freem | so, what I was talking about in #devuan was, I know it would be a lot of work to do that for *all* packages, but, it might be easier for the ones you guys already fork to separate the daemon enabling system from the binary itself | 00:55 |
---|---|---|
fsmithred | freem, what did you have in mind? | 00:55 |
freem | also, I believe this would actually reduce packages complexity. | 00:55 |
freem | The con would be it would increase the number of available packages. | 00:56 |
freem | Let me take a simple example. | 00:56 |
freem | Mpd. Music player daemon. | 00:56 |
fsmithred | you're outside my area of expertise, so I won't have a lot to say about this | 00:56 |
fsmithred | but someone else might | 00:56 |
freem | The deb includes the binary, the manpages, licences, systemd/sysVinit (in debian) files, and pre/post-rm/inst scripts | 00:58 |
freem | This imply one *must* disable it manually, as root, after install and every update, if the goal is only to provide it to users | 00:58 |
fsmithred | ? | 00:59 |
fsmithred | I don't understand what you mean about disabling it | 00:59 |
freem | splitting the package to remove the daemon scripts, and move them to individual packages, would keep the auto-enable feature, while making new init supports easier | 01:00 |
freem | fsmithred: I'm sorry, could you precise the question? WHat did you not understood? | 01:01 |
fsmithred | what do you want to disable? | 01:01 |
freem | I don't want to disable | 01:01 |
freem | I think it would make the alternative init implementations easier by splitting the packages | 01:02 |
freem | apt-file show mpd | grep '\/etc\/' | 01:02 |
fsmithred | five files | 01:03 |
freem | output of this command shows that the init.d script is directly embedded in the package. I'm pretty sure it alsa includes the other supported init files | 01:04 |
freem | it's not about amount of data | 01:04 |
fsmithred | I don't see any service files | 01:04 |
freem | I'm still under debian :) what are service files names like? | 01:04 |
freem | /lib/systemd/system/mpd.service maybe? | 01:05 |
fsmithred | yeah, .service | 01:05 |
freem | well, so, how does debian works to choose one or another? I'll check (in the process right now) but I'm sure it does it with pre-rm/post-rm/pre-inst/post-inst scripts | 01:06 |
fsmithred | choose what? | 01:06 |
fsmithred | if systemd is running, it will look for service files | 01:06 |
freem | how to install and enable stuff? | 01:07 |
fsmithred | if sysvinit is running, it will look for init scripts | 01:07 |
freem | what was the way to extract debs already... | 01:07 |
freem | dpkg-deb -R | 01:07 |
fsmithred | dpkg -i | 01:08 |
fsmithred | to install | 01:08 |
fsmithred | ar x | 01:08 |
fsmithred | to extract the contents of a .deb package | 01:08 |
fsmithred | maybe other ways, too | 01:08 |
freem | raw-extract, to have the scripts | 01:08 |
fsmithred | ar x | 01:08 |
freem | dpkg-deb --raw-extract works too :) | 01:09 |
fsmithred | and then unpack the control archive | 01:09 |
freem | got some pre-inst and alike scripts in debian mpd deb | 01:10 |
freem | the prerm is... only related to systemd. | 01:10 |
freem | conforts me in my opinion, let me provide it (other files are bigger) | 01:10 |
freem | http://paste.debian.net/1152259/ | 01:11 |
fsmithred | yeah, I'm pretty sure you can ignore that | 01:12 |
freem | most of the stuff in the 3 scripts *is* related to init system | 01:12 |
freem | what I mean here, is not about systemd (you and I don't care about it I think) | 01:12 |
freem | it's just that, I think it would be easier to have a package with binary, as now, and alt packages that, for example, pre-depend on the 1st one, and depend on a specific init | 01:13 |
freem | those would install a specific init support. Would make it pretty simple imo to support more inits, without maybe even touching the debian original scripting | 01:14 |
freem | but maybe there are issues I don't see, I'm no expert after all. | 01:16 |
fsmithred | seems like that would only be needed if the parts of the different init systems conflict with each other | 01:18 |
freem | is there is any package I could inspect in devuan that supports severa init systems? So that I could stop emitting noise? | 01:19 |
fsmithred | cron, cups dm-event (whatever that is) | 01:20 |
fsmithred | man-db, mdadm, rpcbind | 01:20 |
freem | ty, I'll read the code :) | 01:20 |
fsmithred | I'm just listing some of the .service files I have | 01:20 |
fsmithred | and I know for sure most of those have init scripts | 01:21 |
freem | but another init as well? | 01:21 |
fsmithred | no, just those two | 01:21 |
freem | two? | 01:21 |
freem | sysV and? | 01:21 |
fsmithred | sysvinit and systemd | 01:21 |
freem | openrc? | 01:21 |
freem | oh | 01:21 |
freem | like mpd, then | 01:21 |
fsmithred | openrc uses the sysvinit scripts | 01:22 |
fsmithred | for now | 01:22 |
freem | is there is any package that already have another but systemd and rc.d support? | 01:22 |
freem | even in unstable? | 01:22 |
fsmithred | I don't think so | 01:23 |
fsmithred | unless parazyd put a new openrc in the repo | 01:23 |
