💾 Archived View for gemini.ctrl-c.club › ~phoebos › logs › kisslinux-2022-04-19.txt captured on 2024-05-10 at 14:14:48.
⬅️ Previous capture (2022-04-28)
-=-=-=-=-=-=-
[2022-04-19T00:41:07Z] <deathmist> guess I did something wrong by choosing a user over 8 chars in length.. the ls output truncates the rest and am_owner in /usr/bin/kiss depends on this and so I can't e.g. "kiss u" as my regular user. not sure if this is just a glasnost issue of also present on base KISS
[2022-04-19T00:43:23Z] <deathmist> I feel like this should be fixed to use stat directly instead of ls which appears to not truncate the username
[2022-04-19T00:45:41Z] <deathmist> so what would be the preferred way to patch this: add the patch in https://github.com/kiss-community/repo/tree/master/core/kiss or PR to https://github.com/kiss-community/kiss?
[2022-04-19T02:12:29Z] <jedavies> deathmist: have seen this before, think it's a bug in kiss. Will check logs
[2022-04-19T02:19:06Z] <humaid_> Hi
[2022-04-19T02:21:42Z] <humaid_> I'm trying mount an SD card, on my laptops internal card reader. It's mounted, but I can't create files in it.
[2022-04-19T02:22:13Z] <humaid_> mount says: Read-only filesystem.
[2022-04-19T02:23:36Z] <humaid_> "mount -t vfat -o remount,rw /dev/sdb1 /mnt/usb" didn't change anything. SD drivers are enabled(I think) in kernel.
[2022-04-19T02:27:36Z] <testuser[m]> Hi
[2022-04-19T02:44:45Z] <humaid_> Any thoughts?
[2022-04-19T07:34:43Z] <deathmist> humaid_: anything maybe in dmesg? SD cards also can have a read-only lock so check that isn't on
[2022-04-19T07:45:38Z] <deathmist> also afaik you're giving mount an extra arg, try just "mount -t vfat -o remount,rw /mnt/usb"
[2022-04-19T09:10:51Z] <deathmist> jedavies: oh hey you are here, awesome work on Glasnost. would you like another maintainer for the project? I'd be down to help with various things currently and have documented all my issues with the setup process and such
[2022-04-19T09:29:51Z] <jedavies> deathmist: yeah contributions are welcome. You can raise issues or PRs on GitHub. Right now I'm testing the upgrade to llvm 14.0.1, getting new tarballs ready etc.
[2022-04-19T09:34:31Z] <jedavies> Including for the hot new architecture everyone will be using this summer: aarch64_be xD
[2022-04-19T09:56:59Z] <deathmist> jedavies: I have one concern about https://github.com/glasnostlinux/kiss which is the way it's updated right now, you should prefer rebases over merging code by hand if at all possible since right now it's almost 500 commits behind upstream while technically being up-to-date with the 5.5.28 upstream tag afaik
[2022-04-19T11:19:29Z] <jedavies> There's lots of changes since 5.5.28 - not sure if these have been tested though? I think regular kiss is still on 5.5.28: https://github.com/kiss-community/repo/blob/master/core/kiss/version
[2022-04-19T11:19:44Z] <jedavies> Dylan did not release a new version for a while
[2022-04-19T11:21:16Z] <jedavies> Anyone here running the latest version of the package manager, or are people using 5.5.28? Any known issues with the latest version?
[2022-04-19T11:25:59Z] <deathmist> jedavies: I mean rebasing to the 5.5.28 tag which is 30 or so commits behind master, not literally using the latest hash
[2022-04-19T11:30:32Z] <jedavies> OK will check later today. btw what arch are you running on? Are you cross building?
[2022-04-19T11:30:49Z] <deathmist> just x86_64 for now without cross building anything
[2022-04-19T11:31:03Z] <deathmist> may take a look at running it on armv7 and aarch64 later both cross-compiled from x86_64
[2022-04-19T11:35:13Z] <illiliti> deathmist: re username truncation: https://github.com/kisslinux/kiss/pull/289
[2022-04-19T11:35:31Z] <illiliti> tldr stat isn't portable, hence can't be used.
[2022-04-19T11:56:21Z] <deathmist> damn... that's the only thing I thought of instantly. guess we'd have to lookup the full username based on UID
[2022-04-19T11:57:31Z] <deathmist> guess I could just get a shorter username, but really, 8 characters in 2022?
