[15:21] <seb128> hum, was there any autopkgtest infra sadness?
[15:22] <seb128> https://people.canonical.com/~ubuntu-archive/proposed-migration/update_excuses.html#libinput
[15:22] <seb128> or is that just libinput doing something weird?
[15:44] <Laney> seb128: Looks like triggers with systemd but not gnutls28 are broken
[15:44] <Laney> is that some missing package relationship?
[15:46] <rbalint> since rhonabwy rebuild fixed test failure with gnutls28 i suspect there is an untracked ABI break
[15:47] <rbalint> Laney, but not the test is passing, thus both systemd and gnutls should migrate in the next run this will be solved
[15:48] <rbalint> xnox,^
[15:49] <Laney> you're confident there aren't any other undetected breakages?
[15:50] <rbalint> Laney, hm, regarding gnutls28 no, but i think systemd is ok
[15:50] <Laney> I did mean gnutls28
[15:51] <rbalint> xnox?
[15:51] <Laney> I'm guessing that rhonathingy was only detected because it has an autopkgtest
[15:55] <rbalint> Laney, filed LP: #1897599 to let us figure that out
[15:55] <Laney> nice
[15:55] <Laney> I was going to say, should that have block-proposed, but it does
[15:55] <rbalint> Laney, :-)
[15:56] <Laney> I should probably kill this in progress run in case it migrates it
[15:56] <Laney> ah except ppc64el failed
[15:57] <Laney> oh no, it's still running
[17:09] <xnox> rbalint:  Laney: no abi break....
[17:10] <xnox> new point release of gnutls adds more constants; rhonabwy has code that is compiled when new constants are detected at buildtime, but it doesn't check if new constants/algos are available at runtime and tries to use them.
[17:10] <xnox> rhonabwy is the one broken, gnutls28 is not.
[17:11] <Laney> that sounds backwards from what happened?
[17:11] <xnox> also rhonabwy builds a fresh library & fresh unit tests, and then tries to execute unit-tests against system installed (old) librhonabwy and that has missmatched expectations.
[17:12] <Laney> or are you talking about building during the tests?
[17:13] <xnox> Laney:  autopkgtest builds brand new library & unit-tests which have matching expectations, but then it executes just compiled unit-tests against system library, which has been built against old gnutls28, without exposing new curve that unit-tests think library should know about.
[17:13] <xnox> and vice versa.
[17:13] <Laney> ok, that makes sense
[17:13] <Laney> next, why does systemd break?
[17:14] <xnox> (so gnutls28 is forward compatible, and ghonabwy tries to go back in time, but that's not really supported, or matters)
[17:14] <xnox> systemd vs gnutls28 => no idea =)
[17:14] <xnox> didn't look
[17:15] <Laney> alright I'll leave that little question in the air and go out for my daily walk :>
[17:15] <xnox> wlroots saysa unkonw/regression, yet it is pass superfecial?!
[17:15] <xnox> did it drop tests or something?
[17:15] <xnox> looks all green otherwise.
[17:16] <xnox> oooh systemd smells like nftables migration issue?!
[17:17] <Laney> should reproduce for you, just try testing something (e.g. happend with input) with --apt-pocket=proposed=src:systemd{,src:gnutls28}
[17:17] <Laney> then the network breaks and you can have fun using the serial console to break into the machine
[17:17] <Laney> ubuntu/ubuntu
[17:17] <Laney> s/input/libinput/
[17:47] <ItzSwirlz> o/
[17:48] <ItzSwirlz> Laney: Do you have a pre-made .deb (All of them, preferably) of the built version of libmozjs78-0 in Ubuntu/mozjs78 in Debian?
[17:48] <ItzSwirlz> Working with cjs right now as me and Debian Cinnamon Team and others work to port cjs to mozjs78 (from 52)
[18:38] <rbalint> Laney, xnox the systemd vs. gnutls issue is just udev being upgraded without systemd where i saw it
[18:40] <rbalint> Laney, xnox i'm thinking about thightening dependencies between udev & systemd to avoid such partial upgrades, but apt could know better, too https://objectstorage.prodstack4-5.canonical.com/v1/AUTH_77e2ada1e7a84929a74ba3b87153c0ac/autopkgtest-groovy/groovy/amd64/m/mir/20200928_132812_257f0@/log.gz
[19:31] <Laney> ItzSwirlz: no, but you can get it from the archive can't you?
[19:31] <Laney> rbalint: ok, well if that's it then tightened dependencies would make sense
[19:32] <ItzSwirlz> Can't seem to find it there
[19:32] <ItzSwirlz> Do you have a direct link or something? I've searched around the Debian Package Tracker and packages.ubuntu.com
[19:32] <Laney> apt download libmozjs-78-0?
[19:32] <Laney> or install ubuntu-dev-tools and pull-lp-debs mozjs78
[19:32] <ItzSwirlz> There we go. Cuoldn't remember the command for getting it, thanks
[19:40] <Laney> I retried wlroots, hopefully that makes systemd go in
[19:43] <rbalint> Laney, i retried already, too
[19:44] <Laney> ah well the results weren't in when I looked
[19:45] <Laney> better hope it's not flaky :>
[19:51] <rbalint> Laney, all of them passed, so it is just crossing fingers that's left :-)