💾 Archived View for gemini.ctrl-c.club › ~phoebos › logs › freenode-kisslinux-2020-11-23.txt captured on 2024-05-26 at 16:20:41.
⬅️ Previous capture (2021-12-17)
-=-=-=-=-=-=-
2020-11-23T00:00:51 #kisslinux <midfavila> gonna test with gnu find to be triple-sure
2020-11-23T00:08:12 #kisslinux <midfavila> yeah that seems to have done it
2020-11-23T00:09:49 #kisslinux <soliwilos> Nice.
2020-11-23T00:25:34 #kisslinux <soliwilos> sbase works here as well with kiss if I replace sbase find with busybox find.
2020-11-23T00:25:57 #kisslinux <midfavila> Awesome, looks like that's definitely the problem then
2020-11-23T00:26:08 #kisslinux <midfavila> not sure how to deal with it though...
2020-11-23T00:26:16 #kisslinux <midfavila> maybe add an echo statement to warn the user in the buildfile?
2020-11-23T00:27:21 #kisslinux <soliwilos> A post-install message, perhaps?
2020-11-23T00:27:29 #kisslinux <midfavila> Yeah, that would work.
2020-11-23T00:29:35 #kisslinux <soliwilos> Maybe the problem won't exist with the C version of kiss.
2020-11-23T00:29:39 #kisslinux <midfavila> that's what I'm thinking
2020-11-23T00:30:28 #kisslinux <midfavila> goddamnit, my client froze again
2020-11-23T00:30:31 #kisslinux <midfavila> shoot me
2020-11-23T00:30:51 #kisslinux <soliwilos> Which client is that?
2020-11-23T00:30:56 #kisslinux <midfavila> pidgin
2020-11-23T00:31:12 #kisslinux <midfavila> I use it because I'm autistic about having as little redundancy in my system as possible
2020-11-23T00:31:32 #kisslinux <midfavila> so bundling IRC, telegram, discord, matrix, etc, all into a single low-resource client is nice
2020-11-23T00:31:40 #kisslinux <midfavila> also, it's gtk2, so it obeys theming nicely
2020-11-23T00:32:30 #kisslinux <soliwilos> I remember giving up multi-clients like that since they never seemed to be as good as the specialized ones.
2020-11-23T00:32:46 #kisslinux <midfavila> well, the only other desktop irc client I've used is hexchat
2020-11-23T00:32:50 #kisslinux <midfavila> and hexchat is just annoying
2020-11-23T00:32:56 #kisslinux <midfavila> like it actually pisses me off
2020-11-23T00:33:06 #kisslinux <midfavila> because it warps your cursor into the window every time you get a message
2020-11-23T00:33:39 #kisslinux <soliwilos> Before I gave kiss linux a shot, I was using irssi. Been using kirc a good while now.
2020-11-23T00:33:44 #kisslinux <midfavila> but I guess I can see where you're coming from. pidgin is definitely only as good as the plugins you've loaded
2020-11-23T00:34:16 #kisslinux <midfavila> honestly, I know it kind of defeats the point, but I'm not huge on console applications
2020-11-23T00:34:30 #kisslinux <midfavila> (the point of kiss and minimalist software and whatnot, that is)
2020-11-23T00:34:44 #kisslinux <soliwilos> That's fair enough.
2020-11-23T00:34:47 #kisslinux <midfavila> Yeah.
2020-11-23T00:34:53 #kisslinux <midfavila> It's great for getting actual work done, don't get me wrong
2020-11-23T00:35:02 #kisslinux <midfavila> any time I need to do serious work I pull up a terminal or three
2020-11-23T00:35:20 #kisslinux <midfavila> but in terms of just regular usage I prefer a GUI
2020-11-23T00:35:30 #kisslinux <midfavila> except for file management. way too slow
2020-11-23T00:35:42 #kisslinux <soliwilos> I was tempted to give dino a try over profanity for xmpp, but I've stuck with profanity.
2020-11-23T00:35:52 #kisslinux <midfavila> I've never tried xmpp
2020-11-23T00:36:05 #kisslinux <midfavila> isn't it pretty similar to irc?
2020-11-23T00:37:17 #kisslinux <soliwilos> I guess it's somewhat between irc and the more "normal" chatting services, in that you have a roster/friends list.
2020-11-23T00:38:32 #kisslinux <midfavila> Ah, alright
2020-11-23T00:38:39 #kisslinux <soliwilos> You can also get E2E, though I've not tried that in multiuser chats on xmpp. I've mostly used it for private chats with close friends.
2020-11-23T00:40:40 #kisslinux * midfavila_ nods
2020-11-23T00:41:20 #kisslinux <soliwilos> Filesharing and inline images are easily usable with xmpp clients as well.
2020-11-23T00:41:29 #kisslinux <midfavila> also, wow, just surpassed my record on how little memory I use at idle with a full DE
2020-11-23T00:41:35 #kisslinux <midfavila> 124 megs
2020-11-23T00:41:55 #kisslinux <midfavila> filesharing could be useful I suppose
2020-11-23T00:42:01 #kisslinux <midfavila> inline images... eh... I dunno
2020-11-23T00:43:49 #kisslinux <soliwilos> Niice, that's pretty slim memory usage.
