| al1r4d | ^same question | 07:16 |
|---|---|---|
| gnarface | you could try just building it and packaging it and installing it... | 07:17 |
| gnarface | the question kinda reeks of a troll agenda, because it's worded in a way that falsely implies there aren't already a bunch of alternate init systems in devuan "for fun" | 07:18 |
| gnarface | the real question should be, why isn't shepherd in debian already? is it something wrong with shepherd, or is it something wrong with debian? (i'm gonna assume both for now, because this came out of the blue like an organized psy-op) | 07:18 |
| gnarface | but if shepherd were something *I* was interested in, i'd at least try to compile it myself once first before trying to leverage psychology to try to make other people do the work for me, and i'd definitely stick around longer than 10 minutes if i was needing and expecting help with it | 07:22 |
| paculino | That is strange, I would have thought that Debian Gnu/Hurd was using Shepherd as an option, but apparently only runit is used now according to the mailing lists | 07:22 |
| gnarface | i heard of shepherd for the first time yesterday | 07:23 |
| paculino | You can use guix on debian-based similarly to flatpak, and guix has shepherd | 07:23 |
| al1r4d | i have a question: does anyone can enable hardware-acceleration in chromium/brave-browser? | 07:23 |
| al1r4d | Description:Devuan GNU/Linux 6 (excalibur/ceres) | 07:23 |
| al1r4d | :) | 07:23 |
| paculino | My partner uses guix, so I knew of it already. Iirc, someone on the forum made a guide for removing systemd in any distro, and it used shepherd | 07:24 |
| gnarface | al1r4d: i think there is a way to do that but you kind sabotage the primary point of using chromium instead of firefox or chrome, because it requires adding the closed-source DRM "wildvine?" implementation | 07:24 |
| gnarface | al1r4d: there might even be an optional package for that already in the repo, but worst case you just rebuild chromium with it, i've seen instructions somewhere... | 07:24 |
| al1r4d | i give up from firefox unfortunately | 07:25 |
| al1r4d | gnarface: thx! | 07:25 |
| gnarface | al1r4d: firefox has it already and you just have to enable it by checking a box in settings. the chromium package in the repos might actually have to be rebuilt, but i'd check for an optional package first | 07:25 |
| hagbard | just a PSA for other devuan mirror operators: claude.ai (User Agent ClaudeBot/1.0) disobeys robots.txt, and will recursively download the entire archive, repeatedly, continiously. | 10:34 |
| onefang | Ouch. | 10:35 |
| hagbard | So you might want to have a look in your webserver logs, and you might save some traffic by blocking that user agent. | 10:35 |
| onefang | No sign in my logs. | 10:42 |
| hagbard | Good. Maybe it's just my luck. A rewrite rule in the apache site config to block that Agent did the trick for me. Most entries in the access log have been just from them. | 10:47 |
| amarsh04 | hmm, bind9 seems to have stopped calling resolvconf and I can't figure out when it happened | 12:30 |
| gnarface | what release? | 12:33 |
| amarsh04 | that's the problem, I've tried all versions of bind9 available in Devuan and still hitting the same problem, so I don't know if it is actually a problem in some other package like libc6 | 12:34 |
| gnarface | hmm, weird | 12:35 |
| gnarface | is wpasupplicant or any types of network management daemons involved? | 12:35 |
| gnarface | avahi-daemon, network-manager, etc? | 12:35 |
| amarsh04 | not that I know of | 12:35 |
| gnarface | well it's easy to be sure | 12:35 |
| gnarface | some of these things might be part of the default install | 12:35 |
| gnarface | check the output of "ps aux --forest" | 12:36 |
| gnarface | if you didn't configure /etc/network/interfaces by hand, the answer very likely could be yes, actually... | 12:36 |
| gnarface | dhcp daemons too | 12:36 |
| gnarface | check for those | 12:36 |
| amarsh04 | well, dhclient triggers resolvconf, but then so should bind9 | 12:37 |
| amarsh04 | my workaround is to edit /etc/dhcp/dhclient.conf and uncomment prepend domain-name-servers 127.0.0.1; | 12:38 |
| gnarface | i think that's the right way to do it... | 12:38 |
| gnarface | bind9 is supposed to query the root nameservers by default | 12:38 |
| gnarface | it wouldn't use your resolv.conf i don't think | 12:38 |
| gnarface | although personally at that point i would question the usefulness of including dhcp in this loop | 12:39 |
