| -queuebot:#ubuntu-release- New binary: google-android-installers [amd64] (noble-proposed/multiverse) [1710437545-3] (no packageset) | 00:04 | |
| vorlon | removing cmake-data from release | 00:06 |
|---|---|---|
| vorlon | wow, a tremendous number of packages uninstallable in the kde stack due to -data packages. I'll give the uploads a bit more time to continue and let e.g. riscv64 builds start, then work on removing them all | 00:09 |
| vorlon | setting amd64 build failures to be retried hourly | 00:10 |
| vorlon | gtk4 ftbfs on armhf + i386, that's unhelpful | 00:11 |
| vorlon | well, starting to see build failures already because of deps on python3-distutils | 00:20 |
| vorlon | ah possibly needs python3-defaults rebuilt first for amd64 | 00:20 |
| vorlon | unfortunately that didn't get frontloaded | 00:21 |
| jbicha | vorlon: is it too late to let mutter & gnome-shell migrate before rebuilding? just mentioning because I believe Ubuntu Desktop was uninstallable yesterday because some things migrated without those 2 | 00:31 |
| vorlon | jbicha: they are not migratable, the amd64 binaries were removed | 00:45 |
| vorlon | we are not leaving any amd64 binaries in the archive that were built while the compromised liblzma was on the system | 00:46 |
| vorlon | which is the actual reason for ubuntu-desktop being uninstallable | 00:50 |
| jbicha | I mean there are things like yaru-theme-gnome-shell that got migrated in the big time_t migration before the xz mitigation, but ok, it'll get fixed later | 01:30 |
| === chris14_ is now known as chris14 | ||
| === alucardromero6 is now known as alucardromero | ||
| === arraybolt3_ is now known as arraybolt3 | ||
| vorlon | well that's no good https://launchpad.net/ubuntu/+source/gnutls28/3.8.3-1.1ubuntu3/+build/27979041 | 06:27 |
| vorlon | wgrant: where can I see the buildd chroot tarball currently used by launchpad? | 06:30 |
| vorlon | new apt conflicts currently-build libnettle8. temporarily removing apt from noble-proposed to sort out | 06:52 |
| vorlon | hmph bad bugginess then in my script to order the builds, it thought nettle was unseeded and produced no libs :/ | 06:54 |
| vorlon | because I had to regenerate the list after the copies to noble-updates and noble-updates lacks the Task fields apparently, sigh | 06:56 |
| vorlon | so much for trying to order uploads smartly | 06:58 |
| vorlon | well, redoing the list against snapshot.u.c for the remainder | 07:12 |
| doko | vorlon: cmake : Depends: cmake-data (= 3.28.3-1build5) but it is not going to be installed. is the action to remove cmake-data in the release pocket? otoh I don't see a new cmake upload yet | 08:02 |
| juliank | vorlon: not really a bug in gst-plugins-good1.0, but stupid solver doing stupid things. I think u-r-u fares better, but I need to do analysis on the workaround after all the holidays | 08:22 |
| doko | all builds with chroot failures retried | 08:54 |
| doko | didn't help: now at: apt : Depends: libgnutls30 (>= 3.7.5) but it is not going to be installed | 09:29 |
| -queuebot:#ubuntu-release- New sync: python-ring-doorbell (noble-proposed/primary) [0.8.7-1] | 11:46 | |
| -queuebot:#ubuntu-release- New: accepted google-android-installers [amd64] (noble-proposed) [1710437545-3] | 11:56 | |
| -queuebot:#ubuntu-release- New: accepted python-ring-doorbell [sync] (noble-proposed) [0.8.7-1] | 11:56 | |
| doko | a dist-upgrade from a mantic to a noble chroot succeeds, however apt is kept back. apt-utils is removed | 12:34 |
| doko | juliank: iirc, you proposed changing the apt dependency on libapt-pkg6.0 to change to libapt-pkg6.0t64 | libapt-pkg6.0 ? | 12:35 |
| juliank | doko: no, bumping the soname to libapt-pkg6.1 is something I can do, then it's co-installable | 12:36 |
| juliank | doko: what you're looking for here would be adding more Conflicts to apt | 12:37 |
| juliank | Such that it Conflicts with all its pre-t64 dependencies | 12:37 |
| juliank | To force upgrading them | 12:37 |
| juliank | But it's not always clear if that works | 12:38 |
| doko | so soname bump looks safer? | 12:38 |
| juliank | Collect a dump with -o Dir::Log::Solver=path and I'll look at it Tuesday before the other logs | 12:38 |
