💾 Archived View for gemini.ctrl-c.club › ~phoebos › logs › freenode-kisslinux-2020-09-04.txt captured on 2024-05-12 at 16:07:42.
⬅️ Previous capture (2021-12-17)
-=-=-=-=-=-=-
2020-09-04T02:03:03 #kisslinux <dilynm> dylanaraps: I'd make a GitHub issue but I think this problem is solely of my creation, though I know not how
2020-09-04T03:25:56 #kisslinux <dilynm> I'm sorry that I'm spamming but this got super weird
2020-09-04T03:26:14 #kisslinux <dilynm> konimex: csn you replicate this at all? Mesa is a fine test package
2020-09-04T03:26:35 #kisslinux <dilynm> If I comment out m4 in flex's deps, everything works fine
2020-09-04T03:30:05 #kisslinux <konimex> what to replicate?
2020-09-04T03:32:37 #kisslinux <konimex> you know, it might be a cyclical problem
2020-09-04T03:32:42 #kisslinux <dilynm> Kiss exiting with no prompt to install packages
2020-09-04T03:32:52 #kisslinux <dilynm> Yeah I think it's exactly a problem with m4
2020-09-04T03:33:02 #kisslinux <dilynm> Can you build bsd m4 without flex
2020-09-04T03:33:19 #kisslinux <konimex> it explicitly needs a lex(1) program to build
2020-09-04T03:33:26 #kisslinux <konimex> but I'll see what I can do
2020-09-04T03:34:29 #kisslinux <konimex> does building flex require any m4 implementation?
2020-09-04T03:35:05 #kisslinux <dilynm> Flex has m4 as a dep, m4 has flex as a make dep
2020-09-04T03:35:26 #kisslinux <dilynm> A dirty fix is just dropping one, but that doesn't really solve the actual issue huh
2020-09-04T03:35:33 #kisslinux <konimex> what if you set m4 as a make dep?
2020-09-04T03:35:37 #kisslinux <konimex> in flex
2020-09-04T03:35:42 #kisslinux <konimex> <dilynm "A dirty fix is just dropping one"> yeah
2020-09-04T03:36:05 #kisslinux <dilynm> Setting as make dep works
2020-09-04T03:36:26 #kisslinux <dilynm> Which makes sense
2020-09-04T03:36:27 #kisslinux <konimex> alright
2020-09-04T03:39:06 #kisslinux <konimex> wait, I think I encountered your bug now
2020-09-04T03:39:26 #kisslinux <dilynm> Kiss just needs a way to handle circular dependencies xD
2020-09-04T03:40:54 #kisslinux <konimex> you know, fuck it, I'm removing flex as a makedep from m4, unfortunate situation but necessary
2020-09-04T03:41:08 #kisslinux <dilynm> Lmao
2020-09-04T03:41:42 #kisslinux <dilynm> Just means you'll need a copy of m4 handy if you ever remove it
2020-09-04T03:43:28 #kisslinux <dilynm> Poor bsd m4 is just unbuildable. Behold the superiority of gnu m4
2020-09-04T03:47:16 #kisslinux <konimex> since m4, flex, and byacc are toolchain pkgs anyway, would it make sense that in a system, the packages are expected to be installed anyway?
2020-09-04T03:47:23 #kisslinux <konimex> regardless of deps
2020-09-04T03:50:39 #kisslinux <konimex> it's odd that the package manager doesn't ignore makedeps on kiss install though
2020-09-04T03:51:02 #kisslinux <konimex> since makedeps are usually prone to cyclic dep
2020-09-04T03:53:16 #kisslinux <dilynm> It probably does
2020-09-04T03:53:25 #kisslinux <dilynm> I bet if I installed flex the problem would also be solved
2020-09-04T03:53:46 #kisslinux <dilynm> And it is
2020-09-04T03:54:04 #kisslinux <konimex> I got looped on m4 and flex with deps referencing each other *with* clarifying them as make dependencies
2020-09-04T03:54:06 #kisslinux <mcf> in case anybody's interested: i wrote a lexer for openbsd m4 to avoid the circular dependency: https://github.com/oasislinux/oasis/blob/master/pkg/openbsd/patch/0020-m4-Use-hand-written-lexer-to-avoid-cycle-in-bootstra.patch
2020-09-04T03:55:04 #kisslinux <dilynm> Michael forney is the hero we don't deserve
2020-09-04T03:55:08 #kisslinux <konimex> nice
2020-09-04T04:01:22 #kisslinux <konimex> alright, crisis averted
2020-09-04T04:04:42 #kisslinux <konimex> mcf: thanks for the patch
2020-09-04T04:05:28 #kisslinux <mcf> np
2020-09-04T06:20:39 #kisslinux <dylanaraps> Fixed m4, thanks everyone.
