| mason | golinux: I'd be happy hand-maintaining it. | 00:29 |
|---|---|---|
| mason | golinux: Until then, it's just a renamed hunk of systemd, which is countrary to the goals. | 00:29 |
| mason | I guess I need to grab the source package and see what they're doing with it, and what dbus library or functionality they were using in 9 and before. | 00:30 |
| mason | bbiab, dinner | 00:30 |
| golinux | mason: Please read this thread: https://lists.dyne.org/lurker/message/20190820.160454.487c51ed.en.html | 01:12 |
| mason | golinux: I disagree with the phrase "libsystemd0: I thought removing the libsystemd0 dependency | 01:16 |
| mason | everywhere was not even a long-term goal in Devuan, since we | 01:16 |
| mason | now have libelogind0" | 01:16 |
| mason | golinux: From https://github.com/elogind/elogind see " | 01:16 |
| mason | The systemd project's "logind", extracted to a standalone package | 01:16 |
| golinux | IIUC, the conclusion was that removing libsystemd0 was unnecessary work now that we have elogind and that those hours maintaining it could better be used elsewhere. | 01:19 |
| mason | Anyway, I'll dig in a bit more and see if there are any hurdles to ripping it out. I can't imagine there'll be much and it'll fit in with my overall plans. | 01:19 |
| mason | golinux: elogind *is* systemd. | 01:19 |
| mason | It is, in the words of the folks who are maintaining it, "The systemd project's "logind", extracted to a standalone package" | 01:20 |
| mason | I think we want to have done with it to the extent possible. | 01:20 |
| golinux | Have you read the above link? Dmitri Bogatov did that with apt-cacher-ng and it was basically rejected. | 01:21 |
| mason | Yes, you said to read the thread, so I did. | 01:21 |
| golinux | The thread is unfortunately a bit broken. Mailman strikes again . . . | 01:21 |
| mason | Even if I missed bits of it, it remains that dropping systemd is still an important goal. | 01:22 |
| golinux | Those man hours could be put to better use elsewhere. | 01:22 |
| mason | Possibly, but it just seems wildly ironic to have systemd at the heart of Devuan's packaging tools. | 01:22 |
| golinux | At this time we do mot have the resources to repackage every instance where it appears. | 01:23 |
| golinux | LydiaK did that for jessie but there were fewer infected packages then | 01:24 |
| golinux | Oops . . . Lydia_K | 01:24 |
| mason | I'll look at what amesser did I guess and see if it seems reasonable, as a start. | 01:24 |
| golinux | Lydia_K: Let's catch up sometime. I never see you around these days. Hope you're OK. | 01:26 |
| mason | But first I have to finish my struggle with systemtap, in which it appears to have me in an unbreakable chokehold. | 01:26 |
| * golinux goes to dinner | 01:26 | |
| golinux | Pad for this week's meet: https://pad.dyne.org/code/#/2/code/edit/ZBHHORyHjyIYkPFVWb41SijQ/ | 02:40 |
| Evilham | mason: forked packages have to be maintained, if enough people are willing to do the work of not having libsystemd0 dependencies and prove time and time again that they are able and willing to keep things up to date (e.g. security), then yes, that can be a goal for devuan | 12:08 |
| LeePen | mason: If you really want to remove any packages that have their origins in systemd, then you will need to replace eudev as well as elogind. | 12:20 |
| jaromil | mason: is this systemtap you mention? https://sourceware.org/systemtap/ | 12:24 |
| fsmithred | To move a package from my personal project to devuan-packages, do I just 'git remote set-url origin...' or do I have to do something through the web interface? | 13:29 |
| LeePen | fsmithred: When I moved amesser's tomcat9 last week I just cloned it to devuan-packages using the web interface. | 13:34 |
| fsmithred | ok | 13:34 |
| fsmithred | so then I'd still need to change the origin locally so I push to the right place (or clone the devuan-packages repo?) | 13:35 |
| LeePen | Or set up devuan-packages as a separate origin so you can control where you push to? | 13:36 |
| LeePen | When I tried, I couldn't push to create a new public repo in devuan-packages. | 13:37 |
| fsmithred | I can't remember how I did it before. | 13:40 |
| fsmithred | but it sounds easy. | 13:40 |
| mason | jaromil: That's the one, yes. Like dtrace, but for Linux. | 15:52 |
| mason | LeePen: eudev comes from udev, which was in wide use before systemd subsumed it. Very slightly different. | 15:53 |
| mason | Evilham: Agreed about needing maintainers, and I'll certainly step up to track changes and fixes in the stuff I care about most. | 15:53 |
| mason | jaromil: I'm just learning it, but I needed to write a probe to track and identify what was sending signals to a jboss process on a customer's system. | 15:55 |
| jaromil | good luck :^) | 16:07 |
| mason | jaromil: I ended up finding a reasonable solution. The syntax was C-like enough to not be a problem, and it was just a matter of understanding what I could read to narrow down the source and destination of signals. | 16:08 |
| mason | jaromil: The end result ended up being trivially simple, as often happens, with some syntax borrowed from examples and some stuff cobbled up from the docs: https://bpaste.net/UI7Q | 16:09 |
| mason | jaromil: FWIW, Debian ships systemtap, and thus we ship it. I'm installing it on a box now to exercise it. | 16:10 |
| mason | (Beowulf) | 16:11 |
| mason | Evilham: I don't say this to troll, but I can't see how the mandate of the project - don't have systemd forced on us - is compatible with just renaming it but still having it as a required core of some of the most important software. | 16:17 |
| mason | jaromil: So, it looks like while we ship systemtap, we'd also need to provide dbdsym repositories. Here's what I get running the set-up step: https://bpaste.net/YHPQ I don't see that Devuan has this yet, unless I'm missing something. | 16:20 |
| jaromil | we use debian's kernel so no need to ship our own | 16:37 |
| mason | I believe they're build artifacts, so it might follow the same course as how we interact with the main repositories. | 16:40 |
| mason | jaromil: What's your take on this libelogind0 thing? I'm honestly surprised it doesn't have more people agitated. | 16:41 |
| jaromil | did not have time to look into, but yes elogind is running on my system so i should | 16:41 |
| jaromil | i'm so behind... have still notes on an article i want to write on the unique id story | 16:42 |
| Evilham | mason: check the code of libelogind0, it basically does nothing | 16:50 |
| mason | Evilham: Almost nothing: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=941937 | 16:52 |
| Evilham | that's the answer to a different question | 16:54 |
| Evilham | I am telling you "libelogind0 does basically nothing, check the code", you are telling me "debian links too much against libsystemd0" | 16:54 |
| Evilham | and yeah, you are right | 16:54 |
| mason | Evilham: Are you saying that libelogind0 isn't just a fork of libsystemd0? Maybe I'm labouring under a misconception. | 16:55 |
| mason | I guess your "read the code" answer is still the right answer. :P | 16:55 |
| Evilham | it's not zero functionality, but it's significantly slimmer than libsystemd0 | 16:59 |
| Evilham | it just happens to offer the same ABI which means less packages have to be forked which means devuan still works without having an army of packagers | 17:00 |
| Evilham | so, striving to remove it for beowulf is pointless, for chimaera I guess it can be talked about, but on linux land it's harder and harder to do desktop without something like elogind | 17:01 |
| mason | consolekit | 17:06 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!