gemini - kennedy.gemi.dev

💾 Archived View for gemini.ctrl-c.club › ~phoebos › logs › kisslinux-2021-09-28.txt captured on 2024-05-10 at 14:21:14.
-=-=-=-=-=-=-
[2021-09-28T00:03:25Z] <tleydxdy[m]> hmm how would I create system user for a package?
[2021-09-28T00:03:58Z] <tleydxdy[m]> I'm trying to package fcron and it wants a fcron user
[2021-09-28T00:05:11Z] <illiliti> post-install hook
[2021-09-28T00:06:09Z] <illiliti> https://kisslinux.org/wiki/package-system#007
[2021-09-28T00:10:03Z] <illiliti> you can also use pre-remove hook to delete user
[2021-09-28T00:19:56Z] <tleydxdy[m]> yes, but the build process needs that user already
[2021-09-28T00:20:08Z] <tleydxdy[m]> I assume to install some files or something
[2021-09-28T00:26:50Z] <illiliti> patch it out
[2021-09-28T00:27:48Z] <tleydxdy[m]> well, if it's doing what I assume it's doing like creating some file owned by fcron then that wouldn't really work tho
[2021-09-28T00:28:55Z] <tleydxdy[m]> kinda surprised that the build is not done in a chroot tbh
[2021-09-28T00:32:33Z] <acheam> how would you handle dependencies in a chroot tho
[2021-09-28T00:32:49Z] <acheam> I guess you could copy their files into it?
[2021-09-28T00:32:54Z] <acheam> seems io heavy though
[2021-09-28T00:39:00Z] <tleydxdy[m]> you just install the dependency into the chroot
[2021-09-28T00:39:21Z] <tleydxdy[m]> this way you would find out if you left out a dep, as a bonus
[2021-09-28T00:42:52Z] <GalaxyNova> I've switched to seamonkey
[2021-09-28T00:43:00Z] <GalaxyNova> It's nice
[2021-09-28T00:43:38Z] <acheam> as a main web browser?
[2021-09-28T00:43:48Z] <acheam> does it still use an up to date gecko?
[2021-09-28T00:44:07Z] <GalaxyNova> as a main web browser yes
[2021-09-28T00:44:17Z] <GalaxyNova> acheam: IDK but honestly I coudn't care less
[2021-09-28T00:44:29Z] <GalaxyNova> most websites that aren't shit render fine
[2021-09-28T00:49:52Z] <GalaxyNova> I don't know if that's some kind of security vulnerability though..
[2021-09-28T00:49:52Z] <illiliti> acheam: hardlinks or bind mounts can be used to handle deps in chroot without excessive copying
[2021-09-28T00:50:54Z] <acheam> oh yeah, forgot about bind mounts. Hardlinks would still have the problem of duplicated file sizes, though, right?
[2021-09-28T00:54:21Z] <illiliti> iirc hardlink is a symlink at the filesystem level. so there's shouldn't be a problem of dup sizes
[2021-09-28T00:54:34Z] <acheam> oh okay
[2021-09-28T01:04:58Z] <tleydxdy[m]> hardlink causes problem in case of modification
[2021-09-28T01:05:09Z] <tleydxdy[m]> so reflink is needed for that
[2021-09-28T01:06:05Z] <acheam> well they wouldnt be modified
[2021-09-28T01:06:17Z] <illiliti> ^
[2021-09-28T01:09:06Z] <tleydxdy[m]> they would in the case we are talking about here
[2021-09-28T01:14:54Z] <illiliti> i think everything in /etc/ should be just copied
[2021-09-28T01:15:42Z] <illiliti> if you have chromium sources in /etc/, that's your problem
[2021-09-28T01:16:37Z] <tleydxdy[m]> yeah
[2021-09-28T01:17:10Z] <tleydxdy[m]> I mean it shouldn't be that big a problem to make a copy in chroot really
[2021-09-28T01:17:27Z] <tleydxdy[m]> it's just the deps a package needs
[2021-09-28T01:17:31Z] <tleydxdy[m]> not the full system
[2021-09-28T01:17:49Z] <tleydxdy[m]> * just the make deps a
[2021-09-28T01:18:24Z] <tleydxdy[m]> big oof, failed install leaves a dirty system
[2021-09-28T01:18:39Z] <acheam> ?