| juliank | Haven't tried the soname bump yet either tbh | 12:38 |
| doko | so a mantic -> noble dist-upgrade with these options? | 12:39 |
| juliank | It should affect less the solver and more the installation ordering | 12:39 |
| juliank | ubuntu-release-upgrader should work a lot better | 12:39 |
| doko | ... but not used on the buildds | 12:39 |
| juliank | It has the logic to collect conflicts/break+replaces against old packages | 12:39 |
| juliank | But why try to upgrade a mantic buildd | 12:40 |
| juliank | Just debootstrap noble from the February snapshot | 12:40 |
| doko | because that is the buildd chroot that we are starting from. | 12:40 |
| doko | or even earlier. it hasn't been updated yet | 12:40 |
| juliank | It's possibly something else like gnutls, gcrypt, db5.3-especially for apt-utils removal | 12:44 |
| doko | juliank: that option doesn't write any more output for me | 12:47 |
| juliank | Did you give it a relative path? Then it writes to /car/log/apt | 12:47 |
| ahasenack | component-mismatches is a bit nutty lately | 12:48 |
| juliank | I can't do much today or tomorrow I really gotta focus on recovery or I'll be hopelessly overstressed | 12:49 |
| juliank | I was starting to get feverish from the late t64 nights past week | 12:49 |
| doko | still wants to upgrade libdb5.3, not installing libdb5.3t64 | 12:50 |
| doko | ahasenack: that will stay like this until the rebuilds are done | 12:50 |
| juliank | The solver may change behaviour once libdb5.3 is no longer available | 12:51 |
| juliank | Obsolete packages it cares about less | 12:51 |
| juliank | But if you want to experiment locally, whacking a Conflicts: libdb5.3 (<< t64) to apt-utils in the Packages files may show you if that works | 12:52 |
| doko | ahh, I think I have the problem. the new apt upload still built with the old db5.3, and apt-utils still depends on libdb5.3 | 12:54 |
| doko | vorlon: ^^^ | 12:54 |
| vorlon | doko: I woke up this morning realizing I needed to also get gnutls rebuilt and apt built against it | 15:14 |
| vorlon | because new apt has a versioned conflicts against old libnettle8; but just getting new nettle built doesn't help when it now builds libnettle8t64 and that's not what things depend on | 15:15 |
| doko | vorlon: gnutls is not directly needed by apt | 15:15 |
| vorlon | apt depends on libgnutls30 | 15:16 |
| vorlon | so new gnutls28 needs built, then another apt uploaded to build against it | 15:16 |
| vorlon | hmm we may have to rebootstrap apt actually | 15:16 |
| doko | but that requires an installable cmake | 15:16 |
| vorlon | by temporarily dropping the gnutls dep again, to make it possible to install another apt's build-deps in a chroot | 15:17 |
| vorlon | doko: cmake-data is not in the release pocket now, did that not fix it? | 15:18 |
| vorlon | oh, we shouldn't have to rebootstrap apt because libgnutls30t64 provides: libgnutls30 | 15:18 |
| vorlon | so, I've just removed apt again, and have retriggered gnutls28 to build once that removal has published | 15:18 |
| doko | a ppa build with only -updates activated failed with: | 15:21 |
| doko | The following packages have unmet dependencies: | 15:21 |
| doko | cmake : Depends: libcurl4t64 (>= 7.16.2) but it is not installable | 15:21 |
| doko | E: Unable to correct problems, you have held broken packages. | 15:21 |
| vorlon | where do you have a cmake that depends on libcurl4t64? | 15:22 |
| doko | an i386 it failed with | 15:22 |
| doko | The following packages have unmet dependencies: | 15:22 |
| doko | cmake : Depends: cmake-data (= 3.28.3-1build5) but it is not going to be installed | 15:22 |
| doko | that's I'm trying to figure out ... | 15:22 |
| vorlon | 3.28.3-1 doesn't, it was uploaded before | 15:22 |
| doko | yes, but there's 3.28.3-1build5 | 15:23 |
| vorlon | not on amd64 | 15:23 |
| vorlon | and if you're trying to fix a build on !amd64 it has to wait | 15:23 |
| vorlon | I do see that cmake didn't get uploaded last night because my gpg agent timed out, sigh. Will work through that this morning | 15:26 |
| doko | cmake 3.28.3-1build5 depends on libcurl4t64 | 15:29 |
