libera/#devuan-dev/ Thursday, 2020-01-23

gnu_srs1LeePen: I'm back. Should we upgrade the transitional packages {,lib}udev*, etc to 3.2.9-3 or change the dependencies of {,lib}eudev* to a fixed version?12:20
gnu_srs1I'll test upgrades next from Debian Buster to Devuan Beowulf/unstable when we have decided.12:22
gnu_srs1fsmithred: I saw that you had tested Beowulf on an encrypted disk. Can you check if eudev 3.2.9-3 is fine, see #26112:28
LeePenI think it is better not to have to update udev transitional packages every time, if that will work.12:28
LeePenDoes eudev need versioned depends on udev at all?12:29
fsmithredgnu_srs1, eudev did not upgrade. Still at 3.2.7-612:31
gnu_srs1fsmithred: eudev 3.2.9-3 is in unstable12:32
fsmithredyeah, I see it12:32
fsmithredI think I just did 'apt upgrade' and didn't do dist-upgrade12:32
fsmithredwill do it now12:32
gnu_srs1LeePen: We need the epoch for eudev packages: Conflicts: udev (<< 1:${source:Version})12:33
gnu_srs1Otherwise Debian udev versions takes precedence. I can change to e.g.:  udev (<< 1:3.2.7+devuan1.1), the latest transitional package available.12:35
LeePenSeems resonable and stops us having to bump and rebuild src:udev.12:39
gnu_srs1LeePen: The udev transitional packages are old. Maybe we could build new versions 1:3.2.9-4 and new eudev versions 3.2.9-4 having a fixed dependency on udev 1:3.2.9-4?12:39
LeePenWhy are the transitional packages in a separate source anyway?12:40
gnu_srs1I'd really like to fix some stuff in src:udev.12:40
LeePenCrack on!12:41
gnu_srs1The transitional packages cannot be included in src:eudev. Combining epoch and no epoch is not possible. We have tried that before.12:41
LeePenOK. I see now.12:44
fsmithredI got the newer eudev on dist-upgrade. No udev is installed. It boots. Anything specific I should check?12:52
gnu_srs1fsmithred: see #261: eudev: System with encrypted root fs and swap fails to shutdown cleanly13:12
fsmithredgnu_srs1, I'm working on a fix for that13:15
fsmithredif there's a different way to fix the delay, please tell me about it.13:17
fsmithredmxlinux made a package that uses dpkg-divert to replace cryptdisks.functions (or cryptdisks-functions in beowulf)13:21
fsmithredso I cloned it:
gnu_srs1fsmithred: Seems like #261 is different?, no lvm: On a fresh Devuan ASCII install with encrypted root and swap (no LVM, no RAID),...13:24
fsmithredthere are two fixes that people have been using13:25
fsmithredone works for just luks, the other works for lvm13:25
fsmithredgnu_srs1, were you thinking of repackaging cryptsetup?13:26
gnu_srs1Maybe I should ask the bug reporter about #261. Are the fixes for luks/lvm related to eudev or only cryptsetup?13:26
fsmithredseems to be just cryptsetup13:27
fsmithredthat's all anyone has messed with to get rid of the shutdown delay13:27
gnu_srs1No I do not intend to repackage cryptsetup. Is that package forked by Devuan?13:27
gnu_srs1If not, then bugs should be filed in Debian BTS?13:29
jaromildepends, the bug may not affect debian, sorry I am just jumping in but don't have enough information13:30
jaromili'm just triggered by cryptsetup since is a rather important package to me :^P13:30
fsmithredhere's the debian wontfix report
fsmithredyeah, apparently if you use systemd it shuts down too fast to see the error messages.13:31
jaromillol yea I see it now13:31
jaromilbut well I confirm is a chicken-egg problem unmounting the root partition13:32
fsmithredyeah, one idea that's been discussed is to go to initramfs at shutdown13:32
jaromiland we cannot rly fix it realistically13:32
jaromili see... well..13:32
jaromil... sorry i use tomb :^) I don't care13:32
fsmithredgetting rid of the delay seems to be sufficient13:32
jaromilit works better and doesn't keeps the keys on the disk13:33
jaromildisclaimer: wrote it myself :^S13:33
fsmithredyeah, I've looked at it13:33
fsmithredseems like a convenient tool13:33
fsmithredso should I go ahead and make the cryptsetup-modified-functions package?13:34
fsmithredif so, I have some questions about packaging details13:34
masonfsmithred: I tried to spin up a test VM yesterday and it choked on MD-RAID1 /boot, which was surprising.16:08
masonSo I didn't get a chance to complete the install. It also choked thinking that it was on an EFI system, which it wasn't, which was also surprising. Couldn't install GRUB. So I tossed it for now.16:10
masonBut other than the odd quoting, that LVS-related patch seems reasonable, and the quoting probably doesn't break anything.16:10
fsmithredmason, were you using an installer iso or a live iso?16:59
masonfsmithred: Always installer for this kind of thing, but I might do it again with a live image and my scripting just to get it done.17:00
fsmithredI had a hard time doing it17:00
fsmithredwith live iso17:00
fsmithredso which quoting seems odd to you?17:00
masonI need to look at the error... Might not have been related to the RAID /boot, in which case I should open a bug for it.17:00
masonfsmithred: There were double quotes around things that seemed unnecessary. Let me find it.17:01
fsmithredprobably vgs=17:02
fsmithredwith the sed, there are a lot of quotes17:02
fsmithredI haven't dissected that sed command.17:02
masonThere shouldn't be any quotes needed there though.17:20
masonfsmithred: I'll whip up a demonstration.17:20
fsmithredI usually quote in similar situations.17:21
masonfsmithred: But it's incorrect. Anyway:
masonWay, way better than the questionable regex with the bad quoting.17:29
fsmithredum, ok. All the variables in the rest of that file are quoted, including the command substitutions.17:33
fsmithredI have to go. bbl.17:33
masonfsmithred: You're right, that is all over the place in there. :/17:43
masonI don't know. Most code is horrible, so I guess this is no different. I have no objections to however you want to commit it, anyway, not that it's matter all that much even if I did. :)17:44
masonMy general heuristic is that if an element of code does nothing, or nothing useful, it shouldn't be there, with the rare exception of being explicit with braces and such to clarify blocking.17:45
golinuxIt's meet time!21:35
masonGods, I don't have a meeting! I can join!21:35
onefangI'm awake, but I have to duck out now, before it gets too hot.21:53
masonIt was good to finally make a meeting. Been a while.22:42
golinuxNice to have you back!22:45
LeePenfsmithred: Just move the purpy themes to beowulf and chimaera in dak.22:51
gnu_srs1LeePen: Around tomorrow I'll have updated src:eudev and src:udev packages to build for unstable.23:09
gnu_srs1Should we build or do you want me to install Debian Buster, create a local repo and then23:09
gnu_srs1upgrade to Devuan Beowulf/unstable before to test? That will delay things some time!23:09
LeePengnu_srs1: Not sure. Let's get them built for unstable first and get debootstrapping ceres fixed.23:11
LeePenWe are on the point of freezing beowulf.23:12
gnu_srs1So you will be testing debootstrapping of ceres. OK:)23:13
LeePenYes, we know that is broken by this at the moment.23:13
gnu_srs1I think the udev and eudev packages should be a part of Beuwulf.23:14
fsmithredLeePen, I'm not aware that I have access to dak.23:16

Generated by 2.17.0 by Marius Gedminas - find it at!