gnu_srs | LeePen: I don't understand why I need to create an UNRELEASED version before pushing to unstable? | 10:12 |
---|---|---|
gnu_srs | What about just add to changelog:"* debian/changelog: Fix indentation, commit locally" and then push? | 10:13 |
LeePen | gnu_srs: Sorry, I think I misunderstood your question. I thought you were asking in general about a workflow for managing commits and changelog updates. | 10:53 |
LeePen | In this case, just update the changelog in a new commit and push. | 10:54 |
gnu_srs | LeePen: I pushed again. OK now? | 11:40 |
gnu_srs | Next time I'll take a closer look of how to use gbp-dch :D | 11:41 |
gnu_srs1 | LeePen: Can I try to issue a build of eudev, or maybe somebody else could do that? | 17:24 |
fsmithred | gnu_srs1, I think I could do it, or I could walk you through it if you want. | 18:43 |
gnu_srs1 | fsmithred: I think I found how to do that, but maybe you can issue the build this time. Thanks :) | 18:47 |
fsmithred | stick around so I'm sure I have the right page | 18:47 |
fsmithred | ok, 7 hours ago must be the right place | 18:49 |
fsmithred | done | 18:50 |
fsmithred | it's failing | 18:54 |
gnu_srs1 | Why?? | 18:54 |
fsmithred | there will be a link soon | 18:54 |
fsmithred | and you can check the console output | 18:54 |
gnu_srs1 | I see: dpkg-deb: error: conffile '/etc/udev/udev.conf' is present but is requested to be removed and | 19:02 |
gnu_srs1 | dh_builddeb: error: dpkg-deb --root-owner-group --build debian/eudev .. returned exit code 2 | 19:02 |
fsmithred | I trust that you understand that better than I. | 19:05 |
gnu_srs1 | Strange: I also see this: Error: Failed to build with cowbuilder. | 19:06 |
gnu_srs1 | This seems to be the real problem?? error: branch 'suites/unstable' not found. | 19:08 |
fsmithred | ???? | 19:08 |
fsmithred | I can find that branch in my web browser | 19:09 |
fsmithred | and it shows that you were the last one to push to that branch | 19:09 |
fsmithred | where are you seeing that error about suites/unstable? | 19:12 |
bb|hcb | gnu_srs1: shall i try a local build of that? maybe it will show what is the problem | 19:16 |
bb|hcb | i have skimmed through the build log, the problem is the config file, as stated above. But the reason is unclear | 19:27 |
bb|hcb | E: eudev: missing-systemd-service-for-init.d-rcS-script eudev | 19:45 |
bb|hcb | W: eudev-udeb udeb: appstream-metadata-missing-modalias-provide lib/udev/rules.d/50-udev-default.rules | 19:45 |
bb|hcb | W: libeudev-dev: breakout-link usr/lib/x86_64-linux-gnu/libudev.so -> lib/x86_64-linux-gnu/libudev.so.1.6.3 | 19:45 |
bb|hcb | W: eudev: debian-changelog-line-too-long line 4 | 19:45 |
bb|hcb | W: libeudev-dev: debian-changelog-line-too-long line 4 | 19:45 |
bb|hcb | W: libeudev1: debian-changelog-line-too-long line 4 | 19:45 |
bb|hcb | N: 18 hints overridden (11 warnings, 7 info) | 19:45 |
bb|hcb | locally it builds OK; first error obviously should be ignored, line 4 from changelog shoudl be split; I am not sure about the breakout-link warning | 19:46 |
bb|hcb | And I tried one more thing - doing a clean build by: gbp buildpackage --git-pbuilder --git-dist=ceres (this assumes that there is a ceres cowbuilder chroot prepared) and the error is same as the one seen in jenkins | 20:14 |
gnu_srs1 | bb|hcb: Thanks for looking into this. OK, I'll change the changelog line 4. But that will not make it build, right? | 20:30 |
bb|hcb | 1 by 1 :) line 4 needs to be changed anyways | 20:35 |
bb|hcb | see in changelog - eudev (3.2.2-13); maybe there shouldn't be a conf file at all? | 20:38 |
bb|hcb | so my scientific guess would be that the first line from eudev.install should be removed | 20:39 |
bb|hcb | i'll run another gbp build and will let you know | 20:40 |
bb|hcb | I can confirm that is the fix | 20:49 |
bb|hcb | You can see my changes here: https://git.devuan.org/bbonev/eudev/src/branch/suites/unstable | 20:51 |