2020-11-23T00:44:08 #kisslinux <midfavila> about thirty of that was the task manager :p
2020-11-23T00:44:26 #kisslinux <midfavila> i'm not very interesting so minimizing my setup's memory footprint is kind of a hobby
2020-11-23T00:44:27 #kisslinux <soliwilos> haha
2020-11-23T00:44:42 #kisslinux <midfavila> let me see how much X is using on its own
2020-11-23T00:45:02 #kisslinux <midfavila> yeah X is using 77mb
2020-11-23T00:45:09 #kisslinux <midfavila> rip
2020-11-23T00:45:17 #kisslinux <midfavila> still not sure how to minimize X's footprint
2020-11-23T00:45:30 #kisslinux <soliwilos> Swap to wayland. :p
2020-11-23T00:45:41 #kisslinux <midfavila> that would require I replace every single program I use
2020-11-23T00:45:42 #kisslinux <midfavila> big pass
2020-11-23T00:45:43 #kisslinux <midfavila> :p
2020-11-23T00:45:53 #kisslinux <soliwilos> hehe, understandable.
2020-11-23T00:45:55 #kisslinux <midfavila> besides, how can I use Xaw on wayland?
2020-11-23T00:46:00 #kisslinux <midfavila> checkmate, shill
2020-11-23T00:46:31 #kisslinux <midfavila> even though I have like
2020-11-23T00:46:31 #kisslinux <midfavila> literally one program that uses it
2020-11-23T00:56:11 #kisslinux <soliwilos> Looks like sbase and ubase does not completely fill in for all of busybox's stuff.
2020-11-23T00:56:24 #kisslinux <midfavila> Yeah, they don't
2020-11-23T00:56:50 #kisslinux <midfavila> I usually complement them with gnused, gnugrep, one true awk, mksh, dash, and my own packaging of inetutils and net-tools
2020-11-23T00:56:56 #kisslinux <midfavila> along with sysmgr and hummingbird
2020-11-23T00:57:17 #kisslinux <midfavila> plus some other stuff like shadow and whatnot
2020-11-23T00:58:44 #kisslinux <soliwilos> Mhm.
2020-11-23T00:58:51 #kisslinux <soliwilos> I like mksh as well.
2020-11-23T00:58:56 #kisslinux <midfavila> oh, and of course findutils, pciutils, libarchive, and __usually__ usbutils, that last one seems broken atm
2020-11-23T00:59:10 #kisslinux <midfavila> normally I'd use yash, since it's even slimmer than dash while remaining useful interactively
2020-11-23T00:59:17 #kisslinux <midfavila> but it's kinda wonky as of late on kiss
2020-11-23T01:06:39 #kisslinux <soliwilos> As for init and process manager, I like s6.
2020-11-23T01:07:39 #kisslinux <midfavila> I've never bothered with s6
2020-11-23T01:07:42 #kisslinux <midfavila> seems neat tho
2020-11-23T01:20:04 #kisslinux <soliwilos> s6 is quite nice, but does require a fair bit of reading to figure out how it all fits together.
2020-11-23T01:21:25 #kisslinux <E5ten> dylanaraps: I might've already sent something like this (or maybe exactly this) but here's a patch to get rid of nftw usage in k that I think should work, feel free to give it a try http://ix.io/2F9C
2020-11-23T01:21:37 #kisslinux <E5ten> (iirc that's the only XSI thing k uses?)
2020-11-23T01:26:26 #kisslinux <mcf> midfavila: do you have a way to reproduce the find issue so i can try to fix it?
2020-11-23T01:26:41 #kisslinux <midfavila> build sbase, install it, build ubase
2020-11-23T01:26:47 #kisslinux <midfavila> that's an easy way to demo it
2020-11-23T01:27:03 #kisslinux <mcf> i mean, some minimal test case that doesn't involve me installing kiss
2020-11-23T01:27:18 #kisslinux <midfavila> I dunno then.
2020-11-23T01:27:25 #kisslinux <midfavila> I didn't have any other issues with it.
2020-11-23T01:27:31 #kisslinux * midfavila shrugs
2020-11-23T01:27:42 #kisslinux <mcf> can you do a bit of debugging?
2020-11-23T01:27:58 #kisslinux <midfavila> kinda busy, but in what way?
2020-11-23T01:28:22 #kisslinux <midfavila> pretty sure it's something to do with kiss itself in this case
2020-11-23T01:29:01 #kisslinux <mcf> determine the find command that kiss is using, the output that you get with GNU find, and the output you get with sbase find
2020-11-23T01:31:29 #kisslinux <midfavila> well, looking through kiss, it's using some arguments that sbase find doesn't support
2020-11-23T01:31:37 #kisslinux <midfavila> in this case the most obvious is -type
2020-11-23T01:31:38 #kisslinux <mcf> are they POSIX?
2020-11-23T01:31:58 #kisslinux <midfavila> let me pull up a reference
2020-11-23T01:31:59 #kisslinux <mcf> sbase find supports -type
2020-11-23T01:32:28 #kisslinux <midfavila> oh, so it does. didn't see it listed in the --help
2020-11-23T01:32:30 #kisslinux <midfavila> my bad then.
