[01:02] slangasek: No, but ubuntu-touch-meta and ubuntu-core-meta are the only metas (because of the whole "sdk libraries" thing) that directly depend on libraries, so it makes them much harder to deal with. [01:03] slangasek: Other metas are mostly a non-issue. You need to keep them installable, but that can be done by transitioning a few of their deps at a time, rather than ALL of them. [01:09] infinity: can you please new gunicorn? [01:09] https://launchpad.net/ubuntu/wily/+queue?queue_state=0&queue_text=gunicorn [01:14] off to sleep. good night. there is too much things to new. [01:14] ok thanks. [05:42] xnox, why? https://people.debian.org/~doko/logs/gcc5-20150813/archive-gcc-08-13-2015/gnuradio_3.7.8-1_unstable_gcc5.log you are delaying the transition. we will continue to offload your packages [06:12] infinity, slangasek: all arm64 buildds down. contacted is, but no reply yet === rsalveti_ is now known as rsalveti === nhandler_ is now known as nhandler === locutus_ is now known as LocutusOfBorg1 === hyperair is now known as Guest47721 === Guest47721 is now known as hypera1r === hypera1r is now known as Guest53587 === hypera1r_ is now known as hyperair [08:32] * Laney screams at gnuradio [09:08] Laney: still looking at gnuradio? [09:56] * slangasek cleans up octave Breaks: to match the actual Ubuntu versions [09:57] so what's still wrong with gnuradio? [10:07] now needs a transition [10:07] har, and software-center is rendered uninstallable because debtags in wily-proposed has dropped the python2 support [10:07] Laney: a new one? [10:08] it was already transitioning from 3.7.5 to 3.7.8 [10:08] ah [10:08] xnox: nack on gnuradio 3.7.8-1ubuntu3, this needs reverted [10:12] xnox: for packages that are mixed up in the big transition, it's worth 5 minutes of discussion before upload to save 5 hours of compile time [10:14] xnox: also, to summarize the current status: all first-level libraries that are affected by the ABI change have already been transitioned in wily; so unless you run into a package build failure due to a library in the archive still having old symbols (which will basically mean this library didn't get detected in the Debian rebuild because it FTBFS), there shouldn't be any more transitions needed [10:20] if anyone cares to look at it, pocl has an interesting build failure - rebuild needed for llvm3.5, but build segfaults with a stack trace pointing into llvm3.5 itself [10:22] (not sure why llvm-toolchain-3.5 is still in main in wily, given that mesa has moved to 3.6 - but if mesa's moved I'm not so worried about it that I'll block the 3.5 transition) [10:27] removing pcl binaries on ppc64el, the source looks incompatible with current eigen3 [10:36] so I can't see any reason that the autohinter is including debtags in the hint, considering nothing in update_excuses depends on it and it makes stuff uninstallable. It may be a mis-hint that the autohinter will figure out when we get closer, or it may be something that'll require a manual hint to avoid [10:36] but I am amused that debtags has dropped python2 support before software-center has switched to python3 [10:37] uwsgi FTBFS in some not immediately obvious how to fix way [10:37] looks like yade needs fixing/removing too [10:38] yade was on my list already, I don't know why it didn't make it to removal [10:38] slangasek: ok. [10:38] Laney: you at the venue? [10:38] yeah [10:39] the smaller talk room [10:39] oh actually, yade dropped of the list for this run [10:39] slangasek: the thing that tripped me up, is that it's still red in boost tracker, yet it clearly has been built against boost 1.58. [10:39] * xnox traces. [10:41] xnox: interesting; I would blame the tracker for still getting the wrong answer when there are different binary names between wily and wily-proposed, except gnuradio 3.7.8 did already transition to wily (but has to get rebuilt again and transition again with the group, because uhd) [10:41] xnox: maybe it's NBS binaries? looking now [10:41] xnox: it is not NBS binaries [10:42] anyway, at a bit of a stopping point so wandering to the debconf venue now [10:42] ok [10:43] Laney: ah I remember why I hadn't gotten around to removing yade yet though, it's because it looked suspiciously like a vtk6 bug rather than a yade bug [10:46] I'm hoping it got disentangled [10:46] slangasek: OK to sync new libsidplayfp? [10:46] looks