💾 Archived View for gemini.ctrl-c.club › ~phoebos › logs › freenode-kisslinux-2020-05-04.txt captured on 2023-01-29 at 04:02:56.
⬅️ Previous capture (2021-12-17)
-=-=-=-=-=-=-
2020-05-04T11:56:27 #kisslinux <illiliti> https://github.com/kisslinux/init/blob/master/lib/init/rc.shutdown#L47-L56 << this code not working due to dmsetup requires /proc which was unmounted
2020-05-04T12:07:21 #kisslinux <illiliti> https://github.com/kisslinux/init/blob/master/lib/init/rc.shutdown#L40 << also, why lazy? this command is very dangerous and may lead to data loss...
2020-05-04T12:15:30 #kisslinux <ax> good day
2020-05-04T12:28:34 #kisslinux <dylan03> illiliti: Regarding "lazy" unmounting: This was actually to fix issues like what you mentioned. Context: https://freenode.logbot.info/kisslinux/20200316#c3420388
2020-05-04T12:29:22 #kisslinux <dylan03> The first commit which added this: https://github.com/kisslinux/init/commit/03e5a99d01f0644055c703a5e2d5db59a2c20a6b#
2020-05-04T12:29:38 #kisslinux <dylan03> More context: https://freenode.logbot.info/kisslinux/20200315#c3416846
2020-05-04T12:29:55 #kisslinux <dylan03> paradigm: You might have more to add here (rationale).
2020-05-04T12:31:07 #kisslinux <illiliti> https://github.com/kisslinux/init/blob/master/lib/init/rc.shutdown#L53 << this check is useless. you need to close all containers line by line, not only 0
2020-05-04T12:31:13 #kisslinux <dylan03> illiliti: Regarding the /proc issue; Should I move 'dmsetup' to _before_ we unmount the disks? (Leaving 'cryptsetup' to run where it already does)
2020-05-04T12:31:30 #kisslinux <dylan03> illiliti:
2020-05-04T12:31:35 #kisslinux <dylan03> Alright
2020-05-04T12:32:01 #kisslinux <dylan03> I'll make the changes (as I understand them) and I'll await more comments. I appreciate how much you care about correctness here.
2020-05-04T12:33:13 #kisslinux <dylan03> https://termbin.com/gr4d
2020-05-04T12:33:30 #kisslinux <dylan03> This is for your most recent suggestion. From what I understand, this is what you meant.
2020-05-04T12:33:59 #kisslinux <paradigm> I feel like the comment is sufficient. busybox's `mount -a` does not reliably work if there are nested mount points. If `mount -a -l` is undesirable, another fix would be effectively implementing `mount -a` ourselves with a loop. It's certainly an option, but not as terse as `moutn -a -l`.
2020-05-04T12:35:09 #kisslinux <illiliti> dylan03: you need to preserve /proc or store dmsetup output
2020-05-04T12:36:22 #kisslinux <paradigm> I'd also like to learn moure about how lazy unmounting is dangerous. Most documentation I've read about it mentions its use explicitly during the shutdown process to avoid hanging on things like network mounts.
2020-05-04T12:37:45 #kisslinux <dylan03> illiliti: If you could send a PR for the encryption related code I'd appreciate it. I don't personally use encryption so I'd need you to test each change.
2020-05-04T12:37:59 #kisslinux <dylan03> paradigm: Thanks for chiming in so quickly.
2020-05-04T12:38:03 #kisslinux <paradigm> Happy to help
2020-05-04T12:39:11 #kisslinux <illiliti> dylan03: i'll try
2020-05-04T12:41:50 #kisslinux <paradigm> If the lazy unmount was before the kall 9 I could certainly understand the raised point; processes could remain which could have continued writes after the umount was queued. However, that's not the case here. I could also see potential concerns if it was after the sync call, but again that's not the case here. This seems the proper and expected place to do it, as far as I understand lazy
2020-05-04T12:41:56 #kisslinux <paradigm> unmounting.
2020-05-04T12:43:52 #kisslinux <illiliti> https://github.com/kisslinux/init/blob/master/lib/init/rc.shutdown#L23 << kill stdout here
2020-05-04T12:44:18 #kisslinux <dylan03> Shouldn't it be stderr?
2020-05-04T12:44:19 #kisslinux <illiliti> same for https://github.com/kisslinux/init/blob/master/lib/init/rc.boot#L99
2020-05-04T12:44:23 #kisslinux <dylan03> But yes, you're right.
2020-05-04T12:44:35 #kisslinux <illiliti> yes, it stderr
2020-05-04T12:47:02 #kisslinux <illiliti> https://github.com/kisslinux/init/blob/master/lib/init/rc.lib#L12 << grep -q "$1" /proc/mounts
2020-05-04T12:47:58 #kisslinux <dylan03> Instead of mountpoint?
2020-05-04T12:48:08 #kisslinux <illiliti> yes
2020-05-04T12:48:09 #kisslinux <dylan03> I can do this throughout (and with pure shell).
