[03:16] xnox: why is libboost-python1.67.0 1.67.0-16 using the same package name as 1.67.0-13ubuntu2 despite containing none of the same .so's? [03:17] xnox: correction: containing a partial set of the same .so's [03:18] xnox: this is basically an unmanaged transition that is currently blocking on autopkgtests that correctly fail due to the library deps having been broken with no indication in the packaging [05:14] -queuebot:#ubuntu-release- New binary: codec2 [i386] (focal-proposed/universe) [0.9.2-2] (i386-whitelist, kubuntu) [05:14] -queuebot:#ubuntu-release- New binary: codec2 [s390x] (focal-proposed/universe) [0.9.2-2] (i386-whitelist, kubuntu) [05:15] -queuebot:#ubuntu-release- New binary: codec2 [ppc64el] (focal-proposed/universe) [0.9.2-2] (i386-whitelist, kubuntu) [05:16] -queuebot:#ubuntu-release- New binary: codec2 [amd64] (focal-proposed/universe) [0.9.2-2] (i386-whitelist, kubuntu) [05:21] -queuebot:#ubuntu-release- New binary: codec2 [arm64] (focal-proposed/universe) [0.9.2-2] (i386-whitelist, kubuntu) [05:21] -queuebot:#ubuntu-release- New binary: codec2 [armhf] (focal-proposed/universe) [0.9.2-2] (i386-whitelist, kubuntu) [05:33] -queuebot:#ubuntu-release- New: accepted codec2 [amd64] (focal-proposed) [0.9.2-2] [05:33] -queuebot:#ubuntu-release- New: accepted codec2 [armhf] (focal-proposed) [0.9.2-2] [05:33] -queuebot:#ubuntu-release- New: accepted codec2 [ppc64el] (focal-proposed) [0.9.2-2] [05:33] -queuebot:#ubuntu-release- New: accepted guava-mini [amd64] (focal-proposed) [0.1.2-1] [05:33] -queuebot:#ubuntu-release- New: accepted codec2 [arm64] (focal-proposed) [0.9.2-2] [05:33] -queuebot:#ubuntu-release- New: accepted codec2 [s390x] (focal-proposed) [0.9.2-2] [05:34] -queuebot:#ubuntu-release- New: accepted codec2 [i386] (focal-proposed) [0.9.2-2] [06:05] -queuebot:#ubuntu-release- Packageset: Removed python-numpy from i386-whitelist in focal [08:26] vorlon: all of the failing packages are RC and are due to be removed in debian and ubuntu [08:27] vorlon: it is easier to go through this pain, without the pain of bumping to boost1.72 [08:27] vorlon: and it is correct that (a) upstream finally came up with stable cross-os/distro python sonames (b) all other ones are dropped [08:30] vorlon: to get boost through the next steps are for yade & openblas to migrate [08:30] but openblas still has regressions =( [08:31] and illegal instruction in https://objectstorage.prodstack4-5.canonical.com/v1/AUTH_77e2ada1e7a84929a74ba3b87153c0ac/autopkgtest-focal/focal/amd64/c/cp2k/20191219_014105_aef5c@/log.gz sounds odd [11:06] -queuebot:#ubuntu-release- New binary: spymemcached [amd64] (focal-proposed/universe) [2.12.3+dfsg-2] (no packageset) [11:14] -queuebot:#ubuntu-release- New binary: rust-js-sys [amd64] (focal-proposed/universe) [0.3.28-1] (no packageset) [11:14] -queuebot:#ubuntu-release- New binary: rust-js-sys [ppc64el] (focal-proposed/universe) [0.3.28-1] (no packageset) [11:17] -queuebot:#ubuntu-release- New binary: rust-js-sys [arm64] (focal-proposed/universe) [0.3.28-1] (no packageset) [11:17] -queuebot:#ubuntu-release- New binary: rust-js-sys [armhf] (focal-proposed/universe) [0.3.28-1] (no packageset) [11:29] -queuebot:#ubuntu-release- New binary: rust-js-sys [s390x] (focal-proposed/universe) [0.3.28-1] (no packageset) [14:20] xnox: all of the failing packages are RC> that doesn't change the fact that boost-python is RC buggy by dropping libraries without changing package name and breaking revdeps on upgrade [14:21] xnox: boost1.67 was in disco and eoan; there needs to be upgrade handling [16:58] -queuebot:#ubuntu-release- New: accepted rust-js-sys [s390x] (focal-proposed) [0.3.28-1] [16:59] -queuebot:#ubuntu-release- New: accepted rust-js-sys [amd64] (focal-proposed) [0.3.28-1] [16:59] -queuebot:#ubuntu-release- New: accepted rust-js-sys [armhf] (focal-proposed) [0.3.28-1] [16:59] -queuebot:#ubuntu-release- New: accepted spymemcached [amd64] (focal-proposed) [2.12.3+dfsg-2] [16:59] -queuebot:#ubuntu-release- New: accepted rust-js-sys [arm64] (focal-proposed) [0.3.28-1] [16:59] -queuebot:#ubuntu-release- New: accepted rust-js-sys [ppc64el] (focal-proposed) [0.3.28-1] [17:59] -queuebot:#ubuntu-release- New binary: otb [amd64] (focal-proposed/universe) [7.0.0+dfsg-2] (no packageset) === locutus_ is now known as LocutusOfBorg [18:13] -queuebot:#ubuntu-release- Unapproved: hedgewars (bionic-proposed/universe) [0.9.25-5~ubuntu18.04.1 => 1.0.0-4~ubuntu1.18.04.1] (no packageset) [18:13] -queuebot:#ubuntu-release- Unapproved: hedgewars (disco-proposed/universe) [0.9.25-5build1 => 1.0.0-4~ubuntu1.19.04.1] (no packageset) [19:12] vorlon, can you please help make dkms migrate? (one single i386 failure) [19:18] locutus_: hinted [19:18] lovely thanks [19:18] locutus_: do you know why an i386 test was triggered for nvidia-390 but not for any of the others? [19:18] nope... [19:19] because the driver is still there? https://launchpad.net/ubuntu/+source/nvidia-graphics-drivers-390/390.132-0ubuntu2 [19:19] vorlon, did you run the removal on restricted archive? [19:21] the nvidia packages are in the whitelist [19:59] vorlon: libraries are whitelisted, but not the dkms module [19:59] nvidia-dkms-440 | 440.44-0ubuntu1 | focal/restricted | amd64 [19:59] nvidia-dkms-390 | 390.132-0ubuntu2 | focal/restricted | amd64, armhf, i386 [20:02] xnox: the whitelist is by source package. So does nvidia-graphics-drivers-390 need updated to drop the i386 dkms package? [20:03] xnox: btw what do you think about having boost1.67 carry a delta to drop numpy support on i386? [20:04] (rather than me doing something to matplotlib to make the only i386 binary packages go away on that arch) [20:14] vorlon: i'm happy to make such delta. I think i can even keep the package in sync. [20:15] what's wrong with matplotlib [20:15] xnox: in sync> are you going to make boost's debian/control be generated? since the build-deps must vary based on target distro [20:16] vorlon: we used to have that anyway, so trivial to ressurect for when mpi was in universe and was strictly out of b-d [20:16] ok [20:16] vorlon: i think -dkms packages should be dropped on i386 => however some of them are arch:all [20:17] wrong with matplotlib> it's a math package, it's on i386, and it's currently in the whitelist only because of the arch: all python-matplotlib-doc as a build-dep of numpy which leaves uninstallable arch-dep binaries in the archive [20:18] with further germinate hacking, I suppose we could change it to not add the sources of arch: all packages to the sources output [20:18] but for this corner case, it'd be nice to just excise numpy as a whole [20:18] heh [20:19] and boost is whitelisted, and pulls in numpy too [20:19] yes [20:19] vorlon: which libraries of boost are actually needed on i386? i'm all for making boost more of a bump on i386. [20:19] ie. are there any other boost things i can stop building on i386? [20:19] and I also found that I still have a bug, because frei0r should be pulling in gavl as a build-dependency and is not :/ [20:20] xnox: https://people.canonical.com/~ubuntu-archive/germinate-output/i386.focal/i386+build-depends there are multiple reverse-build-deps of boost in the set (mir, firebird3.0, ceph) [20:20] vorlon: i merged your germinate branch into my foreign arch one, and i get test failure in your test case =/ [20:20] also [20:21] incorrectly %s" % "included" if sense == "no-" else "omitted" [20:21] ah, whe [20:21] e [20:21] "incorrectly %s" % "included" if sense == "no-" else "omitted" [20:21] is not what you think it is [20:21] oops :) [20:21] ("incorrectly %s" % "included") if sense == "no-" else "omitted" [20:21] is how python does it [20:22] well, since you are merged, i had to catch up in my PR. Need to work out what am I doing wrong =) [20:22] i think it is me, because i