[07:26] please remove ldc and its reverse-deps from ppc64el, upstream doesn't want to support that architecture, and Debian asked to remove them too https://github.com/ldc-developers/ldc/issues/2824 [07:26] ldc-developers issue 2824 in ldc "Fails to build from source on ppc64el" [Open] === M_hc[m] is now known as _hc [07:43] anyone looking at the r-cran-rgenoud test regression with glibc? not sure what the issue is there. it's been too many years now since I did r type things, and I never used r even then [07:43] -queuebot:#ubuntu-release- Unapproved: accepted grub2-signed [source] (bionic-proposed) [1.93.5] [07:44] * acheronuk shudders at the memory of finite element in fortran! [08:26] acheronuk: I did have a quick look, they seem to do the same calculation with and without a seed. https://salsa.debian.org/r-pkg-team/r-cran-rgenoud/blob/master/tests/testthat/test-genoud_fixed_seed.R#L92 https://salsa.debian.org/r-pkg-team/r-cran-rgenoud/blob/master/tests/testthat/test-genoud_no_given_seed.R#L84 [08:27] with the new glibc, the results without a given seed are now closer to the old results with a seed [08:30] acheronuk: so adjusting the tolerance there might be safe, but it concerns me that the result only changed on amd64 and armhf [08:30] infinity: do you have any thoughts ^ ? Is this expected? [08:30] indeed [08:54] btw with the new ldc we finally gained full arm64 support, while reverse-deps never really worked / compiled on ppc64el [10:01] -queuebot:#ubuntu-release- New binary: llvm-toolchain-snapshot [ppc64el] (cosmic-proposed/universe) [1:8~svn340819-1] (no packageset) [10:17] acheronuk, ginggs: i've spent far to long poking at r-cran-rgenoud and have nfi what's going on [10:18] it doesn't even call that many libc functions in its core loop [10:42] mwhudson, acheronuk: did either of you find a repo or bug tracker for r-cran-rgenoud? [10:42] ginggs: no [10:42] ginggs: seems like a one man thing to me [10:43] it also has no rdeps so i hope some AA deletes it before i waste too much more time on it [10:48] folks, please ignore symfony tests on armhf and s390x, they always failed, I did ignore test results because of deprecation warnings, and they started passing because of this reason... when I discovered that it was too late [10:53] looks like the unseeded run is wrongly finding the max from the seeded one? fun [10:57] ginggs: authors github? https://github.com/JasjeetSekhon/rgenoud [10:57] 2 commits and out of date! [11:05] acheronuk: thanks, i couldn't find anything. i'll drop them an email [11:11] the unseeded version just seems to pick a random number between 1 and 2^31 to use as a seed https://salsa.debian.org/r-pkg-team/r-cran-rgenoud/blob/master/R/rgenoud.R#L20 but all the failed tests fail with the same result === cascardo` is now known as cascardo [12:17] fun that running unseeded in R locally with old glib mostly gives the same answer, but occasionally it does a swerve and finds one of the other maxima! [12:20] so not sure I believe this is a hugely reliable test to start with [12:22] https://paste.ubuntu.com/p/f3vtTNcw5t/ vs https://paste.ubuntu.com/p/mNqrRK2YXs/ [12:23] ^^ 2 following runs [12:31] -queuebot:#ubuntu-release- Unapproved: rax-nova-agent (xenial-proposed/universe) [2.1.15-0ubuntu1~16.04.0 => 2.1.15-0ubuntu1~16.04.1] (no packageset) [12:43] cjwatson, hello, can you please ack/nack libapache-session-perl sync/merge? (I mean, is this add-recommended-packages change still needed now that Debian has autopkgtests? they even dropped debian/test/control in the last upload... ) [12:44] you did some of that perl changes, I'm merging them but I don't really know how to check if the bug is fixed or not... (and we have no ongoing perl transition fortunately) [12:44] I'd rather it be merged, but the worst case if you sync it is that somebody will have to reinvent the same thing for the next perl transition [12:44] so I don't care very deeply === didrocks999 is now known as didrocks [12:57] -queuebot:#ubuntu-release- New binary: llvm-toolchain-snapshot [i386] (cosmic-proposed/universe) [1:8~svn340819-1] (no packageset) [13:07] so cjwatson reintroduce that file and that's all? [13:08] Sorry, I don't remember and am in the middle of two other things right now [13:08] If it's a difficult merge due to the underlying file not existing any more, then just sync instead and we can sort it out later [13:08] (if necessary) [13:09] -Depends: @, pkg-perl-autopkgtest, libdbi-perl [13:09] +Depends: @, pkg-perl-autopkgtest, libcgi-pm-perl, libdbi-perl [13:09] this is the patch, in debian/test/control, but that file is now removed [13:10] Just drop it and sync then [13:10] it is recommended in the libapache-session-perl package, so maybe we can teach autopkgtest for perl to install recommends by default? [13:10] It was a subtle problem and it's possible it's been fixed; if it hasn't then we'll likely need to fix it somewhere else anyway [13:10] maybe debian did that... [13:11] [ gregor herrmann ] [13:11] * Remove debian/tests/control; add libdbi-perl to Recommends. [13:11] the reason for adding libdbi-perl to recommends is similar, so maybe this is really already fixed [13:11] in case we break on next perl transition, we will have a lot of "ubuntu1" previous packages in release, and we can quickly notice of the dropped delta, lets sync [13:12] ack [13:34] -queuebot:#ubuntu-release- Unapproved: snapd (xenial-proposed/main) [2.34.2 => 2.35.1] (desktop-core, ubuntu-server) [13:37] -queuebot:#ubuntu-release- Unapproved: snapd (bionic-proposed/main) [2.34.2+18.04 => 2.35.1+18.04] (desktop-core, ubuntu-server) [13:37] -queuebot:#ubuntu-release- Unapproved: snapd (trusty-proposed/universe) [2.34.2~14.04 => 2.35.1~14.04] (no packageset) [19:20] -queuebot:#ubuntu-release- New binary: llvm-toolchain-snapshot [amd64] (cosmic-proposed/universe) [1:8~svn340819-1] (no packageset) [20:58] slangasek, do you like the new diaspora-installer autopkgtest? [20:58] I can't in pbuilder environment install the package, not sure why [20:58] http://autopkgtest.ubuntu.com/packages/d/diaspora-installer/cosmic/amd64 [20:58] but it seems good, I'm tempted to remove the block [22:20] hmm [22:21] hmm [22:22] hmm [22:22] looking at twisted/foolscap/txtorcon [22:22] current situation is that there's a txtorcon in proposed that does not build a python 2 package [22:22] but a previous upload built a totally broken one [22:23] i _guess_ the thing to do is remove the broken one and upload a txtorcon-for-python-2 source package that is the old txtorcon source [22:23] but all this requires an AA to sort out [22:26] well unless txtorcon-for-python-2 gets a ridiculous dishonest version i guess but well