2020-11-23T01:32:31 #kisslinux <midfavila> hrm
2020-11-23T01:35:34 #kisslinux <midfavila> well, after skipping through kiss and looking at every instance of find, it doesn't look like there's any improper arguments, referencing sbase's find manpage
2020-11-23T01:35:39 #kisslinux <E5ten> Totally guessing cuz I'm not even at my computer, could it be the find call at line 538?
2020-11-23T01:36:08 #kisslinux <E5ten> Just cuz it has a -o -print after the -exec which seems like it could maybe be an edge case for some reason
2020-11-23T01:36:08 #kisslinux <midfavila> well, that's the only find in pkg_manifest(), so it'd make sense
2020-11-23T01:36:20 #kisslinux <midfavila> and if it's an edge case that would make even more sense
2020-11-23T01:36:24 #kisslinux <mcf> so the error you get is that some things are missing in the manifest?
2020-11-23T01:36:29 #kisslinux <midfavila> Yeah.
2020-11-23T01:36:48 #kisslinux <mcf> can you show the correct manifest and the entries that are missing?
2020-11-23T01:37:02 #kisslinux <midfavila> Where do I pull up the proper manifest?
2020-11-23T01:37:36 #kisslinux <E5ten> Maybe just try running that command directly in some test dir with sbase and busybox find to see if they differ?
2020-11-23T01:38:41 #kisslinux <midfavila> yeah
2020-11-23T01:38:44 #kisslinux <midfavila> output's different
2020-11-23T01:38:59 #kisslinux <midfavila> busybox handles it just find, sbase find has trouble
2020-11-23T01:39:14 #kisslinux <midfavila> this is just being run in the root dir of a kiss-chroot
2020-11-23T01:39:22 #kisslinux <midfavila> find: readdir ./proc/222/task/222/net: Invalid argument
2020-11-23T01:39:22 #kisslinux <midfavila> find: readdir ./proc/222/net: Invalid argument
2020-11-23T01:39:22 #kisslinux <midfavila> find: readdir ./proc/223/task/223/net: Invalid argument
2020-11-23T01:39:22 #kisslinux <midfavila> find: readdir ./proc/223/net: Invalid argument
2020-11-23T01:39:22 #kisslinux <midfavila> find: readdir ./proc/276/task/276/net: Invalid argument
2020-11-23T01:39:23 #kisslinux <midfavila> find: readdir ./proc/276/net: Invalid argument
2020-11-23T01:39:23 #kisslinux <midfavila> find: readdir ./proc/277/task/277/net: Invalid argument
2020-11-23T01:39:24 #kisslinux <midfavila> find: readdir ./proc/277/net: Invalid argument
2020-11-23T01:39:48 #kisslinux <mcf> you probably want -xdev
2020-11-23T01:40:16 #kisslinux <midfavila> Maybe. I'm just copying the command straight from kiss
2020-11-23T01:40:40 #kisslinux <mcf> well, kiss is probably running it on the package installation directory which has no submounts
2020-11-23T01:41:25 #kisslinux <midfavila> Running it with xdev produces no output, so that's something at least
2020-11-23T01:44:28 #kisslinux <E5ten> I meant a test dir like just some simple dir where you have a subdir and a file or 2, to see if the -o -print after the -exec is the problem
2020-11-23T01:44:35 #kisslinux <mcf> i can't really debug the problem off of "sbase find has trouble". i need some tarball of a directory, a find command to run on the contents, and the expected output for that find command
2020-11-23T01:44:45 #kisslinux <mcf> perhaps the ubase binary package would work?
2020-11-23T01:45:46 #kisslinux <midfavila> I don't know. I'm not really experienced with shell scripting and stuff. Sorry.
2020-11-23T01:49:49 #kisslinux <mcf> can you upload your ubase binary package somewhere (maybe 0x0.st)?
2020-11-23T01:50:21 #kisslinux <midfavila> Sure, I guess... let me rebuild it with sbase find
2020-11-23T01:51:39 #kisslinux * midfavila sighs
2020-11-23T01:51:46 #kisslinux <midfavila> and now, randomly, it works.
2020-11-23T01:51:50 #kisslinux <midfavila> what the fuck.
2020-11-23T01:52:43 #kisslinux <mcf> hmm, well if it happens again, let me know
2020-11-23T01:52:50 #kisslinux <midfavila> well, at least it isn't a problem on just my machine... there was someone else here who mentioned they had issues with it
2020-11-23T01:52:56 #kisslinux <midfavila> maybe it's an interaction between find and something else
2020-11-23T01:55:33 #kisslinux <soliwilos> Running the find bit from line 538 on a sourcedir, there's one thing I'm noticing at a quick glance. Busybox find results ends with a "./" but sbase find does not.
2020-11-23T01:57:05 #kisslinux <soliwilos> In it's place looks like an empty line.
2020-11-23T02:00:55 #kisslinux <mcf> that's interesting, thanks. i think the ./ might be getting appended to some previous line
2020-11-23T02:01:05 #kisslinux <mcf> perhaps related to how it spawns the printf commands
2020-11-23T02:03:19 #kisslinux <soliwilos> busybox find: http://ix.io/2F9O
2020-11-23T02:03:29 #kisslinux <soliwilos> sbase find: http://ix.io/2F9P
2020-11-23T02:04:00 #kisslinux <soliwilos> The sbase one there has a empty newline in place of the "./"
2020-11-23T02:08:07 #kisslinux <mcf> i think this might be an interaction between the external printf command, and buffering of stdout
2020-11-23T02:15:30 #kisslinux <mcf> soliwilos: can you test if http://ix.io/2F9V fixes the issue?