do fake provides [20:22] which make arch-all things look arch-any and things go astray [20:24] ah that's why frei0r build-deps aren't included, yeesh [20:24] we only pull in frei0r-plugins-dev which is Arch: all. It does not depend on the arch-dep frei0r-plugins. [20:26] so that's similar to the matplotlib case :/ [20:28] I think I'll just treat that as a manual exclude from the packageset for now [20:50] -queuebot:#ubuntu-release- Packageset: Removed frei0r from i386-whitelist in focal [20:56] vorlon: we drop many python2 packages, which will remain installed on users systems on upgrade [20:57] vorlon: many of which will remain installed, with broken / non-working compiled extension deps. [20:57] xnox: broken/non-working> that should not be the case [20:57] care to elaborate? [20:59] hm [20:59] py2 packages get dropped from the archive, yet remain installed on the users system [20:59] and become incompatible with other arch:all packages they depend on [21:00] ..... of which the set of brokeness is zero?! [21:00] other arch: all packages that aren't called python-foo? [21:00] yes, like plugins to the apps that embed python interpret [21:00] i.e. gimp [21:00] which changes from py2 to py3 [21:01] if they become incompatible with their dependencies, those dependencies should be declaring a Breaks: or Conflicts: and the revdep should be removed on upgrade [21:01] within the archive -> sure [21:01] and we can't do anything for externals right? [21:01] right [21:02] somehow i feel boost1.67-python should always provide boost1.67-python37 and make shlibs generate the dependencies on boost1.67-python37 and not just boost1.67-python [21:02] because we are going to drop python37 abi of boost for example [21:04] vorlon: so should I now add lots of breaks on boost1.67-python package of all the python2 boost using packages since disco|stable ? [21:05] xnox: I think you should do a one-time rename of the boost1.67-python package so that it doesn't satisfy the existing revdeps [21:05] and then rebuild all of those to pick up the new package name [21:05] the same as any other library transition [21:07] vorlon: and you still want me to drop numpy boost on i386? [21:07] xnox: yes please :) [21:08] I can drop matplotlib with a packageset-level hack, but I'd rather drop them both [21:13] vorlon: we should rename pyhon2 packages to python-legacy =) [23:11] -queuebot:#ubuntu-release- New binary: pcapy [s390x] (focal-proposed/universe) [0.11.4-1] (no packageset) [23:11] -queuebot:#ubuntu-release- New binary: pcapy [amd64] (focal-proposed/universe) [0.11.4-1] (no packageset) [23:12] -queuebot:#ubuntu-release- New binary: pyvows [amd64] (focal-proposed/universe) [3.0.0-1] (no packageset) [23:12] -queuebot:#ubuntu-release- New binary: pcapy [ppc64el] (focal-proposed/universe) [0.11.4-1] (no packageset) [23:12] -queuebot:#ubuntu-release- New binary: seafile [amd64] (focal-proposed/universe) [7.0.4-2] (no packageset) [23:13] -queuebot:#ubuntu-release- New binary: pcapy [arm64] (focal-proposed/universe) [0.11.4-1] (no packageset) [23:13] -queuebot:#ubuntu-release- New binary: pcapy [armhf] (focal-proposed/universe) [0.11.4-1] (no packageset) [23:22] -queuebot:#ubuntu-release- New: accepted pcapy [arm64] (focal-proposed) [0.11.4-1] [23:22] -queuebot:#ubuntu-release- New: accepted pcapy [ppc64el] (focal-proposed) [0.11.4-1] [23:22] -queuebot:#ubuntu-release- New: accepted pcapy [armhf] (focal-proposed) [0.11.4-1] [23:22] -queuebot:#ubuntu-release- New: accepted seafile [amd64] (focal-proposed) [7.0.4-2] [23:22] -queuebot:#ubuntu-release- New: accepted otb [amd64] (focal-proposed) [7.0.0+dfsg-2] [23:22] -queuebot:#ubuntu-release- New: accepted pcapy [s390x] (focal-proposed) [0.11.4-1] [23:22] -queuebot:#ubuntu-release- New: accepted pcapy [amd64] (focal-proposed) [0.11.4-1] [23:22] -queuebot:#ubuntu-release- New: accepted pyvows [amd64] (focal-proposed) [3.0.0-1]