like upstream bumped SONAME [10:46] (blocks https://launchpad.net/ubuntu/+source/sidplayfp/1.4.0-1/+build/7793291) [10:48] * Laney does it [12:05] rebuild plasma-mediacenter [12:06] +ing [12:50] Laney: I have no idea :) [12:55] I did it anyway :P [13:13] * cyphermox feels audacious [14:04] Laney: right, so the answer is that libsidplay has added new uninstallables to the transition (audacious-plugins) which were otherwise ready to go, so you get to fix that :P [14:06] and for questions of the general form "can I sync new version of $package_included_in_the_big_autohint that introduces a new ABI transition", the answer is "not until the hint finishes" ;) [14:20] Riddell: why does plasma-workspace in wily-proposed conflict against the latest version of kio-extras present in wily-proposed? plasma-workspace was uploaded 6 days ago, is a new kio-extras coming? [14:23] Riddell: note that this is on the short list of < 6 remaining blocking issues for the main g++5 transition, and as this is a versioned breaks it doesn't look like something we can straightforwardly upload a fix to plasma-workspace for; so if we don't have a new kio-extras in -proposed by the time the hint is ready to go in (which I believe will be EOD today), we will probably need to break kubunt [14:23] u-meta in wily to let the transition through [14:39] Laney: which library transition entangles uwsgi? I don't see any uploads of it to wily-proposed [14:39] Laney: (an upload to wily-proposed would let us all look at the same build failure) [14:40] ok, libgloox [14:40] (uploading) [14:46] right, uwsgi build failure reproduced in the archive now; someone who likes both ruby and GNU make can debug that one [14:46] meanwhile, package removed from wily [14:47] the next p-m run should start to look interesting [14:47] Laney: I've assumed that you are going to do the no-change rebuilds for new libsidplayfp (now that it's past new). If not please let me know [14:52] ok, the debtags problem reveals itself - the old version is in C and depends on libept, the new version drops python2 support [14:59] slangasek: It blocked some other package, and the transition was small - yes, will rebuild [14:59] ty [15:25] anyone already looking at wusgi [15:25] no I am [15:26] * Laney watches him get up and walk out [15:26] * Laney creepy [15:26] Laney: ? [15:27] Building uwsgi now [15:27] Only a crappy fix though, if it works [15:27] ok [15:27] but we have multiple supported ruby versions which it seems breaks this [15:27] doesn't seem to be much else left [15:28] debtags [15:28] AFAICS [15:28] (remove that from proposed?) [15:34] debtags needs a fix; I just talked to enrico had he told me mvo has some insight on this [15:35] s/had/and/ [15:35] The version in wily-release too? [15:36] yes, per above old debtags depends on libept which is an affected C++ lib [15:36] and oh btw I think the hint just succeeded [15:36] oops [15:37] oops because whatever uninstallables someone had left in wily that are now resolved were uninteresting leaf packages, and they've been swapped for um ubuntu-desktop and friends [15:37] so I hope this upgrade I've just started succeeds before the mirrors update? [15:38] I mean that we can reupload a buildX of the last version which had python-debtagshw [15:38] If that helps [15:38] don't see that it worked..? [15:38] * amd64: audacious, audacious-dbg, audacious-plugins, audacious-plugins-dbg, audtty, autoradio, debtags, goplay, kubuntu-desktop, kubuntu-full, lubuntu-desktop, packagesearch, qmmp-plugin-projectm, remuco-audacious, wmauda [15:39] sometimes bits get nibbled off though [15:41] Laney: oh, I see - right it didn't succeed at all it was an invalid hint because someone changed the version of qtdeclarative-opensource-src! [15:42] well I guess I can drop my hint now anyway in favor of the autohint now that we know what's up with debtags [15:44] changed the version of> or possibly that bit got in, as you say [16:07] libsidplayfp has NBS in wily-proposed to clean up [16:11] debtags uploaded [16:11] should be it AFAICS [16:11] please to delete libsidplayfp3v5 [16:12] Laney: done [16:12] kubuntu-desktop will still be uninstallable, because of the kio-extras issue [16:13] what are these new mozc uninstallables in the latest run? [16:15] new upload, needs to clear excuses [16:17] uim-mozc was NBS in wily-proposed, also