2020-11-23T02:16:30 #kisslinux <soliwilos> mcf: Sure.
2020-11-23T02:20:54 #kisslinux <soliwilos> mcf: Problem is still there.
2020-11-23T02:22:32 #kisslinux <soliwilos> This is the error: http://ix.io/2F9W
2020-11-23T02:24:09 #kisslinux <mcf> so the /usr/share/man/man1/pwdx.1 is from the -print, and the ./ is from the -exec printf. somehow they are getting mixed together
2020-11-23T02:26:26 #kisslinux <E5ten> maybe something to do with the grouping of the directory results (with the + after the -exec)? idk how that'd be working but I wonder if the same issue occurs with "{} + ;"
2020-11-23T02:26:37 #kisslinux <E5ten> oh wait no I wrote that wrong lol
2020-11-23T02:26:40 #kisslinux <E5ten> "{} ;"
2020-11-23T02:27:19 #kisslinux <E5ten> soliwilos: can you try running the same thing but with an escaped ; instead of +?
2020-11-23T02:30:10 #kisslinux <soliwilos> busybox seems the same as before, but sbase find get's new stuff.
2020-11-23T02:30:24 #kisslinux <soliwilos> a bit weird..
2020-11-23T02:30:33 #kisslinux <E5ten> can you send the output?
2020-11-23T02:30:46 #kisslinux <soliwilos> "./" and "." and two empty newlines..
2020-11-23T02:32:26 #kisslinux <soliwilos> Like this: http://ix.io/2Fa0
2020-11-23T02:32:46 #kisslinux <mcf> as far as i can tell, the printf command is forked and waited for, so it should be a single unit. but if the puts from -print outputs a partial line, then the printf executes, that would explain the weird output
2020-11-23T02:33:12 #kisslinux <mcf> but if setting stdout to line buffered does not fix the issue, then i'm not really sure
2020-11-23T02:39:40 #kisslinux <E5ten> I'm not on kiss, but just running the find command on 538 using busybox and sbase I get no difference btw
2020-11-23T02:41:19 #kisslinux <mcf> make sure you run it in a pipeline so that stdout isn't buffered line-buffered by default
2020-11-23T02:41:22 #kisslinux <mcf> i.e. find ... | cat
2020-11-23T02:42:03 #kisslinux <mcf> (since it is used in a pipeline in kiss)
2020-11-23T02:42:56 #kisslinux <E5ten> ok ran it through sort, yeah I see the difference
2020-11-23T02:44:39 #kisslinux <E5ten> specifically I ran it in my sbase tree, and the ./ at the top got inserted at the end of ./unlink, replacing the final "k", so it became ./unlin./, and then there's an extra line at the end that's just "k"
2020-11-23T02:45:56 #kisslinux <E5ten> seeing no difference when I add the setvbuf call
2020-11-23T02:55:34 #kisslinux <E5ten> soliwilos: can you try running the find command itself (piping it like mcf suggested) with the added setvbuf line? like instead of running it through kiss can you try just running the command itself and comparing output again, but with that change, and back to using + instead of ;? cuz I also don't see how the issue could still be happening with line buffering enabled
2020-11-23T02:59:57 #kisslinux <soliwilos> The outputs I've posted have been without kiss, apart from the error message.
2020-11-23T03:00:38 #kisslinux <E5ten> did you send an output for running with the setvbuf change? I only see the error for that one and then after that output for using ; instead of +
2020-11-23T03:01:44 #kisslinux <soliwilos> Well the output is identical to before the setvbuf patch.
2020-11-23T03:02:02 #kisslinux <E5ten> ah
2020-11-23T03:06:45 #kisslinux <soliwilos> I need some sleep now, good night!
2020-11-23T03:07:45 #kisslinux <E5ten> see ya
2020-11-23T03:40:58 #kisslinux <mcf> E5ten: so you don't see a difference with setvbuf either?
2020-11-23T03:41:19 #kisslinux <mcf> i think a better solution might be to just fflush(stdout) before spawning a command
2020-11-23T03:51:49 #kisslinux <mcf> it looks like this is unspecified behavior in POSIX: "it is otherwise unspecified whether the invocation occurs before, during, or after the evaluations of other primaries"
2020-11-23T03:52:30 #kisslinux <mcf> it should be easy to fix though (assuming that this is the problem)
2020-11-23T04:26:35 #kisslinux <sge> New install. There's no UTF8 support in the terminal. Any tips?
2020-11-23T04:26:59 #kisslinux <sge> (FreeBSD transplant here, sorry for possibly dumb questons)
2020-11-23T05:15:30 #kisslinux <E5ten> mcf: I do see a difference with the setvbuf
2020-11-23T05:15:31 #kisslinux <E5ten> It worked
2020-11-23T05:16:07 #kisslinux <dylanaraps> o/
2020-11-23T05:16:14 #kisslinux <E5ten> mcf: I think if it's unspecified dylanaraps would wanna fix it?