| gnarface | if you just want a caching nameserver for a single machine that respects your ISP's default dhcp-supplied network configuration, i think dnsmasq is the one that comes out of the box with suitable defaults for that | 12:40 |
| amarsh04 | well, using resolvconf used to "just work" to make bind9 on localhost the nameserver rather than my router's ISP-linked nameserver | 12:40 |
| gnarface | a change to the dhcp client or the network manager perhaps? | 12:41 |
| gnarface | or you forgot you made some change to bind9 maybe but, i think the former | 12:41 |
| amarsh04 | could be, anyway the Australian government mandates blocking dns names of torrent sites so I run a local nameserver | 12:42 |
| amarsh04 | looking at the bind9 source | 13:13 |
| gnarface | i wouldn't think that would be necessary, i think just looking at the documentation for the config file should jog your memory enough | 13:19 |
| gnarface | what you might have done is set your ISP's DNS servers as the primary resolvers manually | 13:19 |
| gnarface | but off the top of my head i can't be sure there isn't some line to just tell it to refer to the system's resolv.conf | 13:20 |
| gnarface | i just know it doesn't do that by default | 13:20 |
| gnarface | and hasn't, as far back as i can remember | 13:20 |
| gnarface | but i've never actually installed it on a system with a dhcp client, a network manager, or avahi-daemon | 13:22 |
| gnarface | so maybe ymmv | 13:22 |
| freem | <gnarface> you could try just building it and packaging it and installing it... | 14:03 |
| freem | Actually, I had read the question as: "is it possible to install multiple inits on a system at the same time, and to have the system to still work without too much work" and my answer would be: no. Because then there won't be any integration, and I suspect things like sysvinit-core will fight against such process. Now, I had not answered because 1) asker left very quickly and 2) I could not turn it into an answer I'd be happy with, there's too much | 14:03 |
| freem | guessing and suppositions in my "answer" | 14:03 |
| freem | I agree on the agenda thing though, even if I didn't thought about it | 14:03 |
| gnarface | well, the thing is they're barking up the wrong tree | 14:06 |
| freem | well, technically having several init installed is just a matter of not naming the binaries "/sbin/init" but making it a symlink pointing at them, but the integration would still miss | 14:07 |
| freem | amusing expression "barking up the wrong tree" :) | 14:07 |
| gnarface | i know of the several that are in the repos, there is some overlap because sysvinit isn't the only one that relies on sysvinit's init scripts | 14:07 |
| freem | in any case, when *I* wanted to toy with other inits I got curious about, I tried, instead of asking, and that certainly worked nicely :) | 14:08 |
| freem | that dependency though is probably related to the fact people re-use old rc.d scripts to inject them in init alternatives, as what is done for runit? It works, but is a relatively poor solution I think (which is why I do the effort myself, it's easy to write a runit service after all). | 14:09 |
| freem | reminds me I need to do that for chrony... | 14:10 |
| gnarface | yea, openrc does that too. it's just a set of useful defaults, you can still use them the normal way other distros do, but debian wanted to reduce effort of replication... | 14:10 |
| gnarface | it does tend to throw people off though, when they're expecting the same experience as some other distro that defaults to their particular chosen init | 14:11 |
| freem | well, I understand taht effort, but I think the way it's done is not exactly clean. Now, I ain't contributing nor talking with maintainers, so should just shut up, obviously. | 14:11 |
| gnarface | (also historically not an uncommon situation with "debianization") | 14:11 |
| freem | " when they're expecting the same experience as some other distro that defaults to their particular chosen init" fun fact, I consider void-linux's runit integration quite poor, if that have not changed since last time I tried | 14:12 |
| freem | I don't know for other init systems and distros, thuogh, nor do I know what people might expect from an init + service manager (since this is really what we are talking about here, the framework, and not just the init) | 14:14 |
| freem | hm... but should not we continue this offtopic, thuogh? | 14:14 |
| freem | if there's more to say, that is. | 14:14 |