LeePen | That may fix the build, but I am not sure it is correct. Surely we need to ship /etc/udev/udev.conf? | 20:52 |
bb|hcb | That is what I was checking now - it is empty and also there is a rm_conffile /etc/udev/udev.conf 1:3.2.2+devuan2.11~ udev | 20:53 |
bb|hcb | in eudev.maintscript | 20:53 |
bb|hcb | Maybe dh is getting more clever and detects the conflict | 20:54 |
bb|hcb | Let me install and reboot to see what happens :) | 20:55 |
bb|hcb | LeePen: you are right, that fix is not the proper one | 20:58 |
bb|hcb | update-initramfs: Generating /boot/initrd.img-5.10.0-8-amd64 | 20:58 |
bb|hcb | cp: cannot stat '/etc/udev/udev.conf': No such file or directory | 20:58 |
bb|hcb | E: /usr/share/initramfs-tools/hooks/udev failed with return 1. | 20:58 |
LeePen | Oddly, suites/unstable builds find in a beowulf chroot!!!! | 20:59 |
bb|hcb | dh got clever and detects the problem; maybe the idea was to replace the old conf with a default one | 21:00 |
bb|hcb | I don't have a single package with conf files and have to read a little bit about their proper handling | 21:01 |
bb|hcb | Maybe the previous way was to remove the existing conf file with whatever config was there and install a new empty one, but that have become an unsupported hack with the new dh | 21:02 |
bb|hcb | gnu_srs1: Shall I try to prepare a more proper fix? | 21:05 |
LeePen | gnu_srs1: bb|hcb: I think this is caused by a change (bug?) in debian/unstable's debhelper 13.5: | 21:46 |
LeePen | * dh_installdeb: Automatically rewrite `rm_conffiles` into the | 21:46 |
LeePen | new `remove-on-upgrade` feature from dpkg when possible. | 21:46 |
gnu_srs1 | So what to do, downgrade debhelper? | 21:49 |
gnu_srs1 | Anyway I will fix the long line in changelog. And since the build did not susseed, I don't need to increase the -10 version? | 21:50 |
LeePen | https://linux.debian.devel.narkive.com/5KIpWSNz/how-to-use-remove-on-upgrade-to-remove-a-configuration-file | 21:51 |
LeePen | gnu_srs1: We will have to understand the correct way to do it with the new debhelper. | 21:55 |
LeePen | See also http://bugs.debian.org/993531 | 21:59 |
LeePen | So the new deb-conffiles(5) explains why dpkg-deb is complaining: https://manpages.debian.org/testing/dpkg-dev/deb-conffiles.5.en.html | 22:08 |
LeePen | New debhelper now uses remove-on-upgrade flag. | 22:09 |
LeePen | So, are the /etc/init.d/udev and /etc/udev/udev.conf rm_conffiles in d/eudev.maintscript actually correct? | 22:10 |
bb|hcb | gnu_srs1: since it did not build and did not get installed, there is no need to bump the -10 | 22:13 |
bb|hcb | downgrading dh is definitely not the proper way; we need to understand what was the point to remove the conf file and fix it properly as LeePen explains above | 22:14 |
golinux | beer: are you coming? | 22:37 |
golinux | Beer: ^^^ | 22:37 |
Beer | Beer: No. | 22:42 |
* Beer hides | 22:42 | |
amesser | LeePen: suites/unstable needs to be reset to commit db391379e1 ( this is tags/debian/2.2.0 ) | 23:06 |
LeePen | amesser: Done. Can you verify? | 23:14 |
amesser | give me a minute | 23:14 |
Beer | It's still on e7dea2f on git.devuan.org | 23:22 |
Beer | Have you force-pushed LeePen? | 23:23 |
LeePen | Yes. | 23:24 |
Beer | to the proper origin? | 23:24 |
Beer | Could you (temporarily) give me rights on the repo so I could try? | 23:25 |
LeePen | Really done now. force-with-lease was not enough. | 23:26 |
Beer | I still see the commit I mentioned on the repo | 23:28 |
amesser | Thx worked. | 23:28 |
amesser | I just pushed the debian updates | 23:28 |
Beer | Are we talking about devuan/apt on the main suites/unstable branch? | 23:31 |
amesser | I think so, yes | 23:31 |
amesser | e7dea2f is the correct commit | 23:32 |
amesser | Before reseting it, it was at 3220383220 | 23:32 |
amesser | (suites/chimaera-proposed-updates) | 23:32 |
Beer | Ah. You mentioned needing to reset to db391379e1 | 23:32 |
mason | This is why I really dislike Git. | 23:33 |
amesser | yep, LeePen reset and I already pushed the new stuff :-) | 23:33 |