2020-11-23T05:16:54 #kisslinux <E5ten> Perhaps 2 find calls, one with -type d and -exec and the other with ! -type d, in braces, and then pipe the braces into the sort?
2020-11-23T05:16:59 #kisslinux <dylanaraps> E5ten: Thanks for the patch. Will apply it soon.
2020-11-23T05:17:20 #kisslinux <dylanaraps> Read the logs about find. If it's UB, I'll 100% fix it
2020-11-23T05:17:21 #kisslinux <E5ten> dylanaraps: make sure to test it thoroughly first though, I don't have kiss installed so I can't
2020-11-23T05:17:54 #kisslinux <dylanaraps> E5ten: Will do. The only place it's used now is in cache cleanup.
2020-11-23T05:18:13 #kisslinux <dylanaraps> There will be similar for manifest generation, package installation, etc
2020-11-23T05:18:20 #kisslinux <E5ten> Makes sense
2020-11-23T05:19:08 #kisslinux <E5ten> But yeah see my suggestion above about how to fix the find ub
2020-11-23T05:19:27 #kisslinux <E5ten> Not great but I'm not sure how you could do it with one find call without that ub
2020-11-23T05:20:50 #kisslinux <dylanaraps> This should do it: https://termbin.com/x57r
2020-11-23T05:22:03 #kisslinux <dylanaraps> Fixed
2020-11-23T05:22:32 #kisslinux <dylanaraps> E5ten: Is there any way to remove() or unlink() a FILE or fd?
2020-11-23T05:23:01 #kisslinux <dylanaraps> unlinkat is the closest thing (if I hold an fd to some starting portion of the path)
2020-11-23T05:28:11 #kisslinux <micr0> sooooo dylanaraps i ran into my first package conflict that wasnt solved automatically (ie. via kiss alternatives)
2020-11-23T05:28:24 #kisslinux <dylanaraps> micr0: Details?
2020-11-23T05:28:28 #kisslinux <micr0> emacs won't build if you have libexecinfo installed
2020-11-23T05:28:44 #kisslinux <micr0> the build scripts use libexecinfo (no way to specify to use or not)
2020-11-23T05:29:07 #kisslinux <micr0> and musl linux shouldnt even have libexecinfo but someone made some kinda-working version that is needed for stuff like pulseaudio
2020-11-23T05:29:14 #kisslinux <dylanaraps> Yeah. I don't support libexecinfo. It's been an endless cause of support issues.
2020-11-23T05:29:45 #kisslinux <dylanaraps> We don't support any of the tack-on libc extensions really
2020-11-23T05:29:46 #kisslinux <micr0> yeah i dont know what libexecinfo does, all i know is that i had to uninstall it to build emacs (surprisingly fast build)
2020-11-23T05:30:14 #kisslinux <dylanaraps> How is libexecinfo related to a missed conflict though
2020-11-23T05:30:16 #kisslinux <dylanaraps> ?*
2020-11-23T05:30:36 #kisslinux <micr0> missed conflict is a red herring
2020-11-23T05:30:52 #kisslinux <dylanaraps> What was the output from kiss when it happened?
2020-11-23T05:30:55 #kisslinux <micr0> i was trying to compliment that ive been on kiss long enough and that there is a package-level conflict that is not a file level conflict
2020-11-23T05:31:26 #kisslinux <dylanaraps> I just woke up so forgive my slowness
2020-11-23T05:31:28 #kisslinux <dylanaraps> ;)
2020-11-23T05:31:37 #kisslinux <micr0> basically, in other package managers, i would add !libexecinfo the depends
2020-11-23T05:31:45 #kisslinux <micr0> and it would bail early stating there is a conflict
2020-11-23T05:32:10 #kisslinux <micr0> but I am not sure if asking for that is a misfeature, or if spending some time to think about it would be better
2020-11-23T05:32:58 #kisslinux <dylanaraps> Well
2020-11-23T05:32:59 #kisslinux <micr0> in my emacs/build the first line I run is something like "kiss list libexecinfo && { printf "please uninstall libexecinfo to build emacs >&2" && false } || true
2020-11-23T05:33:10 #kisslinux <dylanaraps> We don't support libexecinfo
2020-11-23T05:33:23 #kisslinux <dylanaraps> So it's not something I want to add checks for
2020-11-23T05:33:33 #kisslinux <micr0> i wouldnt want checks for libexecinfo specifically
2020-11-23T05:33:59 #kisslinux <dylanaraps> Yeah
2020-11-23T05:34:01 #kisslinux <micr0> i would maybe want support for !somepackage in the /depends file
2020-11-23T05:34:13 #kisslinux <dylanaraps> It's checks for: pkg a cannot be built if pkg b is present
2020-11-23T05:34:22 #kisslinux <dylanaraps> A way to tell the package manager this
2020-11-23T05:34:39 #kisslinux <micr0> yeah its just for packages to give users a bit better of a message
2020-11-23T05:35:06 #kisslinux <dylanaraps> I'd need to see cases of this with supported software
2020-11-23T05:35:15 #kisslinux <micr0> but I worry that a) the ergonomics arent simple b) how many packages would actually use it c) how complex the implementation would be
2020-11-23T05:35:36 #kisslinux <dylanaraps> b) is what I'm worried about
2020-11-23T05:35:44 #kisslinux <dylanaraps> Doesn't seem like much would use it at all
2020-11-23T05:35:57 #kisslinux <E5ten> I don't think you can remove a FILE */fd
2020-11-23T05:36:17 #kisslinux <dylanaraps> E5ten: Yeah, thought so. I thought I'd ask anyway.