removed now [16:18] if nothing else changes, I guess the next run will be it aside from forcing kubuntu-desktop [16:26] where is this kio-extras 4:5.3.2+git20150710.0210? [19:22] The lubuntu desktop live build failed [19:37] ianorlin: I can check, maybe there is something I can do [19:38] it looked to fail build the livefs [19:39] we're mid g++ 5 transition, lots of stuff is failing across the board, not just lubuntu [19:39] in this case the specific failure is [19:39] blueman : Depends: bluez (>= 4.61) but it is not going to be installed [19:39] bluez-alsa : Depends: bluez but it is not going to be installed [19:39] but it's probably just uncoinstallable for real at the moment ... [19:40] well, these should have been to -release already [19:40] oh [19:40] yeah but that probably means that something else in the install set depends on non-v5 libraries [19:48] bluez-alsa is dead. [19:48] I removed it from seeds, but it needs meta updates. [19:48] Someone needs to upload all the metas that include it. [19:48] Which is all the desktops except mate, cause I fixed theirs. [19:48] I can do that. [19:49] Not sure about the blueman->bluez thing, though. [19:49] lubuntu-core [19:49] Oh, that'll be the same failure. [19:49] Because bluez C/R bluez-alsa. [19:50] yes [19:50] So rebuilding all the metas will fix that. [19:50] I'll grab 'em all and upload. [19:50] ok [19:50] * cyphermox is still too slow at this debugging race [19:51] ah ok I grabbed an alternate and it is almost done installing [19:51] cyphermox: Thanks for adding the C/R, BTW. [19:51] infinity: yeah, apparently I'm sure I did it with that previous upload, but someone it didn't get through [19:52] *somehow [19:52] Heh. [19:52] Editors are fickle. [19:52] not that much :) [19:53] I see some stuff appeared in that update_output of doom. [19:53] cjwatson: Does germinate respect apt's proxy configs? [19:53] * infinity guesses no. [19:53] Would make these updates a tad faster. [19:58] * infinity notes that a number of these haven't been updated since vivid... [19:58] And mythbuntu, not since utopic. Awesome. [19:59] Oh well, I needed something to do while waiting on laundry before my flight. [20:17] cyphermox: we are not so far to FeatureFreeze, what about Network Manager 1.0 in wily? [20:25] infinity: It uses Python's urllib.request - should respect the proxy environment variables, but not apt's config [20:30] cjwatson: I probably live in a fantasy world, but I'm of the opinion that anything downloading from Debian archives should respect the system apt proxy configs. [20:30] cjwatson: (I know, I know, patches welcome) [20:35] It's probably not hopelessly difficult with python-apt. [20:36] I don't immediately have time though [20:38] cjwatson: I was just going to parse `apt-config dump`, but switching from urllib to python-apt might be a better long-term thing, dunno. [20:42] ari-tczew: NM is no longer my primary responsibility, so it's up to free time. thanks for reminding me, I'll get at least the git migration started right now [20:44] cyphermox: so who is the next one? [20:44] ari-tczew: unknown, possibly "awe_", or you, if you're interested [20:45] I'd think I would do 1.x and leave the package in a nice enough state for somebody else to take over [20:45] (although the 1.x release shouldn't be very hard to do) [20:45] cyphermox: would be nice, but I think I haven't enough knowledge [20:46] where are you all now? youth hostel, or the hotel? [20:46] doko: I thought you had left already, I'm back to my room [20:46] infinity: Doesn't need to switch to python-apt for the retrieval, just use apt.Config [20:46] I have no idea where the others are [20:46] or whatever it is [20:49] ari-tczew: knowledge comes from experience, we all start with relatively 0 knowledge about package *p [20:50] cyphermox: you're right [20:51] ari-tczew: maybe we should move this to #u-devel [20:51] cyphermox, mvo, Laney, and me are at the hotel bar [20:51] ah! [20:51] cyphermox: +1 [20:52] well, I guess I'll have to pass for tonight :) [20:52] Alright, metas all uploaded, now to finish packing and call a cab... [20:53] cjwatson, infinity: anything to look at for the big migration? [20:57] doko: I've been away for a week, I know nothing [21:00] doko: update-output-helper, unblock the megablob? [21:00] doko: that's the last info I had [21:01] there isn't much left there