| freem | oh, that chronyd's -F option seems like a rabbit hole... what is this seccomp facility now, and how could I use it... from chronyd's man-page, it sounds like something programmers should integrate within their programs. I guess it's not portable though? It kind of looks like those *BSD facilities I heard about, though, from what I can get (without reading much yet) | 14:18 |
| amarsh04 | gnarface, with some more testing I found that bind9 1:9.18.28-1~deb12u2 worked with resolvconf, and bind9 1:9.20.4-3 does not call resolvconf | 14:23 |
| freem | apparently, chronyd double forks, and there's no way to avoid that? | 14:23 |
| freem | maybe should use ntp or openntpd packages instead... | 14:24 |
| freem | ntpsec is written in python lol, probably because of some "managing memory is unsafe" fashion stance | 14:25 |
| freem | guess it will be openntpd then, man-page is clear as to how to avoid the ugly double-fork trick | 14:30 |
| freem | heh. Another trivial runit script done in few secs. I like runit and daemons which have a clear doc about if they double-fork or not, and allow to disable it when they do. | 14:33 |
| amarsh04 | reported the bind9 not calling resolvconf to Debian: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1090804 | 14:35 |
| Hurgotron | freem: I tried ntpd and openntpd on systems which get suspended (laptops...) and they stop working sooner or later. chrony has been reliable in my usecase. On 24/7 running machines I use openntpd | 14:39 |
| freem | are you not contradicting yourself? " openntpd ... stop working sooner or later. chrony has been reliable in my usecase. ... I use openntpd"? | 14:40 |
| freem | Hurgotron: also, what do you mean, stopped working? | 14:41 |
| Hurgotron | No? "On 24/7 running machines I use openntpd" ie, not laptops | 14:41 |
| freem | ah | 14:41 |
| freem | got it | 14:41 |
| Hurgotron | I think the deamon died... maybe didn't like switching networks | 14:41 |
| freem | that computer is my desktop, definitely not running 24/7, but chronyd's doc is a bit lacking, so I don't feel too trusty about it | 14:41 |
| freem | ah | 14:41 |
| freem | well, crashes or the likes will be handled by runit I guess. It's not critical anyway, it's just about not having stupid TLS "unsafe area" all the time when browsing the web... | 14:42 |
| Hurgotron | But I also had issues on my desktop, which I also hibernate. It has a fixed ipv4, but I guess my ipv6 prefix chansged every now and then, so probably the same cause | 14:43 |
| freem | ah, yeah, seems likely | 14:44 |
| Hurgotron | actually I don't like autorestart on daemons. If they crash they should be fixed, if hey don't crash I don't need autorestart | 14:44 |
| freem | I don't hibernate though, system boots faster with cold boots than with hibernate/wakingup | 14:44 |
| freem | I still think it's weird openntpd would crash on such events? But maybe. | 14:45 |
| Hurgotron | ah, not hibernate, I suspend the desktop too. Keep confusing it. "suspend to RAM" is more clear | 14:46 |
| freem | I mean, openbsd have a rather good reputation in security, and crashes are not exactly what i'd expect (well, better a crash than a hole, sure) | 14:46 |
| freem | I see | 14:46 |
| Hurgotron | freem: Not sure if it crashes, but at some point I wondered why the system time is off and found that the deamon was not running anymore | 14:47 |
| freem | well, I guess I shuold be fine, since I just call /sbin/poweroff... should suspend-to-ram more often, really | 14:47 |
| freem | but I never think about doing this | 14:47 |
| freem | Hurgotron: thx for feedback though, it's good to know. Why not run chronyd on those 24/7 systems, too? Just to avoid fixing what is not broken, or is there is another reason? | 14:55 |
| Hurgotron | basically it's what I always did, the switch to chrony on the other systems is somewhat recent (this year, ISTR) | 14:56 |
| freem | got it, and makes sense | 14:56 |
| Hurgotron | But I have to use chrony at work (RHEL & co) so I know what to do with it, too. | 14:57 |
| freem | well, my only use case is, as said, to still be able to browse the web when a computer ran off batteries or drifted too much | 14:58 |
| freem | since asus thought it's a smart idea to remove the cmos battery on laptops... | 14:58 |
| freem | there really are smart people around, sad that it's not a requirement to design systems... | 14:59 |
| Hurgotron | reminds me of foone's rant about 3 key rollover on some keyboards... | 15:02 |
| Hurgotron | but, it's offtopic. Maybe I should go there | 15:03 |