2020-11-23T05:36:42 #kisslinux <dylanaraps> This forces me to do path building with strings then
2020-11-23T05:37:00 #kisslinux <dylanaraps> It's a shame that I can't just pass fds/FILE * around
2020-11-23T05:37:14 #kisslinux <E5ten> Path building for what out of curiosity?
2020-11-23T05:37:42 #kisslinux <dylanaraps> it's just joining paths together really.
2020-11-23T05:37:53 #kisslinux <dylanaraps> cache directory + pkg + dest_dir + source basename
2020-11-23T05:38:01 #kisslinux <dylanaraps> repository directory + pkg + file
2020-11-23T05:38:02 #kisslinux <dylanaraps> etc
2020-11-23T05:38:05 #kisslinux <E5ten> Ah
2020-11-23T05:38:09 #kisslinux <dylanaraps> Lots of this
2020-11-23T05:38:35 #kisslinux <dylanaraps> I can do the joining via the *at() functions
2020-11-23T05:38:48 #kisslinux <dylanaraps> But then I realize that I need the full path...
2020-11-23T05:39:13 #kisslinux <dylanaraps> Another reason for keeping full paths is error messages
2020-11-23T05:39:21 #kisslinux <E5ten> Makes sense
2020-11-23T05:39:46 #kisslinux <dylanaraps> It's fine though
2020-11-23T05:40:04 #kisslinux <E5ten> I might try to make a more general version of that nftw-lite rm_rf impl to allow things other than just removing
2020-11-23T05:40:08 #kisslinux <dylanaraps> My buf impl lets me alloc once at program start and reuse until finish
2020-11-23T05:40:13 #kisslinux <E5ten> Like having the caller pass a func
2020-11-23T05:40:16 #kisslinux <dylanaraps> Yeah
2020-11-23T05:40:24 #kisslinux <E5ten> Just so I can copy it in anything I might find it useful for
2020-11-23T05:40:42 #kisslinux <E5ten> And I guess probably put it up on github for others to do the same if it seems decent
2020-11-23T05:41:35 #kisslinux <dylanaraps> Yeah
2020-11-23T05:41:41 #kisslinux <dylanaraps> That'd be great
2020-11-23T05:41:45 #kisslinux <E5ten> I'd probably just want it to act like nftw would when given whatever seems like an "optimal or most used" set of flags kinda thing
2020-11-23T05:42:16 #kisslinux <E5ten> Like nftw seems most useful for the depth-first option, so it does that cuz it needs to for removing, the staying on the same mount thing seems useful so it does that
2020-11-23T05:42:33 #kisslinux <dylanaraps> Yeah
2020-11-23T05:42:35 #kisslinux <dylanaraps> Makes sense
2020-11-23T05:42:42 #kisslinux <E5ten> And I'd check if there's any other relatively simple and often-used nftw flags I guess
2020-11-23T05:44:22 #kisslinux <dylanaraps> What's cool about k is that you can use it with/without dependencies. For sha256 you have the choice between internal, openssl or bearssl. For tar you have the option between libarchive or the tar command. etc etc
2020-11-23T05:45:24 #kisslinux <dylanaraps> I'm working on my own tar impl locally too
2020-11-23T05:45:46 #kisslinux <dylanaraps> libarchive is not ideal
2020-11-23T07:41:35 #kisslinux <E5ten> Yeah that is cool
2020-11-23T09:38:08 #kisslinux <testuser[m]> dylanaraps: I'm having this weird issue with KISS where if i try to build dhcpcd, it builds fine, however the package manager just exits after the `Generating etcsums` message and no errors are generated
2020-11-23T09:38:22 #kisslinux * testuser[m] posted a file: kiss_5.2.1.log (26KiB) < https://matrix.org/_matrix/media/r0/download/matrix.org/IqkHaueVDHEOUCKfuQbwetOq/kiss_5.2.1.log >
2020-11-23T09:38:45 #kisslinux * testuser[m] posted a file: kiss_5.2.0.log (26KiB) < https://matrix.org/_matrix/media/r0/download/matrix.org/LZaIqoEwaVnnNKntNuZnuOiq/kiss_5.2.0.log >
2020-11-23T09:39:01 #kisslinux <testuser[m]> Works fine with v5.2.0 but not with 5.2.1
2020-11-23T09:43:42 #kisslinux * testuser[m] sent a long message: < https://matrix.org/_matrix/media/r0/download/matrix.org/DGKUKSFQbTsLePVgdchxGSyN/message.txt >
2020-11-23T13:34:03 #kisslinux <testuser[m]> works fine if i revert commit 8973e18abd095929c726896103b503a2c31072f8
2020-11-23T15:22:55 #kisslinux <dilynm> himmalerin: testuser[m]: what's you're guys' /bin/sh? I had similar problems when sh was... Loksh iirc? Some ksh derivative
2020-11-23T15:23:05 #kisslinux <dilynm> Never figured it out and nobody else could reproduce it
2020-11-23T15:26:06 #kisslinux <testuser[m]> `ash`
2020-11-23T15:27:07 #kisslinux <dilynm> Hm
2020-11-23T15:28:44 #kisslinux <dilynm> Can't help you there :o
2020-11-23T16:07:16 #kisslinux <E5ten> mcf: I'm looking at sdhcp, couldn't it use the POSIX timer_* stuff instead of timerfd for the same purpose? And then use timer_getoverrun instead of poll? (except you'd still use poll for the socket that isn't a timer obviously)
2020-11-23T16:34:07 #kisslinux <micr0> when kiss build mentions that readelf found a missing dependency on 'gcc', does that mean its a runtime dependency? i.e. that kiss looks at all binaries with readelf to see if they are linked against libgcc?
