rwp | In Ceres the iproute2 package changed program executable locations and the bind9 has a hard code path which breaks starting named. | 05:37 |
---|---|---|
rwp | This was discussed in #devuan-offtopic where it is not logged so I thought I would mention it here. | 05:37 |
rwp | I filed this bug in Debian for the minimum change to get a working init script again. Wanted more but feared that asking for more would get less. | 05:37 |
rwp | bind9: iproute2 change ripples through bind9 sysvinit script preventing start https://bugs.debian.org/1055429 | 05:37 |
onefang | So another case where just having the symlinks fixes things, by the look of it. But yeah, relying on PATH should do it to. | 06:01 |
rwp | If I add "ln -s /usr/bin/ip /bin/ip" then what happens when the inevitable UsrMerge comes through? Will the extra files in /bin be an obstacle? But of course that is an easy workaround too. | 06:33 |
rwp | In another topic my systems are collecting gcc versions. I have 5 different versions installed. | 06:34 |
rwp | dpkg -l | awk '$2~/^gcc-[0-9][0-9]*$/' | 06:34 |
rwp | I can't yet see why those extra ones are not candidates for autoremove. apt-mark showauto and apt-mark showmanual seem to indicate that they should be candidates. | 06:35 |
onefang | The usrmerge script does those symlinks and moves the files for you. Simples. B-) | 06:36 |
rwp | The default gcc is a symlink "/usr/bin/gcc -> gcc-13" so the others are extra ones. | 06:36 |
rwp | I have not looked at the usrmerge script yet, I should, but the result is a symlink from /bin -> /usr/bin and /sbin -> /usr/sbin so it can't be recreating those. It would need to remove them. | 06:37 |
onefang | Ah you meant recreating and removing symlinks you already made? | 06:38 |
rwp | Yes. If I create new symlinks, or if I add scripts and programs to /bin, then what does usrmerge script do in that case? Simply remove them? It must. | 06:39 |
onefang | The script written by the dpkg people that reverts that, reverts that. So you can go both ways. | 06:40 |
rwp | I think that would have the high potential for data loss and so I find it hard to believe they would do it that way. As a coder writing something like that would terrify me of the ramifications released into the wild. | 06:40 |
onefang | The usrmegse scripts moves the /bin/* files to /usr/bin/ and same with the others. then makes the symlinks. | 06:40 |
rwp | Also the dpkg people are the ones in strong opposition to the current UsrMerge and it is other people that are doing the migration. | 06:41 |
onefang | Ah, you are back in the "usrmerge is evil, they must do it badly" rut? | 06:41 |
rwp | So what does it do when it encounters something in /bin that already exists in /usr/bin? | 06:41 |
onefang | You can always just look at the scripts, both directions, see if they fuck it up. | 06:42 |
rwp | I am in the rut that I fear they will do something unexpected no matter what and therefore it is best if I stay out of the way entirely. | 06:42 |
rwp | I should look at the scripts. Unfortunately life and time is why everything does not happen all at once. | 06:42 |
Unit193 | helmut has basically taken over the migration, and think what you will about the others but he's one very solid DD. His page alone on the possible pitfalls and problems that may arise shows how much thought he's put into it (which sad to say, appears to be more than the rest combined.) | 06:44 |
onefang | Doing it or not doing it, at this stage both ways will end in "what if someone fucks it up" paranoia. | 06:44 |
rwp | But FTR I don't have anything in /bin that is not stock and therefore am trying to avoid adding anything there that is not installed by dpkg. | 06:44 |
onefang | Unit193: you got a URL for Helmut's page? | 06:45 |
rwp | But sure I could "ln -s /usr/bin/ip /bin/ip" and add a symlink workaround. But then if something "mv /bin/* /usr/bin/" then of course there would be a collision. I prefer to avoid needing to figure out what happens in that case. | 06:46 |
onefang | Anything not handled by apt I put into /usr/local, or /opt/foo | 06:46 |
rwp | Good plan. And you can see why I am avoiding it. The original problem is a hard coded path to /bin/ip which is no longer there in Ceres. Hard coded paths == Bad. | 06:46 |
onefang | The "just symlink it and be done" U keep mentioning is to symlink the directories, not individual files. | 06:47 |
onefang | Er "I keep mentioning". | 06:47 |
rwp | Ahem we are talking about current Devuan and so far Devuan has not followed UsrMerge. I have not done anything to hinder it. | 06:47 |
