💾 Archived View for gemini.ctrl-c.club › ~phoebos › logs › kisslinux-2021-12-17.txt captured on 2024-05-26 at 16:11:35.
⬅️ Previous capture (2022-01-08)
-=-=-=-=-=-=-
[2021-12-17T00:02:09Z] <noocsharp> only a masochist would try to implement POSIX
[2021-12-17T00:02:31Z] <midfavila> i never said I wasn't a masochist
[2021-12-17T00:02:40Z] <phoebos> midfavila: getopt(3)
[2021-12-17T00:02:49Z] <midfavila> yeah, I know that that exists now
[2021-12-17T00:03:00Z] <phoebos> also, i'm [slowly] doing the same, albeit with different approaches to you
[2021-12-17T00:03:18Z] <midfavila> a userland? nice
[2021-12-17T00:03:57Z] <phoebos> https://git.bvnf.space/bore/
[2021-12-17T00:04:00Z] <midfavila> i need to do some work on my implmentation of yes
[2021-12-17T00:04:14Z] <phoebos> lol
[2021-12-17T00:04:15Z] <midfavila> it could use a buffer instead of a loop to produce more output per syscall...
[2021-12-17T00:04:18Z] <phoebos> is yes in SUS
[2021-12-17T00:04:25Z] <phoebos> why do you need yes to be fast
[2021-12-17T00:04:27Z] <midfavila> it's in posix which is a subset of SUS
[2021-12-17T00:04:32Z] <midfavila> and yes is good as a benchmark
[2021-12-17T00:04:39Z] <phoebos> it's not in POSIX
[2021-12-17T00:04:40Z] <midfavila> or for generating lots and lots of garbage data quickly
[2021-12-17T00:04:43Z] <midfavila> uh
[2021-12-17T00:04:46Z] <midfavila> i'm 99% sure yes is posix
[2021-12-17T00:05:28Z] <midfavila> well shit, I guess it's not
[2021-12-17T00:05:31Z] <midfavila> fuck me I guess
[2021-12-17T00:05:53Z] <noocsharp> a good start to implementing posix would be to figure out what is posix
[2021-12-17T00:06:08Z] <bibliocar> um, cat /dev/rand
[2021-12-17T00:06:14Z] <phoebos> ^
[2021-12-17T00:06:20Z] <phoebos> how often do you use yes anyway
[2021-12-17T00:06:26Z] <midfavila> fairly often
[2021-12-17T00:06:32Z] <phoebos> what for
[2021-12-17T00:06:32Z] <midfavila> as in "multiple times a day"
[2021-12-17T00:06:40Z] <midfavila> i use it to forcefully initialize my laptop's cellular modem
[2021-12-17T00:06:53Z] <midfavila> through eldritch devfs magic
[2021-12-17T00:07:08Z] <phoebos> ok. why not something like cat /dev/rand
[2021-12-17T00:07:40Z] <midfavila> anyway, i've just been consulting s/ubase and the posix manpages
[2021-12-17T00:07:40Z] <midfavila> and /dev/random can be fine for producing binary data, but sometimes I don't want that
[2021-12-17T00:07:59Z] <midfavila> sometimes I specifically want to output a fuckton of a single string
[2021-12-17T00:08:11Z] <phoebos> while : ; echo y; done
[2021-12-17T00:08:19Z] <midfavila> significantly less efficient
[2021-12-17T00:09:32Z] <phoebos> like 500KiB/s, which is a little bit faster than my eyes can go
[2021-12-17T00:09:44Z] <phoebos> i still don't understand why you need something faster
[2021-12-17T00:10:28Z] * midfavila shrugs
[2021-12-17T00:10:37Z] <midfavila> you don't need to understand. it's just a personal preference.
[2021-12-17T00:11:23Z] <noocsharp> the lisp guy is concerned about the performance of yes now?