freem | I see. Do you know how runit-init is managed actually in debian? | 01:23 |
fsmithred | not sure if that happened yet | 01:23 |
fsmithred | nope | 01:24 |
fsmithred | nor in devuan | 01:24 |
freem | it's... not handled in devuan stable, at least. For Buster, there are seeds. | 01:24 |
fsmithred | I just know that runit was added to the choose-init section of the installer | 01:24 |
fsmithred | and you'll see it in the first point-release of beowulf | 01:24 |
freem | good! | 01:25 |
fsmithred | but I have no idea what it does | 01:25 |
freem | I've read the debian /etc/runit/2 file | 01:26 |
parazyd | fsmithred: OpenRC needs no changes since http://git.savannah.nongnu.org/cgit/sysvinit/insserv.git/commit/?h=1.22.0&id=6a65f4412a2d24d78741f49e64e1128993341e65 | 01:26 |
fsmithred | I don't know what any of that means | 01:27 |
freem | shortly said, for each sysvinit stuff it checks if there's a runit folder with same name. If yes, it does not run the sysv one and waits for runsvdir to do it's job | 01:27 |
freem | it's pretty fragile imo. | 01:27 |
freem | oh, you guys are on savannah? I thought this thing dead for years | 01:28 |
parazyd | No, it's just sysvinit's upstream | 01:28 |
freem | good to see it's still possible to avoid github/gitlab on some projects | 01:28 |
freem | oh | 01:28 |
fsmithred | that's not us | 01:29 |
freem | oups? :) | 01:29 |
parazyd | btw. OpenRC can use both sysv and openrc initscripts and does so transparently | 01:29 |
parazyd | It's pretty stable | 01:30 |
freem | all init can, no? | 01:30 |
parazyd | I don't know if runit can, and I'm pretty sure upstart doesn't. | 01:30 |
parazyd | (Might be totally wrong here though) | 01:30 |
freem | runit runs... run files. That's all. | 01:30 |
freem | if the run file exits, then, it runs finish. | 01:31 |
freem | when done, it reruns the run file. | 01:31 |
freem | my run files are usually 3 liners: shebang, empty line, the command :) | 01:31 |
parazyd | ok | 01:32 |
freem | debian systemd last or maybe 2 versions ago was able to run those, too. Support was incomplete. | 01:32 |
* freem remembers the international mailing list at that time | 01:33 | |
freem | so, you guys plan to implement all inits inside the debs? | 01:34 |
fsmithred | I don't know | 01:35 |
freem | I wonder if I can write a source deb that shows my idea. It would be the best way for you to decide if it's a good idea or not, I think. | 01:36 |
fsmithred | the benefit has to outweigh the additional work | 01:37 |
freem | of course | 01:37 |
fsmithred | most of what I see in /etc/init.d came from packages we don't fork | 01:38 |
freem | indeed. But AFAIK, in debian, stuff in /etc/init.d is only "supported" by alternative inits the same way sysV supported it: no auto restart if daemon crash | 01:39 |
fsmithred | yeah, I don't see auto restart | 01:40 |
freem | I doubt a shim for runit-init could be easyly writen, dunno for openrc | 01:40 |
fsmithred | I also don't see services crashing and needing to be restarted | 01:40 |
freem | and I know for certain and by experience that any of those init.d scripts are a lot more complex than runit ones | 01:40 |
freem | lucky you then :) | 01:41 |
freem | I personnally had enough problems with hardware embedding sysvinit and pppd that would not die on failure and thus not restart, making a physical intervention (several dozens of kilometers minimum) to reboot the hard | 01:42 |
freem | it's nice to know this experience is rare. | 01:43 |
freem | I mean, the bad stuff. | 01:43 |
freem | (damned kortex routers) | 01:44 |
freem | well, I'll write a PoC deb. It'll serve me anyway, even if you guys don't use it :) I 'll try to do that this week and I'll share it here. | 01:45 |
freem | but probably just a binary one, as 1st step. Don't want to overengineer stuff. | 01:46 |
fsmithred | ok, cool | 01:46 |
freem | (binary as in, what you would have with dpkg-deb -R foo.deb ./foo | 01:47 |
freem | if you like the idea, I win by having help, otherwise, I'll be back to my plan to do it myself anyway :) | 01:48 |
freem | (like you guys did by forking, I guess) | 01:48 |
parazyd | How can I make an orga on gitrea? | 11:38 |
parazyd | *gitea | 11:38 |
yeti | https://gitea.devuan.dev/?repo-search-tab=organizations | 19:26 |
onefang | fsmithred: I'm guessing you want to know how to get syslinux + EFI working for ISOhybrid images? | 21:45 |
onefang | So far I have syslinux + BIOS / EFI + GPT on a 400 GB USB stick, booting ASCII that's on a EXT4 partition. Not all of the syslinux modules work when using EFI, and it's very version dependant. Debian SIDs version is the most reliable, the devs say that themselves. | 21:49 |
onefang | And one of the reasons it took so long is I spent a day wondering why there was no keyboard or mouse after booting ASCII. Found out I had been using an AMD64 kernel for the 686 user space. Surprised it worked at all. lol | 21:51 |
onefang | I even managed to get Syslinux to boot Grub Invaders. B-) | 21:54 |
onefang | So tell me what you need, and I'll see if I can get it working. | 21:56 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!