2020-11-23T16:34:56 #kisslinux <micr0> nevermind I just answered my own question with `nvim $(which kiss)` and /readelf . So convinient.
2020-11-23T16:37:38 #kisslinux <soliwilos> Thanks for looking into the kiss/find issue.
2020-11-23T16:38:20 #kisslinux <soliwilos> dylanaraps: Your patch for find in pkg_manifest() works. :)
2020-11-23T16:45:09 #kisslinux <midfavila> Oh, was a fix found for that?
2020-11-23T16:46:01 #kisslinux <soliwilos> Yes.
2020-11-23T16:46:06 #kisslinux <midfavila> Based.
2020-11-23T16:48:02 #kisslinux <soliwilos> If you don't find it in the logs I can re-post the link he gave with the patch.
2020-11-23T16:48:22 #kisslinux <midfavila> I'll look through some time later today
2020-11-23T17:29:10 #kisslinux <testuser[m]> Can someone help me figure out why this firefox build is failing ?
2020-11-23T17:31:41 #kisslinux * testuser[m] posted a file: firefox-2020-11-23-17:21-1371 (6041KiB) < https://matrix.org/_matrix/media/r0/download/matrix.org/PbkKMTjyknqZqMctQmeuqVdT >
2020-11-23T17:31:42 #kisslinux <testuser[m]> "Error" is around line 10046, rust 1.48.0, something related to "packed_simd" failing to compile
2020-11-23T17:32:18 #kisslinux <testuser[m]> This is on a fresh GKISS chroot, but i compiled ff with the same rust version on my gkiss host just yesterday
2020-11-23T17:34:34 #kisslinux <testuser[m]> I had some problem with python not being able to use some SHM stuff but that was fixed after i rebuilt it with proper perms on /dev/shm
2020-11-23T18:09:31 #kisslinux <micr0> so i packaged tmate today, which lets people share terminal sessions (read-only or read-write)
2020-11-23T18:09:49 #kisslinux <micr0> the nice thing is that I can share a terminal without you installing anything special, just ssh
2020-11-23T18:10:49 #kisslinux <micr0> so if you want to connect to my machine and view something: ssh ro-bAexdC6Qr5nhnhSuE7F8fx6jX⊙nti
2020-11-23T18:15:47 #kisslinux <testuser[m]> Sounds cool
2020-11-23T18:47:41 #kisslinux <midfavila> that's pretty neat.
2020-11-23T19:34:38 #kisslinux <E5ten> dylanaraps: in k, for fopenatat, you might wanna use O_DIRECTORY (along with the O_RDONLY that's already there) for the first openat, to avoid unnecessarily opening and closing fd when it won't work as the dirfd for the second openat anyway
2020-11-23T19:44:24 #kisslinux <E5ten> same thing for all the openat's that're used to get a dirfd, like in cache_mkdirat, cache_fopen
2020-11-23T19:44:49 #kisslinux <E5ten> cache_init_all with the open and openat
2020-11-23T19:47:35 #kisslinux <E5ten> dylanaraps: also, imo you should probably use AT_EACCESS for all the faccessat calls, and probably change the access calls to faccessat, also using AT_EACCESS
2020-11-23T19:48:02 #kisslinux <E5ten> cuz that'll actually reflect what the user can access
2020-11-23T19:53:04 #kisslinux <mcf> E5ten: yeah, there are some threads on the mailing list about that (Nov 2018 and Feb 2019). i think one issue with POSIX timers is they need to be recreated after the fork
2020-11-23T19:53:56 #kisslinux <mcf> i always use sdhcp with -f and a service manager, though
2020-11-23T19:56:57 #kisslinux <E5ten> would it be possible to move the daemonization to before anything else happens, assuming -f isn't used, and then just create the timers once, after the fork?
2020-11-23T19:57:24 #kisslinux <mcf> iirc, the main problem i have with sdhcp is that if you already have a configured interface, it will send your dhcp requests on that interface rather than broadcast. you need to construct and send raw packets you construct yourself to fix that
2020-11-23T19:57:52 #kisslinux <mcf> it doesn't daemonize until it successfully configures an address, and you need the timers before that, too
2020-11-23T19:58:03 #kisslinux <E5ten> ah
2020-11-23T19:58:33 #kisslinux <E5ten> I guess you could move the timer creation to a function, and then call that where it's called normally, and also after the fork if -f isn't used?
2020-11-23T19:59:27 #kisslinux <mcf> yeah. Sean MacLennan has some changes posted to the ML, but sdhcp is one of those projects with no real maintainer
2020-11-23T20:04:21 #kisslinux <E5ten> so just looking at it, the actual setting of the timers (not creation) is done again right after the fork anyway, so it's just timer_create calls that'd need to be factored out and done after the fork again right?