[2021-12-17T00:11:31Z] <bibliocar> yes was the first terminal command I learned. I had to turn off my computer because I didn't know ctrl-z
[2021-12-17T00:11:47Z] <midfavila> >implying lisp isn't a performant language
[2021-12-17T00:11:50Z] <midfavila> i'll literally fight you
[2021-12-17T00:12:03Z] <midfavila> meet me in the mcdonalds parking lot after class, nerd
[2021-12-17T00:12:35Z] <noocsharp> find me a lisp that can produce a faster yes than c, without writing your own compiler
[2021-12-17T00:12:50Z] <phoebos> lisp io haha
[2021-12-17T00:13:06Z] <phoebos> anyway, man 1p for posix
[2021-12-17T00:13:16Z] <phoebos> regcomp and regexec are really nice
[2021-12-17T00:13:23Z] <phoebos> made writing a basic grep(1) super easy
[2021-12-17T00:14:51Z] <midfavila> i'm willing to argue that because most if not all major operating systems today are written primarily in C and would therefore have a lower runtime overhead than programs written in other, non-native languages, that's not a fair comparison, noocsharp. even still, in terms of just speed, SBCL is likely competitive
[2021-12-17T00:14:52Z] <midfavila> yeah I know how to reference the posix manpages
[2021-12-17T00:15:47Z] <midfavila> there's also stalin scheme
[2021-12-17T00:15:55Z] <midfavila> but idk if that would even run on modern systems
[2021-12-17T00:16:01Z] <noocsharp> runtime overhead has nothing to do with what language the system is written in
[2021-12-17T00:17:50Z] <noocsharp> you could write a kernel in lisp, and c programs would still run faster, because c is easier to translate to machine code
[2021-12-17T00:18:20Z] <noocsharp> you can't construct functions on the fly in c
[2021-12-17T00:18:28Z] <noocsharp> you can in lisps, and this has runtime overhead
[2021-12-17T00:20:11Z] * midfavila shrugs fair enough then
[2021-12-17T00:22:23Z] <midfavila> anyway, even if lisp isn't going to beat C in some, or even many contexts regarding raw speed, it has other advantages that I consider valuable enough to negate the speed penalty in my limited experience, such as its simpler syntax and superior support for functional programming
[2021-12-17T00:24:12Z] <bibliocar> https://opencores.org/projects/igor what if lisp is the machine language?
[2021-12-17T00:25:35Z] <noocsharp> well if you wrote c for the underlying risc isa, it would probably be faster
[2021-12-17T00:25:39Z] <noocsharp> but that's an interesting project
[2021-12-17T00:26:16Z] <midfavila> there've been a few implementations of lisp in hardware over the years other than lispms
[2021-12-17T00:26:24Z] <bibliocar> sorry, I was making a joke. I don't know what functional programming is?
[2021-12-17T00:26:54Z] <midfavila> functional programming is, as I understand it, a programming paradigm in which the global state of variables doesn't change
[2021-12-17T00:27:02Z] <midfavila> once a value is bound to a variable it doesn't change
[2021-12-17T00:27:41Z] <midfavila> so things like while(x < 10){foo();x++} are impossible
[2021-12-17T00:28:31Z] <midfavila> instead you would need to write a recursive function that calls itself in order to iterate. that's a fairly simple example, but it illustrates the core concept, I think
[2021-12-17T00:28:46Z] <bibliocar> but rewriting things in memory is fairly trivial?
[2021-12-17T00:28:54Z] <midfavila> it's not about performance
[2021-12-17T00:29:02Z] <midfavila> it's about mechanical simplicity
[2021-12-17T00:29:17Z] <noocsharp> midfavila: i think you can do something like your while example in lisp
[2021-12-17T00:29:23Z] <midfavila> functional, recursive procedures are often slower than their iterative counterparts, but once you understand them, it can sometimes be much easier to work with
[2021-12-17T00:29:26Z] <midfavila> and yes, you can, very easily
[2021-12-17T00:29:31Z] <midfavila> because lisp isn't a functional language
[2021-12-17T00:29:39Z] <noocsharp> no i mean a literal translation
[2021-12-17T00:29:51Z] <midfavila> oh. well, i mean, considering there's a loop construct, yeah
[2021-12-17T00:29:52Z] <noocsharp> you can modify variable values
[2021-12-17T00:29:55Z] <midfavila> but I don't use it
[2021-12-17T00:30:14Z] <midfavila> admittedly I have much less experience in Lisp than C as of late, which I need to correct...