Beer | (which is debian/2.2.0) | 23:33 |
Beer | mason: Why? | 23:33 |
mason | Beer: It bites faces indistriminately. | 23:33 |
amesser | lol | 23:33 |
mason | indiscriminately* | 23:34 |
Beer | And what do you like? | 23:34 |
mason | Beer: CVS, Subversion | 23:34 |
mason | Hey, Perforce even. | 23:34 |
Beer | Oof. | 23:34 |
Beer | OK then | 23:34 |
mason | Follows the grand tradition of "if it ain't broken..." | 23:35 |
amesser | actually, I'd love to have git at work. we have subversion there. It so slow via VPN. almost every action needs a connection.... | 23:35 |
Beer | Thos are proper torture when you start trying to collaborate | 23:35 |
mason | amesser: If it's slow, maybe have the admins check that atime is turned off. atime can be a real killed on big repositories. | 23:35 |
Beer | Well, they are broken :D | 23:35 |
amesser | well, our admins have no idea about linux | 23:35 |
mason | Beer: I've done both professional and avocational software engineering for decades using traditional VCS and it's been Just Fine. | 23:36 |
amesser | they run VisualSVN ... | 23:36 |
mason | Ooh. =cough= | 23:36 |
Beer | And you are perfectly right: to each his own taste | 23:36 |
mason | I even had to use VisualSourceSafe once. I didn't like that. VisualSVN just gave me a scare. | 23:37 |
amesser | Yes SourceSafe is pain in ass | 23:37 |
Beer | I would personally hate to be locked in a CVS or SVN prison, and I'm happy with all the possibilities Git enables | 23:37 |
mason | Beer: But it leads to things like "Really done now. force-with-lease was not enough." | 23:38 |
amesser | The funny thing with SourceSafe: "Checking out a file" means no one else can change it within VCS anymore. | 23:38 |
amesser | Until the file is "Checked in" again by the same person. | 23:39 |
mason | amesser: They really mean "checking it out". :P They turn the digital into a fake physical resource. | 23:39 |
Xenguy | amesser, Weird, that reminds me of RCS | 23:39 |
amesser | And if that person was used to subversion, cvs etc and has left the company in the meanwhile - well its a big fun to reset that "checked out" state | 23:40 |
Beer | mason: Never encountered such a problem on any force pushes I did. ANd I do a lot on personal remotes | 23:40 |
amesser | I think they ended up rea-activating the account in ActiveDirectory, reset password, logged into windows and then started SourceSafe :-D | 23:40 |
mason | Beer: LeePen just hit that fifteen minutes ago. :P | 23:41 |
Beer | mason: Yes, and? | 23:42 |
Beer | mason: force-with-lease is special, per manual | 23:43 |
mason | git, the systemd of version control =cough= | 23:43 |
Beer | I personally don't understand its need, since force-pushing is _in essence_ destructive and shall avoided at all cost on collaborative branches. PR ftw | 23:43 |
Beer | (or merge) | 23:44 |
Beer | mason: OK now that's proper trolling :oD | 23:44 |
amesser | I think, in our company they are a bit afraid of replacing subversion with git because of the "senior" developers. Git can be confusing if you come from subversion, and humans sometimes hardly able adapt. | 23:45 |
amesser | however, some departements are already using git internally | 23:46 |
amesser | well, our department is also afraid of using c++ instead of c :-D | 23:47 |
Beer | Hard to let go the "mandatory centralised truth" logic ;) | 23:47 |
LeePen | gnu_srs: bb|hcb : This patch fixes the build failure by not removing udev conffiles that are in the eudev package. | 23:47 |
amesser | tankships - hard to steer | 23:47 |
LeePen | http://ix.io/3xMP | 23:47 |
Beer | Even people/organisations using Git have trouble dealing with forks and sharing stuff without going throug ha central repository... even though Git (and its ideas) are nowhere near new... | 23:48 |
LeePen | But I am unlcear why they were being removed anyway: seems wrong to me. | 23:48 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!