[2022-04-19T11:59:02Z] <testuser[m]> lol i just realized my username is exactly 8 ch
[2022-04-19T11:59:18Z] <illiliti> me too xd
[2022-04-19T12:02:09Z] <deathmist> the Linux kernel (and not to mention "useradd") can handle usernames up to 32 characters, I don't think it makes sense going back to 5 decades in logic in this particular case, KISS is a Linux distribution after all right?
[2022-04-19T12:03:21Z] <deathmist> are there modern coreutils or something that breaks with usernames over 8 chars?
[2022-04-19T12:06:36Z] <deathmist> busybox purposefully limits the output to 8 chars which happens to break the kiss package manager in this case https://github.com/mirror/busybox/blob/1_35_0/coreutils/ls.c#L519-L530
[2022-04-19T12:07:56Z] <deathmist> sadly the more limitations I learn about POSIX the more I'm starting to hate it
[2022-04-19T12:08:18Z] <testuser[m]> u can switch to gnu coreutils
[2022-04-19T12:08:20Z] <testuser[m]> for the ls command
[2022-04-19T12:11:23Z] <deathmist> I know there are workarounds but I'm starting to think maybe KISS isn't the right choice for me after all, "going back to 1970's" is nice in some areas but I'm certainly not a fan of still forcing the limitations of yesteryears in a modern Linux environment
[2022-04-19T12:12:35Z] <deathmist> especially if the kernel isn't exchangeable for say BSD and everything is built around a Linux kernel
[2022-04-19T12:14:17Z] <illiliti> kiss(package manager) must be portable
[2022-04-19T12:15:21Z] <testuser[m]> kiss is nothing but a package manager deathmist
[2022-04-19T12:15:29Z] <testuser[m]> u can use any system packages with it
[2022-04-19T12:15:34Z] <testuser[m]> no defaults
[2022-04-19T12:15:46Z] <deathmist> at this point why am I not using LFS
[2022-04-19T12:15:58Z] <testuser[m]> cuz no pkg manager
[2022-04-19T12:16:20Z] <testuser[m]> btw u can just apply the patch to ur kiss locally
[2022-04-19T12:16:46Z] <testuser[m]> https://github.com/kisslinux/kiss/pull/289
[2022-04-19T12:19:33Z] <deathmist> I understand, but if my 9 character username breaks a spec from 1980's while in a modern Linux env it's perfectly valid and supported even by the coreutils user creation I'm not a fan of this ideology, this just should be something I shouldn't have to be dealing with and keep applying a local patch for
[2022-04-19T12:20:04Z] <deathmist> even blocking creation of >8 char usernames would be better since then it at least respects the spec, right?
[2022-04-19T12:20:38Z] <testuser[m]> i really dont see what the problem is here
[2022-04-19T12:20:49Z] <testuser[m]> if u dont agree with the posix spec for any utility
[2022-04-19T12:20:52Z] <testuser[m]> just swap it
[2022-04-19T12:21:17Z] <testuser[m]> dylan can go a bit overboard with portability
[2022-04-19T12:22:47Z] <illiliti> to be clear, kiss package manager targets posix-like systems, not only linux. it can be used on any system with posix utilities right now. locking everything to linux defeats the point of portability
[2022-04-19T12:23:05Z] <illiliti> there's tradeoffs of course
[2022-04-19T12:23:14Z] <testuser[m]> anyway its a busybox issue really, not kiss
[2022-04-19T12:24:11Z] <testuser[m]> even alpine uses busybox
[2022-04-19T12:25:18Z] <illiliti> i wouldn't say it's a busybox issue. busybox does exactly what is allowed by posix
[2022-04-19T12:25:53Z] <illiliti> and we should handle that
[2022-04-19T12:26:02Z] <testuser[m]> ye
[2022-04-19T12:26:05Z] <testuser[m]> thats fair
[2022-04-19T12:27:47Z] <testuser[m]> deathmist: https://termbin.com/bh74