| freem | previously I'd just use ntpdate, really, but someone replaced the fine native implementation with a python one "for security" I presume. | 15:03 |
| freem | indeed, we should | 15:03 |
| gnarface | amarsh04: could it have been a feature pulled out by a security patch? | 15:38 |
| yeti | I use openntpd instead of that python thing. | 15:47 |
| amarsh04 | gnarface, could be something not properly tested with each build - I didn't know that it had failed until I needed to bypass my ISP's DNS | 16:09 |
| freem | hm... I wonder how I could suspend in ram without requiring root access (i.e. using suid like I did for /sbin/poweroff)? pm-suspend leads to some shell scripts relying on some shell library which itself includes other shell libs, sysvinit+rc.d hellish's style :/ | 17:47 |
| freem | that shell-lib mess is what drived me to try systemd, long ago, and why I did the effort to move to runit, as well | 17:48 |
| freem | and ofc, making the shell script itself suid does not helps | 17:49 |
| freem | perhaps I should just write my own pm-* tools directly from the kernel interface... it might end up being easier than fighting the system | 17:50 |
| freem | Hurgotron: btw, about naming for this stuff, https://www.kernel.org/doc/html/latest/admin-guide/pm/sleep-states.html uses the terms S2RAM and STD. I think S2RAM and S2DISK with the proper context would more explicit than hibernate and suspend. I'll probably start using those, myself. | 17:53 |
| freem | yep, writting a small C utility which just writes in /sys/power/state is definitely going to be easier than fighting against pm-utils's annoying restrictions | 17:58 |
| freem | and code will be much simpler to understand, as well! | 17:58 |
| freem | oh! Now I understand! pm-utils is a freedekstop.org project :D | 17:58 |
| * freem wonders if there is a tool to count the full KLoC of a shell script, including sourced "shell libraries" | 17:59 | |
| freem | oh! Hurgotron: https://cgit.freedesktop.org/pm-utils/tree/README "* Suspend/resume functionality can be easily modified by installing files into the /etc/pm/sleep.d directory. [...] re-syncing the time with ntp" maybe chronyd puts a file in /etc/pm/sleep.d which allows it to detect a suspended/hibernated state and thus resync without problems? | 18:10 |
| freem | although I don't see anything in `apt-file show chrony` which suggests any direct relation with pm-utils, and similarly for the dependences. Maybe there is some other kind of dark magic at work within other tools, though: there is a "suggests" for both dnsutils and networkd-dispatcher on debian. | 18:14 |
| Hurgotron | I would guess openntpd just doesn't expect the network to change and doesn't know how to handle it, and chronyd reconnects. | 18:21 |
| Hurgotron | The "run script before/after suspend to RAM" is a different can of worms. Took me a while to find the solution, and it's NOT the freedesktop one you cited. At least not for Devuan | 18:22 |
| freem | I am failing to imagine a use-case for that honestly. Do you have one? | 18:24 |
| freem | and what was your solution, too :D (might be handy if the question rises someday, I'd have something to answer) | 18:25 |
| freem | (I mean, I would have a relatively standard solution, instead of the "use my own C hack" solution I will likely have in few hours because I'll get rid of pm-utils) | 18:26 |
| Hurgotron | first, the scripts need to go into /lib/elogind/system-sleep and /lib/elogind/system-shutdown in order to work | 18:27 |
| freem | ok, you use elogind to handle sessions, and I guess it's able to interact with pm-utils, that makes a lot of sense | 18:28 |
| Hurgotron | And I need it because I run slock as a screenlocker, and for some reason that fails just with one laptop. So I call it manually | 18:28 |
| Hurgotron | The other use case was to restore keyboard settings, which usuall get reset by waking up | 18:29 |
| freem | really? | 18:29 |
| Hurgotron | I was a bit surprised too | 18:30 |
| freem | with a pm-suspend, keyboard layout config is lost? | 18:30 |
| freem | I have always seen pm-suspend as a way to tell kernel: "ok, now, enter a while(1){sleep(1);} loop, please" | 18:31 |
| Hurgotron | ah, no. First, this is for xfce. And I man special settings like switch , to . on num pad, and disable caps lock | 18:31 |
| Hurgotron | the tweaks are lost | 18:31 |
| freem | "switch , to . on num pad" ah, french user perhaps? I am totally interested in disabling the damn stupid idiocy of making '.' generate ',' on numpad... it is VERY annoying when using, say, gnumerics or calc. How did you did that? | 18:32 |