2020-09-04T06:21:11 #kisslinux <dylanaraps> illiliti: Will fix now.
2020-09-04T06:33:34 #kisslinux <dylanaraps> Also, circular dependencies are considered a bug which is why the package manager does not support them. fwiw
2020-09-04T06:41:37 #kisslinux <dylanaraps> dilynm: How did you end up with flex uninstalled? The binutils package depends on it. Are you by any chance using llvm/clang in place of binutils/gcc?
2020-09-04T06:45:31 #kisslinux <dylanaraps> nvm
2020-09-04T06:45:40 #kisslinux <dylanaraps> I think I understand the full picture now.
2020-09-04T12:27:46 #kisslinux <jayden> hello beanjermin
2020-09-04T12:28:03 #kisslinux <jayden> rip drunk :c
2020-09-04T13:00:29 #kisslinux <dilynm> Sorry for the confusion dylan 😅
2020-09-04T13:05:55 #kisslinux <jaydenramm> lol
2020-09-04T15:30:16 #kisslinux <micr0> mcf theres a name, a blast from the past (hi from exherbo/jedahan)
2020-09-04T15:34:48 #kisslinux <micr0> so theres something thats been bothering me a tiny bit, and wondering what thoughts people have about it
2020-09-04T15:35:05 #kisslinux <micr0> kiss the package manager uses just KISS_PATH and the package name to build and install packages
2020-09-04T15:36:08 #kisslinux <micr0> so if someone has my repository higher than any of the official repositories in their KISS_PATH, I can easily (intentional or accidentally) break or even takeover someones kiss-running machines during a simple kiss update
2020-09-04T15:36:48 #kisslinux <micr0> as an example, i have 2 forks in my repository - git (with perl, for add --interactive --patch), and now openvpn (with mbedtls since libressl is borked with it right now)
2020-09-04T15:37:42 #kisslinux <micr0> generally, is there a non-complex way of having package resolution include repositories, a la @jedahan/git or @jedahan/openvpn?
2020-09-04T15:42:07 #kisslinux <micr0> i am also thinking of lots of other improvements, for example showing what repository a package will be installing/updating from in stdout
2020-09-04T15:42:19 #kisslinux <micr0> warning if it is switching from one repo to another, etc
2020-09-04T15:42:24 #kisslinux <micr0> some could probably done with kiss hooks
2020-09-04T15:46:42 #kisslinux <merakor> Well, you shouldn't use a repository you don't trust
2020-09-04T15:47:02 #kisslinux <merakor> You are giving the same trust to the distribution
2020-09-04T15:48:12 #kisslinux <merakor> Or use the submodule format, ie. make a compilation of packages that you want to use
2020-09-04T15:50:03 #kisslinux <merakor> This is not limited to the kiss environment either
2020-09-04T15:50:15 #kisslinux <micr0> merakor im looking for ways to make it better for everyone. unfortunately cherry-picking packages from different repos leads to other problems, and maybe theres room for improvements that are better than '0% trust' or '100% trust'
2020-09-04T15:53:06 #kisslinux <micr0> yeah definitely not limited to how kiss works, but would like to start by thinking about how kiss can help generally with package management than trying to tackle the whole chain
2020-09-04T15:54:06 #kisslinux <micr0> also i have an inkling that the package manager having a very limited understanding of repositories would provide other UX benefits besides potential security help
2020-09-04T15:55:18 #kisslinux <merakor> I don't know how one would systematically implement a solution for this
2020-09-04T15:55:55 #kisslinux <merakor> How should the package manager define trust? Like gpg trust?
2020-09-04T15:56:56 #kisslinux <merakor> Could kiss remain configurationless with such a system?