[2022-04-19T13:04:01Z] <deathmist> so $LOGNAME contains the full username, just also tested FreeBSD and there a 9 character user is nothing extraordinary either; $LOGNAME wasn't truncated and even "ls" returns the full username. kiss package manager should be fixed to not rely on output of "ls" since "broken" implementations for it such as busybox exists that truncate the output
[2022-04-19T13:04:08Z] <deathmist> I've got an idea: "ls -ldn $dir" -> get $uid from 3rd field -> "getent passwd $uid | cut -d: -f1" for dir owner username -> compare with $LOGNAME
[2022-04-19T13:04:35Z] <deathmist> works on FreeBSD as well
[2022-04-19T13:05:42Z] <illiliti> getent isn't portable
[2022-04-19T13:07:28Z] <deathmist> ffs, fine let's use grep then it's not like it does much anyway
[2022-04-19T13:08:47Z] <testuser[m]> lol
[2022-04-19T13:09:27Z] <illiliti> grep what? /etc/passwd? we can't assume that /etc/passwd exist
[2022-04-19T13:09:35Z] <testuser[m]> bruh
[2022-04-19T13:09:36Z] <deathmist> I give up
[2022-04-19T13:10:07Z] <deathmist> this mess cannot be fixed
[2022-04-19T13:11:21Z] <deathmist> (portably)
[2022-04-19T13:12:23Z] <testuser[m]> is there an actual platform where stat doesnt exist
[2022-04-19T13:23:40Z] <illiliti> illumos it seems
[2022-04-19T13:23:43Z] <deathmist> are any other coreutils other than gnu and busybox even packaged?
[2022-04-19T13:23:44Z] <illiliti> pretty sure there's more
[2022-04-19T13:23:54Z] <illiliti> also ubase stat doesn't support -c
[2022-04-19T13:24:01Z] <deathmist> I'm not keen on building GNU crap just to get a working package manager
[2022-04-19T13:25:23Z] <illiliti> https://github.com/dilyn-corner/KISS-me/tree/main/core/toybox
[2022-04-19T13:25:39Z] <testuser[m]> deathmist: u could patch busybox ig
[2022-04-19T13:25:40Z] <testuser[m]> the format string
[2022-04-19T13:26:02Z] <deathmist> I could, but I already am hating other aspects of it too so I'd just replace it altogether (with something not GNU if possible)
[2022-04-19T13:26:31Z] <testuser[m]> with toybox even more stuff would need patching
[2022-04-19T13:26:34Z] <testuser[m]> just 2-3 things but still
[2022-04-19T13:26:43Z] <deathmist> eh?
[2022-04-19T13:27:41Z] <deathmist> on a personal note: I once contributed to busybox and however small the patch may have been, my authorship was stripped, so I'd just rather not use any of it
[2022-04-19T13:28:29Z] <testuser[m]> check core/baseinit/patches/toybox.patch
[2022-04-19T13:28:34Z] <testuser[m]> looks like thats the only patch required
[2022-04-19T13:28:42Z] <testuser[m]> i remember some more in his repo but ig they're fixed
[2022-04-19T13:29:17Z] <illiliti> https://github.com/ibara/baseutils
[2022-04-19T13:32:39Z] <omanom> "are any other coreutils other than gnu and busybox even packaged" -- where's mid when ya need them >.>
[2022-04-19T13:36:52Z] <deathmist> illiliti: doesn't appear very maintained, issue about musl build failure from 2020 without comment and PR to fix some segfault open still seemingly without resolution(?)
[2022-04-19T13:38:37Z] <deathmist> not that I need new features or whatever in coreutils often if at all, but fixes are always nice to have
[2022-04-19T13:42:45Z] <illiliti> omanom: mid uses sbase+ubase afaik
[2022-04-19T13:44:44Z] <illiliti> yeah, it seems unmaintained. i just pointed out that it exist :)
[2022-04-19T13:44:57Z] <omanom> acheam did you ever do much with ubase+nbase?
[2022-04-19T13:50:20Z] <testuser[m]> omanom: Hi
[2022-04-19T13:50:55Z] <dilyn> I can't believe someone complained about busybox's `ls` truncating usernames and then suggested we just solve the problem by parsing `ls` output :v
[2022-04-19T13:51:00Z] <dilyn> parsing ls is bad and dangerous
[2022-04-19T13:51:21Z] <testuser[m]> dilyn you';re alive
[2022-04-19T13:51:41Z] <dilyn> been very busy haha
[2022-04-19T13:53:19Z] <illiliti> there's no other way dilyn
[2022-04-19T13:53:39Z] <dilyn> id ?