| Hurgotron | also, my xset settings (screenblank...) get randomly reset. No idea why that happens so far | 18:32 |
| Hurgotron | German user | 18:33 |
| freem | to me, this idea is as stupid as gitlab translating technical terms such as "commit", "tree", etc | 18:33 |
| freem | ah, you guys also use ',' to split integer and decimal parts of numbers? | 18:33 |
| Hurgotron | right :) | 18:35 |
| freem | good to know french users are not the only annoyed ones by this hehe: sharing the misery. | 18:36 |
| freem | although it's weird, with the layout I use, for *most* softwares numpad '.' does what I want it to do... only those "productivity suits" behave stupidly | 18:37 |
| Hurgotron | https://file.io/plrVfOOLzpEL | 18:37 |
| Hurgotron | that is my /lib/elogind/system-sleep/99_keyboardmods.sh | 18:37 |
| freem | that is, openoffice and gnumerics. I'm actually almost certain MS excel does NOT | 18:37 |
| Hurgotron | ah of copurse i shared the wrong file | 18:38 |
| freem | hehe | 18:38 |
| Hurgotron | well the startscript part is still interesing | 18:39 |
| Hurgotron | 6Wm/download/plrVfOOLzpEL | 18:40 |
| Hurgotron | gods I hate this clipboard | 18:40 |
| Hurgotron | https://www.file.io/0ZtC/download/pfKbBRibZATa there. | 18:40 |
| Hurgotron | ah leaked my username. whatever. (hardcoded becasue I'm lazy) | 18:41 |
| Hurgotron | freem: I do networking and I often have to input IP addresses... and then you have a comma on the numpad. | 18:42 |
| freem | error while retrieving :) | 18:43 |
| freem | Hurgotron: that is my uploader (ugly) script: % upl ~/.bin/upl -k | 18:43 |
| freem | https://p.mort.coffee/onh | 18:43 |
| freem | might help you :p | 18:43 |
| freem | <Hurgotron> freem: I do networking and I often have to input IP addresses... and then you have a comma on the numpad. | 18:45 |
| freem | well, ok, but I do my share of networking as well. Never had this problem in terminals, and I use terminals as my main toolset: file exploration, versionning, text editing... | 18:45 |
| freem | and ofc ssh-ing :) | 18:45 |
| Hurgotron | default german keyboard layout... | 18:46 |
| Hurgotron | I could switch to US, but then I'm missing umlauts | 18:46 |
| Hurgotron | and we should switch back to offtopic, didn't notice this was the other channel. | 18:47 |
| freem | that is where it is weird, really. default french keyboard layout is missing this bug :D | 18:47 |
| freem | yep, but originally we were on topic, it was about support for the problem of i18n of numpad | 18:48 |
| cousin_luigi | The default French keyboard was invented by the Devil. | 19:48 |
| freem | perhaps, but it's better than qwerty to code | 19:53 |
| freem | https://upload.wikimedia.org/wikipedia/commons/thumb/d/da/KB_United_States.svg/400px-KB_United_States.svg.png | 19:55 |
| freem | look at all those symbols you poor qwerty users *have* to use more than one finger to get... above numbers, mostly. Many of those are direct access on azerty. I'll stop my off-topic here though :) | 19:56 |
| freem | well, qwerty users can reach [] without combination, but not parens, what a mess. Parens are *far* more useful. | 19:56 |
| freem | oops. | 19:56 |
| jonadab | Honestly, the biggest problem with standard QWERTY keyboard layouts, is that the main buckies (shift, ctrl, alt) are all much too inconvenient, way out at the edges of the keyboard. In an ideal layout, the spacebar would be split, so that one thumb can hit space and the other can hit shift, and then ctrl and alt would probably be home-row positions for the pinkies... | 20:06 |
| jonadab | But the main part of that (splitting the spacebar) requires different keyboard *hardware* so that is obviously not happening. | 20:06 |
| djph | freem: it's not like azerty has more buttons ... | 20:14 |
| freem | djph: indeed, but azerty put the numbers where they need shift to be reached, when you lack a numpad | 20:19 |
| djph | ah, that'd get annoying quick. | 20:20 |
| freem | now, when coding, you rarely use numbers except 0 | 20:20 |
| freem | because it's almost always better to use named constants instead | 20:20 |
| freem | why would that be annoying quick? | 20:20 |
| gnarface | come on guys, take it to #devuan-offtopic | 20:20 |
| freem | that ^ is true | 20:21 |
| freem | gnarface: to my defense, I was pinged in here :) | 20:21 |
| freem | arguable argument, I know | 20:21 |
| djph | gnarface: sorry, misread the tab :( | 20:21 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!