| doko | gnutls28 3.8.3-1.1ubuntu3 is not yet built on amd64 | 15:30 |
| doko | cmake depends on cmake-data (= ${source:Version}), so you need to remove the amd64 binary as well | 15:31 |
| doko | <vorlon> so new gnutls28 needs built, then another apt uploaded to build against it | 15:33 |
| doko | ^^ include the new db5.3 in this build as well, for apt-utils | 15:33 |
| -queuebot:#ubuntu-release- New sync: speech-dispatcher-contrib (noble-proposed/primary) [0.12.0~rc2-3] | 16:34 | |
| vorlon | doko: is db5.3 already built? | 16:52 |
| vorlon | checking | 16:52 |
| vorlon | it is. so yeah, the next apt would of course include this | 16:55 |
| vorlon | and my gpg agent timed out again overnight, so starting more uploads shortl.y | 16:55 |
| vorlon | only ~800 uploaded last night :/ | 16:56 |
| vorlon | moved cmake up to the top of the list too | 16:57 |
| vorlon | doko: "need to remove the amd64 binary as well" the amd64 binary *is removed*, along with all amd64 binaries built since Feb 25 | 16:58 |
| vorlon | so I don't know what you're seeing | 16:58 |
| vorlon | cmake 3.28.3-1 and cmake-data 3.28.3-1 are in noble-updates | 16:58 |
| vorlon | gnutls28 build successful https://launchpad.net/ubuntu/+source/gnutls28/3.8.3-1.1ubuntu3 | 16:58 |
| vorlon | reuploading apt | 16:58 |
| vorlon | doko: https://launchpad.net/ubuntu/+source/cmake/3.28.3-1build6/ - looking at the !amd64 build failures | 17:02 |
| vorlon | ah libwacom | 17:02 |
| vorlon | which needs libgudev | 17:02 |
| vorlon | oops, didn't check for libgnutls30t64 publication before uploading apt. cancelling build | 17:07 |
| vorlon | manually reuploading coreutils,cinnamon,binfmt-support,webkit2gtk which already had newer versions upload but that were deleted | 17:10 |
| vorlon | ah not binfmt-support, that's a case of someone else uploading a non-NCR version | 17:12 |
| vorlon | same with webkit2gtk. So just coreutils,cinnamon at the moment | 17:12 |
| vorlon | libgudev-1.0-dev : Depends: gir1.2-gudev-1.0 (= 238-5build1) but 1:238-5build1 is to be installed | 17:14 |
| vorlon | ... | 17:14 |
| vorlon | how did a package get in here with busted epoch handling | 17:14 |
| vorlon | jbicha: ^ I'm not going to look at this at the moment; don't know if you're around but maybe you want to try to sort it | 17:15 |
| mitya57 | vorlon: maybe you need to re-apply this diff? http://launchpadlibrarian.net/721504992/libgudev_238-3_238-3ubuntu2.diff.gz | 17:51 |
| mitya57 | Not sure why it was dropped when syncing -5, jbicha should know better | 17:52 |
| -queuebot:#ubuntu-release- New: accepted speech-dispatcher-contrib [sync] (noble-proposed) [0.12.0~rc2-3] | 17:59 | |
| doko | mitya57: just uploaded | 18:23 |
| -queuebot:#ubuntu-release- New binary: fwupd-snap [amd64] (noble-proposed/universe) [1.9.x-0ubuntu2] (no packageset) | 18:25 | |
| vorlon | libgnutls30t64 published, building apt now | 19:06 |
| vorlon | queuing up some triggers for libwacom/cmake | 19:11 |
| doko | libwacom built, no triggers needed | 19:47 |
| doko | jbicha: while you're on vacation, please consider fixing your vacation uploads in your vacation time ;-P | 19:48 |
| -queuebot:#ubuntu-release- New binary: libxeddsa [amd64] (noble-proposed/universe) [2.0.0+20221109.06a02c8-2.1build1] (no packageset) | 20:11 | |
| -queuebot:#ubuntu-release- New binary: ipp-crypto [amd64] (noble-proposed/universe) [2021.10.0-0ubuntu3] (no packageset) | 20:19 | |
| -queuebot:#ubuntu-release- New binary: speech-dispatcher-contrib [amd64] (noble-proposed/multiverse) [0.12.0~rc2-3] (i386-whitelist) | 20:27 | |
| -queuebot:#ubuntu-release- New: accepted fwupd-snap [amd64] (noble-proposed) [1.9.x-0ubuntu2] | 20:32 | |
| -queuebot:#ubuntu-release- New: accepted libxeddsa [amd64] (noble-proposed) [2.0.0+20221109.06a02c8-2.1build1] | 20:32 | |
| -queuebot:#ubuntu-release- New: accepted ipp-crypto [amd64] (noble-proposed) [2021.10.0-0ubuntu3] | 20:32 | |
| -queuebot:#ubuntu-release- New: accepted speech-dispatcher-contrib [amd64] (noble-proposed) [0.12.0~rc2-3] | 20:32 | |
| === guiverc2 is now known as guiverc | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!