[2022-04-19T13:53:50Z] <dilyn> testuser: let's be clear, I also don't want to install qt xD
[2022-04-19T13:54:19Z] <dilyn> I think I have to give in at this point though; I need file dialogs now for my work. It's too cumbersome scp'ing to my laptop...
[2022-04-19T13:55:24Z] <illiliti> id isn't intended to grep owner of file
[2022-04-19T13:55:30Z] <illiliti> id basically print uid/username of the current user and all of that things
[2022-04-19T13:58:01Z] <dilyn> is -un not posix?
[2022-04-19T13:58:02Z] <dilyn> f
[2022-04-19T13:58:29Z] <dilyn> it is
[2022-04-19T13:58:39Z] <dilyn> what's the malfunction then? we want to know some other user?
[2022-04-19T13:58:52Z] <testuser[m]> bro just rewrite kiss in compiled language and reimplement broken pozix functionality
[2022-04-19T13:59:09Z] <dilyn> ^
[2022-04-19T13:59:12Z] <dilyn> go to the rescue
[2022-04-19T13:59:19Z] <illiliti> how do you supposed to use id -un to fix this problem?
[2022-04-19T13:59:40Z] <illiliti> we need to get untruncated owner of file
[2022-04-19T13:59:51Z] <dilyn> ah, owner of a file in general
[2022-04-19T13:59:52Z] <dilyn> i see i see
[2022-04-19T14:00:23Z] <illiliti> yes
[2022-04-19T14:00:35Z] <illiliti> we need to somehow exploit the fact that ls can output uid
[2022-04-19T14:00:50Z] <dilyn> why not just always assume root :v
[2022-04-19T14:00:57Z] <dilyn> don't answer that
[2022-04-19T14:05:30Z] <omanom> `python -c 'from os import stat; print(stat("<filename>").st_uid)'` lol
[2022-04-19T14:05:42Z] <dilyn> gotim
[2022-04-19T14:06:20Z] <dilyn> wait do uid:gid get mapped somewhere outside of /etc/{group,passwd}?
[2022-04-19T14:07:56Z] <illiliti> we can't assume that /etc/{group,passwd} exist
[2022-04-19T14:08:11Z] <dilyn> we can if uid and gid are meaningless without them...
[2022-04-19T14:08:59Z] <dilyn> 0:0 doesn't have any actual semantics without /etc/group & /etc/passwd, no?
[2022-04-19T14:09:08Z] <illiliti> posix doesn't define /etc/{group,passwd}
[2022-04-19T14:09:14Z] <dilyn> riiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiight
[2022-04-19T14:09:17Z] <dilyn> of course it doesn't
[2022-04-19T14:09:21Z] <dilyn> what does BSD do? lmao
[2022-04-19T14:10:48Z] <illiliti> essentially, we need something like this: https://termbin.com/q040
[2022-04-19T14:12:02Z] <dilyn> why can't we just fix this in busybox?
[2022-04-19T14:12:09Z] <dilyn> they still accept patches. are they just... obstinate?
[2022-04-19T14:13:20Z] <illiliti> cause it isn't really a bug
[2022-04-19T14:13:31Z] <dilyn> it is one hundred percent a bug
[2022-04-19T14:13:31Z] <illiliti> posix allow this
[2022-04-19T14:14:03Z] <dilyn> smdh
[2022-04-19T14:14:27Z] <illiliti> we can fix this in busybox, but other implementations may have same behavior as busybox
[2022-04-19T14:14:41Z] <dilyn> and they should be fixed too >=|
[2022-04-19T14:14:53Z] <omanom> just get the standard updated duh
[2022-04-19T14:15:59Z] <dilyn> just chown the $file to $(id -u) and be done with it
[2022-04-19T14:16:01Z] <dilyn> you know what
[2022-04-19T14:16:02Z] <dilyn> better
[2022-04-19T14:16:03Z] <dilyn> rm -rf /
[2022-04-19T14:16:09Z] <dilyn> solve every problem
[2022-04-19T14:16:17Z] <testuser[m]> just install windooz
[2022-04-19T14:30:24Z] <agsd> i wanna install kiss linux in a thumb drive but i dont know what configs to enable or disable on the kernel, can someone provide one or tell me where to find a generic one?