[2021-12-17T00:30:56Z] <phoebos> i started trying to do advent of code in scheme
[2021-12-17T00:31:00Z] <phoebos> it's been difficult
[2021-12-17T00:31:13Z] <phoebos> haven't had time over the last couple of weeks either
[2021-12-17T00:31:17Z] <midfavila> if you're not used to lisp then I imagine it would be
[2021-12-17T00:31:22Z] <midfavila> i should probably take a crack at AoC
[2021-12-17T00:31:30Z] <midfavila> feel like I'd get filtered pretty early though
[2021-12-17T00:32:10Z] <noocsharp> i'm trying to get my language to a point where i can do advent of code in it
[2021-12-17T00:32:19Z] <midfavila> that would be pretty cool
[2021-12-17T00:32:20Z] <phoebos> that is cool
[2021-12-17T00:32:36Z] <midfavila> do AoC in bc
[2021-12-17T00:34:11Z] <dilyn> do AoC in MtG trading cards
[2021-12-17T01:58:37Z] <Tom10> I am running into ruby compiling issue (I need it for webkitgtk): https://pastebin.com/2yr3Mshx is anyone else running into this issue?
[2021-12-17T02:10:03Z] <dilyn> that paste doesn't include the actual error
[2021-12-17T02:10:13Z] <dilyn> try updating to the latest version (3.0.3?)
[2021-12-17T02:10:23Z] <dilyn> i had an issue building ruby earlier today and bumping fixed it for me lol
[2021-12-17T02:14:11Z] <Tom10> Thank you! Bumping fixed it lol
[2021-12-17T02:53:11Z] <prestonpan> hey
[2021-12-17T02:54:02Z] <Tom10> hi
[2021-12-17T02:55:49Z] <prestonpan> having some problems where after installation of kiss linux and I reboot, grub works fine until I try loading the kernel. It hangs on "loading linux 5.15.8 ..."
[2021-12-17T02:56:39Z] <prestonpan> and i'm using uefi and followed the installation guide on kisslinux.org
[2021-12-17T02:56:46Z] <Tom10> hello I had the same problem earlier in my case it were my gpu drivers, try setting nomodeset kernel boot parameters and see if it boots correctly. If it does then it's probably your gpu drivers
[2021-12-17T02:57:05Z] <prestonpan> that makes sense, i'll try chrooting back into the install rn
[2021-12-17T02:59:17Z] <prestonpan> only thing is my gpu does not have proprietary drivers, is it still possible to face the same problem as you when that is the case?
[2021-12-17T02:59:47Z] <Tom10> Yes, my gpu is rx 550 and I use amdgpu drivers which are open source
[2021-12-17T03:00:25Z] <Tom10> I had to include the firmware for my gpu directly into the kernel, the gentoo wiki explains how to https://wiki.gentoo.org/wiki/AMDGPU#Incorporating_firmware
[2021-12-17T03:01:36Z] <prestonpan> cool, i'll try setting some boot parameters then
[2021-12-17T03:01:43Z] <prestonpan> I could probably just do it temporarily first
[2021-12-17T03:02:53Z] <prestonpan> still hangs after I add `nomodeset` to boot parameters :(
[2021-12-17T03:03:03Z] <prestonpan> heavily unlucky
[2021-12-17T03:04:55Z] <prestonpan> that is to be expected I think because my friend also had the same problem and he said that it had something to do with grub
[2021-12-17T03:05:06Z] <prestonpan> he installed kiss 6 months ago so he couldn't help me
[2021-12-17T03:09:47Z] <Tom10> try compiling the kernel with a big configuration, make sure you enabled support for video drivers stuff, your filesystems and such
[2021-12-17T03:10:21Z] <testuser[m]> Hi
[2021-12-17T03:10:26Z] <midfavila> hey testuser[m]
[2021-12-17T03:10:38Z] <midfavila> i'd suggest stealing a configuration from something like puppy linux if you were going to try that approach
[2021-12-17T03:11:31Z] <prestonpan> i'll try recompiling my kernel
[2021-12-17T03:11:58Z] <prestonpan> i just used the default configuration initially
[2021-12-17T04:46:30Z] <Featyre> I made a patch, however I could get the patch to work will build file
[2021-12-17T04:46:45Z] <Featyre> It just said file not found
[2021-12-17T04:47:27Z] <Featyre> Nvm I need to add it to sources
[2021-12-17T04:53:35Z] <acheam> midfavila: want to hear something atrocious?