rwp | Yet we have a broken bind9 package, which broke my canary system, which I needed to fix. A typical snag one runs into when running Unstable. | 06:48 |
Unit193 | onefang: https://subdivi.de/~helmut/dep17.html | 06:48 |
rwp | It's only broken on non-systemd hosts that are not yet UsrMerged. So Debian will say they do not care. | 06:49 |
rwp | If not fixed in some way then Devuan would need to fork it. That's extra work that could be avoided if the package just worked on both. As it had up until the recent iproute2 version which moved the location of the ip program. | 06:50 |
onefang | The usrmerge script is in chimaera and daeldalus at least. You can install and run it. I installed and ran it after the install in chimaera, and I'm installing it from the initial mmdebstrap in my daedalus script. So far no problems. So far the actual problems others mention are few. Most are doing what you are doing, inventing theoretical problems. I'm not yet convinced it is as evil as lots of people cl/aim. | 06:50 |
Unit193 | usrmerge *does* break things, but it would appear that for some unmerged is now broken more than merged is, so now might be a good time to jump. | 06:52 |
rwp | onefang, Sorry but I have not invented a theoretical problem. An actual problem hit my system and I am handling it. | 06:52 |
rwp | As I said I have done nothing to hinder the migration. | 06:53 |
onefang | A problem that the usrmerge "move files make symlinks" script should fix. Thus only a problem coz you are fighting it one problem at a time. | 06:53 |
rwp | I am not fighting it. | 06:54 |
rwp | Please stop saying that I am inventing something. Install Devuan, upgrade to Ceres, you will hit the problem too. | 06:55 |
onefang | Not likely to hit it if I'm using usrmerge. | 06:55 |
rwp | As I said I am doing nothing to hinder the migration. | 06:56 |
rwp | Since this conversation is going nowhere fast I am going to leave the keyboard and do something more productive elsewhere. | 06:57 |
* onefang stops arguing. The "usrmerge is evil people" have yet to convince me, and I have yet to convince them to just go with it and stop worrying. I got music to make. | 06:58 | |
Unit193 | onefang: BTW, did you see dh_movetouser? | 06:58 |
Unit193 | https://packages.qa.debian.org/d/debhelper/news/20231019T190525Z.html | 06:59 |
onefang | Never heard of it, and that page doesn't actually say what it is. | 07:01 |
systemdlete | I have a r8152/r8153 usb-ethernet dongle (they both use same drivers) and they are usb 2.0 devices, afaik. Anyway, they are getting attached, somehow, to a usb 1.1 bus. | 12:16 |
djph | is the port actually usb 2? | 12:42 |
gnu_srs1 | Hello. Does anybody remember how to get audio and video working, especially with firefox? Devuan/Daedalus on a laptop. | 13:15 |
hiddener | i start icecat and abrowser with apulse for sound to work | 13:23 |
djph | gnu_srs1: what do you mean "working" ? | 13:32 |
spine-o-saurus | when i put libreoffice on hold how come apt is still pulling in libreoffice-* dependancies? | 14:32 |
djph | you didn't hold all of it? | 14:40 |
spine-o-saurus | if i hold a package, it should hold the shared libs otherwise i'll end up with incompatbile shared libs | 14:51 |
spine-o-saurus | that is just a recipe for breaking program | 14:51 |
systemdlete | djph, yes. This is a relatively recent board (Gigabyte 970A-DSP) | 17:20 |
systemdlete | *970A-DS3P | 17:20 |
gnu_srs1 | djph: aplay futuristic-beat-146661.mp3 | 17:23 |
gnu_srs1 | aplay: main:831: audio open error: Host is down | 17:23 |
gnu_srs1 | alsaplayer futuristic-beat-146661.mp3 ; snd_pcm_open: Host is down (default); Failed to initialize plugin! | 17:24 |
gnu_srs1 | Works with pulseaudio, not with pipewire. | 17:25 |
gnu_srs1 | Still no audio in firefox with pulseaudio or standalone (vlc,mplayer) | 17:25 |
djph | gnu_srs1: dunno, I don't use pipewire ... alsa worked OOTB here (but then I ended up adding pulse because some stupid program needed it | 17:26 |
gnarface | gnu_srs1: maybe you need to add yourself to a group still? | 18:48 |
gnarface | or make sure pipewire is starting or something like that? | 18:49 |
gnarface | (i also managed to do without sound daemons) | 18:49 |
msiism | Do all the Devuan 5.0 package repositories (daedalus, daedalus-security, daedalus-updates, daedalus-backports) have that new `non-free-firmware` area? | 23:25 |
gnarface | yes | 23:25 |
msiism | Okay, thanks. | 23:25 |
golinux | msiism: Nice to see you. It's been a while . . . | 23:29 |
msiism | golinux: Yeah, I just finally realized last weekend that I should upgrade to Daedalus. Worked quite smoothly. | 23:33 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!