2020-09-04T15:57:20 #kisslinux <merakor> These are the questions that come into my mind
2020-09-04T15:59:42 #kisslinux <micr0> so id start with just 'show the user information and let them decide what to do'
2020-09-04T16:00:05 #kisslinux <micr0> for example, when installing or updating a package, show the path of where its coming from
2020-09-04T16:00:40 #kisslinux <micr0> then perhaps when installing a package, copying that path to /var/db/kiss/installed/package/source or something like that
2020-09-04T16:00:57 #kisslinux <micr0> then when updating a package, making a note if the repository path changed
2020-09-04T16:27:41 #kisslinux <merakor> Okay, now that made sense to me
2020-09-04T16:27:58 #kisslinux <merakor> Seems pretty reasonable
2020-09-04T16:28:25 #kisslinux <merakor> I would implement that to my fork
2020-09-04T16:58:13 #kisslinux <eudaldgr> i'm trying to build my dotfiles as a pkg, but when i install it, they have owned by root:root, and symlinks inside /etc dosn't created, but package manager says that they do
2020-09-04T16:58:57 #kisslinux <eudaldgr> tha tarball have the symlinks of /etc
2020-09-04T17:17:52 #kisslinux <micr0> eudaldgr i have done dotfiles as a user by using the -o and -g options for install
2020-09-04T17:18:23 #kisslinux <micr0> eudaldgr example: https://github.com/jedahan/kiss-repo/blob/main/talyn-config/build
2020-09-04T17:20:22 #kisslinux <eudaldgr> well thanks, i made a makefile for the dotfiles and a simple build <make DESTDIR="$1" install> for the pkg
2020-09-04T17:20:36 #kisslinux <eudaldgr> but with your way i can do it
2020-09-04T17:20:42 #kisslinux <eudaldgr> https://github.com/eudaldgr/dotfiles/blob/desktop/Makefile
2020-09-04T17:30:53 #kisslinux <Marijn58> Hi, I am currently installing KISS. I created and switched to another user to make some changes, but I am unable to get back to root. It requires a password, but I never set one for it. How to I switch back to root?
2020-09-04T17:35:17 #kisslinux <micr0> Marijn58 i think the default root password is empty. Might wanna try using `su -` and just hit enter?
2020-09-04T17:36:42 #kisslinux <micr0> eudaldgr neat!
2020-09-04T17:37:27 #kisslinux <Marijn58> micr0 That doesn't seem to work, but I fixed it by exiting the install and chroming back into it :)
2020-09-04T17:37:31 #kisslinux <Marijn58> *chrooting
2020-09-04T17:53:04 #kisslinux <eudaldgr> well don't work
2020-09-04T17:58:41 #kisslinux <randomuser> does anyone know the wifi driver name for a thinkpad x220
2020-09-04T17:58:43 #kisslinux <randomuser> ?
2020-09-04T18:03:48 #kisslinux <randomuser> my card isn't showing up in fresh install of kiss
2020-09-04T18:03:54 #kisslinux <randomuser> causing me to suspect the wifi
2020-09-04T18:19:47 #kisslinux <micr0> iwlwifi maybe?
2020-09-04T18:23:55 #kisslinux <randomuser> yeah, but I'm looking in linux-firmware, and there's a lot of iwlwifi*.ucode
2020-09-04T18:24:03 #kisslinux <randomuser> I just dont know which to choose
2020-09-04T18:24:45 #kisslinux <micr0> oh, i think you can copy the entire dir to /lib/firmware
2020-09-04T18:26:09 #kisslinux <randomuser> just like `cp iwlwifi.ucode /lib/firmware`?
2020-09-04T18:26:17 #kisslinux <randomuser> also I thought it use /usr/lib/firmware
2020-09-04T18:29:21 #kisslinux <micr0> i think like cp -r firmware /lib/
2020-09-04T18:29:32 #kisslinux <micr0> and i think /usr/lib and /lib are symlinked
2020-09-04T18:30:34 #kisslinux <randomuser> oh yeah
2020-09-04T18:30:39 #kisslinux <randomuser> you're right
2020-09-04T18:30:55 #kisslinux <randomuser> so when a kernel is compiled, it looks in /lib/firmware?
2020-09-04T18:31:03 #kisslinux <randomuser> and compiles in the firmware there?
2020-09-04T18:51:20 #kisslinux <micr0> if you specify to compile in yeah
2020-09-04T18:51:29 #kisslinux <randomuser> in the config?
2020-09-04T18:51:30 #kisslinux <micr0> ive had mixed results with compiling firmware in the kernel
2020-09-04T18:51:38 #kisslinux <randomuser> I think it's firmware_lib_something
2020-09-04T18:51:41 #kisslinux <micr0> so for some things i stick with modules so i can modprobe
2020-09-04T18:51:43 #kisslinux <randomuser> I can look in the docs
2020-09-04T18:51:54 #kisslinux <randomuser> well there's no modules here, right
2020-09-04T18:51:57 #kisslinux <micr0> yeah theres a config option, plus another one where you specify exactly what firmware to compile in
2020-09-04T18:52:18 #kisslinux <micr0> randomuser you can configure most stuff as modules. especially things that require firmware.
2020-09-04T18:52:34 #kisslinux <randomuser> but there isnt a initramfs
2020-09-04T18:52:39 #kisslinux <randomuser> at least not in my setup
2020-09-04T18:53:28 #kisslinux <randomuser> i mean, correct me if i'm wrong, but I think you need an initramfs for loadable kernel modules
2020-09-04T18:57:18 #kisslinux <micr0> only if your modules need to be loaded before init for some reason
2020-09-04T18:58:13 #kisslinux <randomuser> so if you're encrypting or something, you need an initramfs, but otherwise you dont?