2020-11-23T20:05:06 #kisslinux <mcf> perhaps, it's been a few years since i looked at the code
2020-11-23T20:05:14 #kisslinux <E5ten> ah
2020-11-23T20:34:13 #kisslinux <nerditup> do people typically include their wireless kernel modules as modules (=m) or baked into the kernel (=y)? I want to remove some modprobing at boot and heard that typically it is better to keep some device drivers as modules
2020-11-23T20:35:31 #kisslinux <nerditup> basically if the driver needs to access /lib/firmware then it's better to build it as a module - just want to see what others do?
2020-11-23T20:47:31 #kisslinux <mcf> i always build all drivers and firmware i need into the kernel. seems simpler that way
2020-11-23T20:47:49 #kisslinux <sh4rm4^bnc> =y
2020-11-23T20:48:48 #kisslinux <sh4rm4^bnc> (and fw is in /lib, and it works anyway)
2020-11-23T21:31:06 #kisslinux <nerditup> nice okay, trying now to boot with =y for my wireless device modules
2020-11-23T21:31:56 #kisslinux <micr0> nerditup i hope it works
2020-11-23T21:32:39 #kisslinux <nerditup> key is to have the drivers available in /lib I think
2020-11-23T21:32:46 #kisslinux <nerditup> during compile time
2020-11-23T21:47:27 #kisslinux <nerditup> the device isn't showing up in /dev lol
2020-11-23T21:47:36 #kisslinux <nerditup> back to the drawing board for me
2020-11-23T22:04:10 #kisslinux <claudia02> nerditup: have a look a kiss-help wiki/kernel/firmware
2020-11-23T22:12:01 #kisslinux <micr0> does it confuse anyone else the format of kiss help?
2020-11-23T22:12:20 #kisslinux <micr0> like when I look at kiss help all the links show @/ but if you pass that into kiss help it cant find things
2020-11-23T22:12:21 #kisslinux <claudia02> When building webkit2gtk with opengl enabled(tested for X), the build breaks with the recent added wayland fix. lol.
2020-11-23T22:13:24 #kisslinux <micr0> `kiss help @/wiki` and `kiss help /wiki` both fail
2020-11-23T22:13:38 #kisslinux <claudia02> Y that would be handy.
2020-11-23T22:14:58 #kisslinux <micr0> yeah maybe if i was more comfy with posix sh i'd `nvim $(which kiss-help)` and strip [@/]+ off $1 ...
2020-11-23T22:16:05 #kisslinux <claudia02> just navigate to /usr/share/doc/kiss ;)
2020-11-23T22:16:19 #kisslinux <claudia02> If you dont want to touch your browser
2020-11-23T22:18:12 #kisslinux <dilyn> note to self: proof-read and thoroughly test your wiki docs before they get merged
2020-11-23T22:18:18 #kisslinux <dilyn> mfw zram devices all failed to init :'(
2020-11-23T22:18:46 #kisslinux <claudia02> You know people how write stuff have other poeple handy to read their stuff?
2020-11-23T22:18:51 #kisslinux <claudia02> *who
2020-11-23T22:19:23 #kisslinux <claudia02> Trees and forest problem..(:
2020-11-23T22:19:28 #kisslinux <dilyn> external proof-readers are overrated
2020-11-23T22:19:39 #kisslinux <dilyn> I just need to become literate and competent! the obvious solution
2020-11-23T22:19:42 #kisslinux <claudia02> just do EVERYTHING yourself
2020-11-23T22:19:47 #kisslinux <E5ten> dylanaraps: also did you see what I said (I think yesterday) about a possible suggestion for the parent dying on SIGTERM issue? (pretty much, would it work to save $ in a variable and use that with kill in the subshell instead of kill 0?)
2020-11-23T22:22:53 #kisslinux <claudia02> dilyn, your storage article has been merged but does not appear in the local documentation
2020-11-23T22:23:05 #kisslinux <claudia02> or am I blind
2020-11-23T22:23:13 #kisslinux <dilyn> yeah i don't know why it isn't in local docs
2020-11-23T22:23:23 #kisslinux <dilyn> It's in the main index, idk how dylan packages the wiki with kiss
2020-11-23T22:24:31 #kisslinux <claudia02> ah the wiki docs are a fixed commit in 'kiss'.
2020-11-23T22:24:41 #kisslinux <claudia02> *package
2020-11-23T22:28:13 #kisslinux <claudia02> Does qt5 still segfault with latest musl while video playback?
2020-11-23T22:29:45 #kisslinux <dilyn> you're the one who's gonna have to test that :P
2020-11-23T22:30:22 #kisslinux <claudia02> :(
2020-11-23T22:30:46 #kisslinux <claudia02> Do you have binaries up for that?
2020-11-23T22:31:26 #kisslinux <claudia02> seems not.
2020-11-23T22:32:13 #kisslinux <dilyn> i do on kiss-kde but you'd need dbus to test
2020-11-23T22:33:26 #kisslinux <claudia02> I have to spent time for that anyway so that does no dealbreaker.
2020-11-23T22:33:31 #kisslinux <claudia02> nice