[2022-04-19T14:41:59Z] <soliwilos_> agsd: For a generic kernel config, maybe grab alpine's? https://git.alpinelinux.org/aports/main/linux-lts/lts.x86_64.config
[2022-04-19T14:42:42Z] <agsd> soliwilos_ ill give it a try
[2022-04-19T14:45:09Z] <agsd> soliwilos_ it throws an invalid request when trying to open the link
[2022-04-19T14:46:44Z] <illiliti> https://git.alpinelinux.org/aports/tree/main/linux-lts/lts.x86_64.config
[2022-04-19T17:04:48Z] <Guest4854> nickserv identify agsd w5XUgjEoSmGKzk
[2022-04-19T17:05:16Z] <omanom> time to change your password :)
[2022-04-19T17:05:40Z] <Guest4854> i think i fucked lol
[2022-04-19T17:14:26Z] <gst164375>
[2022-04-19T17:14:31Z] <gst164375> a
[2022-04-19T17:15:58Z] <illiliti> hmm, looks like sudo and doas already support uid
[2022-04-19T17:16:10Z] <illiliti> i can add support for uid in ssu, no problem
[2022-04-19T17:16:23Z] <illiliti> so the only problem is su
[2022-04-19T17:17:00Z] <illiliti> LOGNAME usage can be replaced with $(id -un) i guess
[2022-04-19T17:50:15Z] <agsd> im sorry for spamming, im testing birch
[2022-04-19T18:03:44Z] <illiliti> https://termbin.com/e5tv
[2022-04-19T18:04:04Z] <illiliti> this fixes username truncation issue at the cost of removing su support
[2022-04-19T18:04:17Z] <illiliti> ssu will be added
[2022-04-19T18:04:49Z] <illiliti> and perhaps i should submit pull request...
[2022-04-19T19:00:21Z] <phoebos> deathmist: the package manager works fine, you're completely welcome to apply the patch to use stat, but upstream kiss is _portable_, and >8 char usernames aren't. noone's preventing you from having longer than 8 char usernames, just swap out the bits you don't like, it's your system.
[2022-04-19T19:08:27Z] <deathmist> yeah I'd rather swap all of my coreutils, already have my issues with busybox anyway
[2022-04-19T19:10:06Z] <phoebos> ok! kiss is a package manager, KISS comes with busybox but only *by default*
[2022-04-19T19:12:21Z] <phoebos> when/if dylan starts swapping bits of shell out for C, this problem will be easier to fix.
[2022-04-19T19:29:15Z] <Guest80> Hi all! This is probably a mega dumb question, but how should I best handle file conflicts with the kiss package manager? I'm trying to install powertop but it wont install because busybox has its own /usr/bin/powertop file
[2022-04-19T19:29:48Z] <phoebos> in your build file, just install the /usr/bin/powertop file normally
[2022-04-19T19:30:06Z] <phoebos> then once you've installed the package, do `kiss a <packagename> /usr/bin/powertop`
[2022-04-19T19:30:38Z] <phoebos> `kiss a` lists all the packages which provide alternative versions of each file on your system
[2022-04-19T19:30:48Z] <phoebos> you can switch them at will
[2022-04-19T19:33:28Z] <Guest80> TIL about 'alternatives', beautiful! thanks phoebos :)
[2022-04-19T19:33:41Z] <phoebos> :)
[2022-04-19T19:50:05Z] * agsd l
[2022-04-19T20:00:53Z] <omanom> hey agsd check out catgirl as well: https://git.causal.agency/catgirl/about/
[2022-04-19T20:01:26Z] <agsd> what is that?
[2022-04-19T20:02:18Z] <omanom> it is another pretty minimal irc client
[2022-04-19T20:02:30Z] <omanom> kirc as well: https://github.com/mcpcpc/kirc
[2022-04-19T20:04:50Z] <agsd> im using birsh atm lol
[2022-04-19T20:05:07Z] <agsd> the most annoying thing is authenticating with nickserv
[2022-04-19T20:24:36Z] <soliwilos> catgirl is pretty nice.
[2022-04-19T20:25:26Z] * soliwilos meows
[2022-04-19T20:42:40Z] <phoebos> ( ˘ ³˘)♥
[2022-04-19T20:45:26Z] <soliwilos> :D
[2022-04-19T20:46:05Z] <soliwilos> Seems like dylan is reawakening a little.