gnu_srs1 | Hi, any ideas on how to avoid usrmerge being installed? | 10:31 |
---|---|---|
gnu_srs1 | I had to pin both usrmerge and usr-is-merged :( | 10:40 |
Unit193 | 'init-system-helpers' pulls it in, at some point it's very possible that things will start breaking for the unmerged. | 10:49 |
bb|hcb | gnu_srs1: Unpleasant topic. Either way something will break. Since Debian is going to stop supporting unmerged systems, Devuan have no option but to follow because forking will be enormous work with zero gain. The question is what is the right time to follow with the usrmerge in order to minimize the problems | 11:01 |
bb|hcb | From my experience (I have many chimaera installs, both fresh and unmerged and converted from bullseye and usrmerged), I did not hit a problem on the merged ones yet. But that does not mean that there aren't monsters lurking in the dark; many of them even being known and documented in Debian bugs. Over time more and more new problems will be introduced in unmerged systems (see above about | 11:10 |
bb|hcb | live-boot) | 11:10 |
gnu_srs1 | bb|hcb: Can you specify: "enormous work with zero gain" Additionally, I don't want it for GNU/Hurd. Maybe it's time for a devuan-hurd distro? | 11:38 |
bb|hcb | Currently we fork ~100 packages only. From what I see to unmerge we will need to fork ~20% and also do the respective testing | 12:14 |
bb|hcb | And the bad news is that this list is going to be growing with time; I also do not like that change and see no benefit from it, only problems introduced... | 12:18 |
bb|hcb | Hm, I am not well aware about Hurd - I'd rather expect that Debian will break it :( | 12:21 |
brocashelm | so, devuan will not only be a debian without systemd, but also a debian without usrmerge (by default?) | 12:21 |
gnu_srs1 | brocashelm: That's my wish. | 12:22 |
brocashelm | it looks like it's just the tip of the iceberg | 12:24 |
gnu_srs1 | As long as the dpkg maintainer opposes to ursmerge, and dpkg-fsys-usrunmess from dpkg exists, I think we can do without usrmerge :) | 12:24 |
brocashelm | yup | 12:26 |
gnu_srs1 | I think we need to fork the init-system-helpers package for Devuan. | 12:33 |
brocashelm | agreed, makes sense anyway | 12:37 |
gnu_srs1 | There is already a devuan package: 1.65.2devuan1 Does this not have the dependency on usrmerge? | 12:37 |
brocashelm | no, i upgraded it without usrmerge crap being included | 12:38 |
gnu_srs1 | good :) | 12:38 |
gnu_srs1 | What are the steps to create a devuan-specific distro? My main interest is GNU/Hurd. | 12:50 |
brocashelm | have you taken a look here? https://www.devuan.org/os/distro-kit https://git.devuan.org/devuan-sdk | 12:54 |
LeePen | I have been lurking on #debian-systemd on irc.oftc.net fort a few days. Some of the conversations are interesting, particularly the extent to which usrmerge is driven by ideology rather than what already works and the degree to which they are prepared to break and ignore problems to achieve it. | 15:01 |
LeePen | Anyway, I patched it out of our latest version of init-system-helpers (1.65.2devuan1) and our debootstrap doesn't force it either. | 15:03 |
LeePen | I think we wait and see how this plays out. As bb|hcb it isn't really the endpoint that is problematic, it is the transition. | 15:04 |
LeePen | But, if Debian makes it work, in the end we probably can't resist. | 15:05 |
LeePen | But I don't see how it can work without dpkg support, and that is being resisted ATM. | 15:06 |
LeePen | Enjoy the show! | 15:06 |
onefang | Just make sure you don't break those of us for whom usrmerge works fine with the symlinks. | 16:08 |
bgstack15 | Yeah, I don't care about the end result. I think this hill isn't that interesting. But the walk over that hill better not be bloody! | 16:23 |
rwp | The part I dread is when I find #!/usr/bin/sh appearing in published scripts because people found it in PATH there and don't know what they don't know. | 21:21 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!