[2021-09-28T01:18:44Z] <acheam> the make deps can be huge sometimes
[2021-09-28T01:19:06Z] <acheam> llvm, gcc, nodejs, etc
[2021-09-28T01:22:36Z] <illiliti> tleydxdy[m]: kiss isn't smart. it doesn't keep state to remove/fix leftovers if installation is failed
[2021-09-28T01:22:39Z] <tleydxdy[m]> they are also huge project tho
[2021-09-28T01:22:59Z] <acheam> tleydxdy[m]: im missing your point
[2021-09-28T01:23:18Z] <tleydxdy[m]> I mean I understand if the post_install fails or something 
[2021-09-28T01:23:33Z] <tleydxdy[m]> but it could check the tar ball before unpacking it
[2021-09-28T01:24:12Z] <tleydxdy[m]> acheam: the point is to not have the make environment tainted by what's in the system
[2021-09-28T01:24:28Z] <acheam> so you want reproducible builds?
[2021-09-28T01:24:48Z] <tleydxdy[m]> not really that far
[2021-09-28T01:24:50Z] <acheam> having the builds autodetect features based on the build environment is fine IMO 
[2021-09-28T01:24:55Z] <acheam> take the mpv package for example
[2021-09-28T01:24:59Z] <tleydxdy[m]> just to have less hard to pin down make problems
[2021-09-28T01:25:17Z] <acheam> eh
[2021-09-28T01:25:57Z] <tleydxdy[m]> "if it builds on my system, it should build on yours" sort of deal
[2021-09-28T01:26:15Z] <tleydxdy[m]> probably a smart idea to make it configurable or something
[2021-09-28T01:26:24Z] <acheam> on kiss people tend to customize their system so much that it wouldnt make a difference
[2021-09-28T01:26:59Z] <tleydxdy[m]> acheam: how so?
[2021-09-28T01:27:16Z] <tleydxdy[m]> none of the customization would be in the chroot environment 
[2021-09-28T01:27:18Z] <acheam> changing compilers, libc, providers of common utilities, etc
[2021-09-28T01:27:32Z] <acheam> then you would need fully static binaries
[2021-09-28T01:27:43Z] <tleydxdy[m]> no?
[2021-09-28T01:28:12Z] <acheam> if the chroot is running musl and the chroot is running glibc, then the binary built in the chroot wont run on the host if it isnt static
[2021-09-28T01:29:12Z] <tleydxdy[m]> I see what you mean, but in that case the package just depends on "libc"
[2021-09-28T01:29:52Z] <tleydxdy[m]> and whatever package the user choose to provide libc is install into the chroot
[2021-09-28T01:30:20Z] <tleydxdy[m]> the point is not to limit customization
[2021-09-28T01:30:22Z] <acheam> but then it wouldn't be helpful for "if it builds on my system, it should build on yours" type discussions, if people's libc package is different anyway
[2021-09-28T01:30:28Z] <tleydxdy[m]> but to make everything documented and replicable
[2021-09-28T01:31:09Z] <tleydxdy[m]> so if build has a problem on some system, I can replicate it
[2021-09-28T01:32:09Z] <acheam> this solves a really small problem
[2021-09-28T01:32:30Z] <acheam> and it would limit customizations
[2021-09-28T01:32:34Z] <tleydxdy[m]> it's a problem that happens all the time
[2021-09-28T01:33:09Z] <tleydxdy[m]> acheam: in what way?
[2021-09-28T01:34:34Z] <acheam> going back to the libc example
[2021-09-28T01:35:00Z] <acheam> I cant change the libc on my system because under your proposed thing, all the builds should be the same
[2021-09-28T01:35:11Z] <acheam> and I cant have a different libc in the chroot and on the host
[2021-09-28T01:35:30Z] <acheam> and even if you were to use a generic "libc" package, then the reproducibility is gone
[2021-09-28T01:35:58Z] <tleydxdy[m]> you customize your libc how you want
[2021-09-28T01:36:14Z] <tleydxdy[m]> to reproduce it you just need a copy of the repos
[2021-09-28T01:36:19Z] <acheam> this is obviously not going anywhere
[2021-09-28T01:36:22Z] <acheam> if you want it, implement it
[2021-09-28T01:36:22Z] <tleydxdy[m]> not the user's entire system
[2021-09-28T01:37:20Z] <acheam> -e s/this/this discussion/g -e s/e$/e but the idea does have potential/g
[2021-09-28T01:38:44Z] <tleydxdy[m]> lol, I was like "wait what is this kiss chroot
[2021-09-28T01:39:01Z] <tleydxdy[m]> then I realized it's just an artifact of how kiss works
[2021-09-28T01:48:47Z] <tleydxdy[m]> err, why is /usr a thing
[2021-09-28T01:48:54Z] <tleydxdy[m]> wth is /usr/var
[2021-09-28T01:49:27Z] <GalaxyNova> I really wish /usr didn't exist
[2021-09-28T01:51:47Z] <tleydxdy[m]> yeah as I suspected make install tries to change the group of some files to fcron
[2021-09-28T01:51:59Z] <tleydxdy[m]> curiously it did not work
[2021-09-28T01:52:25Z] <tleydxdy[m]> what is installed on my system still has group root
[2021-09-28T01:53:01Z] <tleydxdy[m]> right perm in the tar ball
[2021-09-28T02:00:30Z] <tleydxdy[m]> seem like cp in pkg_install_file is missing a -p
[2021-09-28T02:00:39Z] <tleydxdy[m]> not sure if it's intentional tho
[2021-09-28T02:02:22Z] <tleydxdy[m]> nope, not the fault
[2021-09-28T02:04:06Z] <tleydxdy[m]> err, I'm very confused
[2021-09-28T03:00:08Z] <testuser[m]> Hi
[2021-09-28T03:06:00Z] <acheam> hi testuser[m] 
[2021-09-28T03:06:12Z] <acheam> what brings you here this early today?
