18:50] chrisccoulson_: let me check something and see if you can easily eliminate BIND_NOW as a variable without doing a full recompile... [18:50] kees - i'll need to do a recompile anyway, i've borked my build environment [18:50] chrisccoulson_: hah, whoops [18:51] doko_, can you please re-evaluate 662883 ? [18:51]