[2021-12-17T04:58:10Z] <Featyre> I did it, I finally got cryptsetup 2.4.2
[2021-12-17T04:58:33Z] <Featyre> Working under POSIX shell only using kiss without bash
[2021-12-17T04:59:52Z] <Featyre> Well I had to use illiliti's script and my patch to work but I will upload the repo to GitHub eventually
[2021-12-17T05:14:13Z] <acheam> nice
[2021-12-17T05:44:30Z] <noocsharp> what's atrocious acheam?
[2021-12-17T06:14:09Z] <noocsharp> i'll tell you what's atrocious
[2021-12-17T06:14:16Z] <noocsharp> the shear amount of built-in functions that zig has
[2021-12-17T06:14:21Z] <noocsharp> sheer*
[2021-12-17T11:04:44Z] <bibliocar> So, on the topic of a 2d api for graphics cards, it seems openvg is very nice, and was at one time part of gallium. Could probably shrink mesa down to a size that only supports gallium drives, openvg, and va-api(video acceleration)
[2021-12-17T11:07:56Z] <Featyre> I found out that tinyramfs-git from kiss-community has issues
[2021-12-17T11:08:07Z] <Featyre> I fixed it up
[2021-12-17T16:04:08Z] <phoebos> Featyre: consider creating an issue or emailing the maintainer
[2021-12-17T16:04:58Z] <Featyre> I'm considering since I am planning to restart my KISS install
[2021-12-17T16:08:56Z] <Featyre> If I do, I'd be sure to upload my repo before formatting the partition
[2021-12-17T19:27:50Z] <sad_plan> o/
[2021-12-17T19:37:12Z] <midfavila> acheam sure i guess
[2021-12-17T19:37:16Z] <midfavila> try to keep it brainlet level
[2021-12-17T19:37:26Z] <midfavila> didn't sleep last night
[2021-12-17T19:52:13Z] <phinxy> GCC compiled firefox with software rast performed poorly with -O2, but -O3 made 720p play back, just. Could not manage to get clang to compile firefox, weird. Tried 92.0 and 94.0, the error was: https://termbin.com/chxv
[2021-12-17T19:55:13Z] <phinxy> Trying to get the ancient GPU to function under firefox is a dead end, I think mozilla dropped opengl2.0 support in favor for 3.0. swaywm recognizes the gpu though. http://termbin.com/e9ax
[2021-12-17T20:21:11Z] <dilyn> phinxy: export CFLAGS="$CFLAGS -I/usr/include/c++/v1" should fix it iirc
[2021-12-17T20:23:06Z] <phinxy> Would a alternative c++ standard library have to be installed for clang or does it use the same as GCC?
[2021-12-17T20:24:05Z] <dilyn> perhaps for building with clang to 'just work', i'm not sure
[2021-12-17T20:25:06Z] <dilyn> such an export doesn't have to be done on wyverkiss, which makes me think it's a "using clang with libstdc++" thing
[2021-12-17T20:25:21Z] <dilyn> which seems wrong, but I'm uncertain what the details are which would make it so:(