[2021-09-28T03:08:39Z] <testuser[m]> Class timings
[2021-09-28T03:27:22Z] <noocsharp> so for the past day, my domain has been pointing at the wrong ip address
[2021-09-28T03:30:35Z] <acheam> lmao
[2021-09-28T03:30:37Z] <acheam> how'd that happen
[2021-09-28T03:31:33Z] <acheam> hmm... dig nihaljere.xyz is timing out for me
[2021-09-28T03:32:33Z] <acheam> and the names not resolving
[2021-09-28T03:34:18Z] <noocsharp> i put a wrong digit in the ip address
[2021-09-28T03:34:26Z] <acheam> oof
[2021-09-28T03:34:35Z] <noocsharp> it resolves fine now for me
[2021-09-28T03:37:37Z] <acheam> weird
[2021-09-28T04:58:32Z] <aspirus> ERROR: 'perl' not found
[2021-09-28T04:59:35Z] <aspirus> nvm it fixed itself
[2021-09-28T05:00:56Z] <aspirus> HI
[2021-09-28T05:01:49Z] <testuser[m]> HI
[2021-09-28T05:02:21Z] <aspirus> Sorry, I didn't introduce myself. I just found out about KISS Linux and in the proceess of installing.
[2021-09-28T05:08:03Z] <aspirus> Should i ask for support here?
[2021-09-28T05:12:28Z] <testuser[m]> Perhaps
[2021-09-28T05:12:40Z] <testuser[m]> Find your terminal first
[2021-09-28T05:13:07Z] <GalaxyNova> lmao
[2021-09-28T05:13:18Z] <GalaxyNova> poor terminals getting lost :(
[2021-09-28T06:36:29Z] <GalaxyNova> Yoooo
[2021-09-28T06:36:31Z] <GalaxyNova> fossil is amazing
[2021-09-28T06:36:41Z] <GalaxyNova> i might just migrate my gitea to just a list of fossil servers
[2021-09-28T06:44:54Z] <testuser[m]> Rich
[2021-09-28T08:22:55Z] <aspirus> I just installed KISS Linux, I can't get Internet to work
[2021-09-28T08:25:09Z] <aosync> What does fossil do better than git ?
[2021-09-28T08:26:05Z] <aspirus> Did i had to enable something during making the kernel to get dhcpcd to use my android tethering for internet
[2021-09-28T08:26:35Z] <aosync> aspirus: could be get more diagnostics? the output of ifconfig, your wireless card model, what you enabled when configuring the kernel?
[2021-09-28T08:27:50Z] <aspirus> Intel Corporation Wireless 7265 (from lspci in void linux where wifi works with iwd and iwctl)
[2021-09-28T08:28:52Z] <aspirus> i didnt edit the default .config apart from what the install guide told
[2021-09-28T08:30:17Z] <aspirus> i copied all folders from the firmware tar to /usr/lib/firmware
[2021-09-28T08:49:18Z] <aosync> ok, and did you add this folder to the firmware folder thing in the kernel config
[2021-09-28T08:49:44Z] <aosync> and then listed the iwlwifi-7265D.ucode in the field of thing to load ?
[2021-09-28T09:42:46Z] <aspirus> aosync: i didnt change anything in the kernel config
[2021-09-28T09:43:02Z] <aosync> yep, you gotta do some stuff
[2021-09-28T09:43:48Z] <aspirus> should i just use void linux 's kernel?