2020-09-04T18:59:03 #kisslinux <micr0> yep
2020-09-04T18:59:07 #kisslinux <randomuser> cool
2020-09-04T19:41:32 #kisslinux <E5ten> mcpcpc: in kirc raw(), you shouldn't multiply the size for malloc by sizeof(char), because char's size is by definition always 1
2020-09-04T19:48:34 #kisslinux <mcpcpc[m]> <E5ten "mcpcpc: in kirc raw(), you shoul"> E5ten: ahhh yes... learning one mistake at a time :) ty ty!
2020-09-04T19:53:53 #kisslinux <mcf> micr0: hey :) good to see you around
2020-09-04T19:59:03 #kisslinux <mcf> micr0: btw, you can set add.interactive.usebuiltin to get --patch/--interactive without perl
2020-09-04T20:02:11 #kisslinux <mcf> i patched that to be the default on oasis. might make sense for kiss, too, since it avoids perl as well
2020-09-04T20:02:16 #kisslinux <mcf> https://github.com/oasislinux/oasis/blob/master/pkg/git/patch/0003-Enable-add.interactive.usebuiltin-by-default.patch
2020-09-04T22:31:31 #kisslinux <E5ten> mcpcpc: in printw(), wordwidth is int, but it's used to store strlen's return value, so it should be size_t, also in the same function you have printf("%c", I think that could just be a putchar() call
2020-09-04T22:33:04 #kisslinux <E5ten> also there's a call to fgets in input_handler() that I get a warning for because of unchecked return value, and the same with write() in raw()
2020-09-04T22:34:29 #kisslinux <E5ten> and in raw(), you allocate MSG_MAX + 1 to cmd_str, but pass MSG_MAX for the size to vsnprintf, so a byte is being wasted, because the size value for snprintf includes the space for the null-byte, so however much space is allocated for the buffer it's filling should be the size argument
2020-09-04T22:46:13 #kisslinux <randomuser> I have a question
2020-09-04T22:47:09 #kisslinux <randomuser> i've compiled linux with the CONFIG_EXTRA_FIRMWARE_DIR to /lib/firmware and had the iwlwifi stuff in there
2020-09-04T22:47:28 #kisslinux <randomuser> but when I've finished installing, my wifi card isn't recognized
2020-09-04T22:47:35 #kisslinux <randomuser> am i being an idot
2020-09-04T22:49:22 #kisslinux <mcpcpc[m]> E5ten: woo. alright. several things to address. cant give enough thanks :)
2020-09-04T23:00:58 #kisslinux <Evil_Bob> mcf: nice work on cproc. i just tested it on OpenBSD, it only needs a small change in config.h there: "-D", "_MACHINE_ENDIAN_H_",
2020-09-04T23:01:21 #kisslinux <Evil_Bob> to avoid using the inline assembly in the file amd64/endian.h
2020-09-04T23:01:58 #kisslinux <Evil_Bob> (it seems you already noticed it because of the openbsd.diff ;)), the GNUC #ifdef might not be a good solution because OpenBSD uses clang by default now
2020-09-04T23:08:44 #kisslinux <mcf> Evil_Bob: thanks. i seem to recall that there was also an issue with stdarg.h due to a recent-ish change in openbsd, which prevents bootstrap
2020-09-04T23:09:24 #kisslinux <Evil_Bob> ah i havent tried bootstrap yet, but now installed it in my ~/bin and testing it on some of my projects :)
2020-09-04T23:11:32 #kisslinux <mcf> clang also defines __GNUC__, so i don't think that should be an issue
2020-09-04T23:11:52 #kisslinux <Evil_Bob> ah
2020-09-04T23:15:01 #kisslinux <mcf> i really should finish up the preprocessor some time
2020-09-04T23:17:17 #kisslinux <Evil_Bob> i had a program which gave the message: <stdin>:3467:20: error: volatile store is not yet supported
2020-09-04T23:17:44 #kisslinux <Evil_Bob> it was for the line: volatile sig_atomic_t sigstate = 0;
2020-09-04T23:18:36 #kisslinux <Evil_Bob> (some other programs compiled succesfully without any changes, really nice)
2020-09-04T23:18:50 #kisslinux <mcf> ah, yeah. qbe has no way to specify that a load/store should be volatile, so it just errors for now. however, i'm pretty sure that qbe won't optimize away loads/stores for global variables, so -Dvolatile= might be an acceptable workaround for now
2020-09-04T23:19:15 #kisslinux <Evil_Bob> yeah
2020-09-04T23:25:41 #kisslinux <mcf> Evil_Bob: feel free to bring any further discussion about cproc to #oasislinux. happy to chat more there :)