[2021-09-28T09:45:11Z] <aosync> no, configure your own, it's pretty nice
[2021-09-28T09:45:52Z] <aspirus> i didnt understood anything in menuconfig so i just left it along
[2021-09-28T09:46:15Z] <testuser[m]> did you check the wiki
[2021-09-28T09:46:22Z] <testuser[m]> it should have some info about kernel confi
[2021-09-28T09:49:31Z] <aosync> aspirus, mess around in menu config, find out how to specify the firmware dir and your firmware file
[2021-09-28T09:49:53Z] <aosync> https://wiki.gentoo.org/wiki/Iwlwifi
[2021-09-28T10:01:39Z] <aspirus> and after then make again right?
[2021-09-28T10:06:07Z] <aspirus> should i disable other things i dont recognize
[2021-09-28T10:08:08Z] <aspirus> what firmware i need for usb tethering
[2021-09-28T10:15:41Z] <aspirus> rebuilding kernel
[2021-09-28T10:33:12Z] <aspirus> exit
[2021-09-28T12:24:37Z] <testuser[m]> phoebos: https://termbin.com/dxxp
[2021-09-28T12:33:25Z] <phoebos> testuser[m]: thanks! i've never really got the hang of the f* counterparts of read, write, open etc
[2021-09-28T12:33:36Z] <phoebos> could you mention in the commit message that the patch is in the public domain?
[2021-09-28T12:35:30Z] <phoebos> if that's something you want to do
[2021-09-28T12:38:18Z] <phoebos> also could you do the fopen line by itself, then check if (fp != 0)
[2021-09-28T12:39:16Z] <testuser[m]> ok
[2021-09-28T12:39:42Z] <phoebos> thank you!
[2021-09-28T12:40:02Z] <testuser[m]> i like to do all assignments in 1 if statement to reduce clutter, and assignment is explicit since the it is in an extra bracket
[2021-09-28T12:40:11Z] <testuser[m]> s/ the//
[2021-09-28T12:40:11Z] <cotangent> <testuser[m]> i like to do all assignments in 1 if statement to reduce clutter, and assignment is explicit since it is in an extra bracket
[2021-09-28T12:40:36Z] <testuser[m]> is !fp ok ?
[2021-09-28T12:40:44Z] <testuser[m]> or you want == NULL
[2021-09-28T12:41:44Z] <testuser[m]> https://termbin.com/2knbo
[2021-09-28T12:41:46Z] <testuser[m]> fixed return code
[2021-09-28T12:42:17Z] <phoebos> i prefer the verbosity of == NULL, particularly in other functions that aren't so easy to understand with the grammar of !fp
[2021-09-28T12:42:21Z] <f1> is there even any advantage to using FILE * over just read(), write(), etc. and fd's?
[2021-09-28T12:42:24Z] <phoebos> but in this case either is ok
[2021-09-28T12:47:39Z] <phoebos> thanks testuser[m], applied
[2021-09-28T12:51:47Z] <testuser[m]> f1: you can make do with either but i just preferred `while ((c = fgetc(fp))` instead of `if ((read(fd, &c, 1))`
[2021-09-28T12:51:48Z] <testuser[m]> phoebos: np
[2021-09-28T17:03:03Z] <noocsharp> f1: with fread and fwrite, short counts only occur if there's an error
[2021-09-28T17:03:37Z] <noocsharp> so you don't have to loop the way you do with read and write
[2021-09-28T18:52:31Z] <illiliti> https://www.sizeofvoid.org/posts/2021-09-26-openbsd-wayland-report/
[2021-09-28T18:59:56Z] <GalaxyNova[m]> <illiliti> "https://www.sizeofvoid.org/posts..." <- Very excited for that
[2021-09-28T19:00:36Z] <GalaxyNova[m]> Will probably switch back to sway when there's sufficient support
[2021-09-28T19:09:15Z] <illiliti> openbsd devs should implement evdev or convince libinput devs to support wscons
[2021-09-28T19:09:28Z] <illiliti> otherwise openbsd is out of the game
[2021-09-28T19:09:47Z] <GalaxyNova[m]> I would prefer the latter
[2021-09-28T19:09:54Z] <illiliti> me too
[2021-09-28T19:09:59Z] <GalaxyNova[m]> evdev sucks
[2021-09-28T19:10:16Z] <illiliti> indeed
[2021-09-28T19:11:41Z] <illiliti> patching libinput is not bad but upstream support is much better
[2021-09-28T19:58:47Z] <ryoshu> so openbsd wants libudev too
[2021-09-28T20:01:13Z] <aosync> i don't think so
[2021-09-28T20:15:55Z] <illiliti> ryoshu: it's not possible to implement libudev because openbsd doesn't have a way to enumerate devices
[2021-09-28T20:18:18Z] <GalaxyNova[m]> <ryoshu> "so openbsd wants libudev too" <- No