2020-05-04T12:48:53 #kisslinux <illiliti> grep -q more performant theoretically
2020-05-04T12:49:52 #kisslinux <paradigm> I think using awk with == against the specific field would be slightly more robust/correct than grep; consider the admitedly unlikely possibility that a directory being checked has regex special characters in the name.
2020-05-04T12:50:22 #kisslinux <dylan03> Yup
2020-05-04T12:50:29 #kisslinux <dylan03> My shell implementation will do this
2020-05-04T12:50:33 #kisslinux <paradigm> grep could also false-positive against the "device" field with a bind-mount
2020-05-04T12:51:42 #kisslinux <dylan03> Yup
2020-05-04T12:52:32 #kisslinux <dylan03> https://termbin.com/0bxo
2020-05-04T12:53:01 #kisslinux <dylan03> This grabs the second field of each line and compares it to the function's input.
2020-05-04T12:53:18 #kisslinux <dylan03> mountpoint is used instead if available.
2020-05-04T12:54:10 #kisslinux <dylan03> Does /proc/mounts always exist?
2020-05-04T12:54:21 #kisslinux <dylan03> If so, I'll remove the mountpoint code.
2020-05-04T12:54:24 #kisslinux <paradigm> if procfs is mounted on /lproc, yes
2020-05-04T12:54:33 #kisslinux <dylan03> Regardless of kernel configuration?
2020-05-04T12:55:05 #kisslinux <paradigm> I haven't reviewed configuration options recently enough to confirm that definitively
2020-05-04T12:56:04 #kisslinux <dylan03> > Before kernel 2.4.19, this file was a list of all the file systems currently mounted on the system. With the introduction of per-process mount namespaces in Linux 2.4.19, this file became a link to /proc/self/mounts, which lists the mount points of the process's own mount namespace. The format of this file is documented in fstab(5).
2020-05-04T12:56:08 #kisslinux <dylan03> https://linux.die.net/man/5/proc
2020-05-04T12:56:25 #kisslinux <paradigm> If that's compiled out, I don't think `mountpoint` would work either. I'd bet any reasonable `mountpoint` implementation reads one of /proc/mounts or /proc/self/mounts or /proc/<pid>/mounts or /proc/<pid>/mountinfo
2020-05-04T12:56:52 #kisslinux <dylan03> https://git.busybox.net/busybox/tree/miscutils/mountpoint.c?h=1_1_stable&id=eeed2306fceacf2baea77712b309b4fa40e671b6#n80
2020-05-04T12:57:17 #kisslinux <dylan03> busybox mountpoint doesn't seem to use /proc?
2020-05-04T12:58:05 #kisslinux <paradigm> ooooh, I see
2020-05-04T12:58:26 #kisslinux <paradigm> That's a less robust solution than /proc, but yes, that works too
2020-05-04T12:58:50 #kisslinux <illiliti> https://git.suckless.org/ubase/file/mountpoint.c.html#l75
2020-05-04T13:01:12 #kisslinux <dylan03> Nice
2020-05-04T13:01:15 #kisslinux <dylan03> That works.
2020-05-04T13:02:01 #kisslinux <paradigm> file/directory metadata contains information about the device ("st_dev") its on. If a given file/directory's device name differs from the parent, you know it has to be a mount point; that's the only way to get that relationship with different device names. However, I think can false-negative if you bind-mount; in that case, there's multiple mount points with the same device name. ubase's solution
2020-05-04T13:02:07 #kisslinux <paradigm> there is more robust; the kernel knows where the mount points are, and that's how it tells you. busybox's solution is a good-enough one that will fail in complicated scenarios. Same reason we needed the lazy unmounting; busybox's `umount -a` also makes assumptions that aren't guarenteed.
2020-05-04T13:03:10 #kisslinux <dylan03> Interesting
2020-05-04T13:04:19 #kisslinux <paradigm> busybox's find silently ignores -xdev when combine with -delete and/or -depth, which I learned the hard way when a `find -xdev -delete` started deleting things on different mount points. busybox is nice and small, but it gets there by sometimes cutting corners.
2020-05-04T13:05:03 #kisslinux <dylan03> It's more that the development has no plan at all
2020-05-04T13:05:15 #kisslinux <dylan03> Random people send in random patches for random things.
2020-05-04T13:05:29 #kisslinux <dylan03> So long as everything works "good enough", it's likely that no one will touch it again.
2020-05-04T13:05:51 #kisslinux <dylan03> Here's my mountpoint implementation: https://termbin.com/z457
2020-05-04T13:07:15 #kisslinux <paradigm> To be hyper pedantic, non-directory files can be mount points; try `mount --bind /path/to/file /path/to/another/file`
2020-05-04T13:07:25 #kisslinux <dylan03> Alright
2020-05-04T13:07:35 #kisslinux <dylan03> [ -e ... ] would be better then
2020-05-04T13:09:16 #kisslinux <paradigm> That's actually a good test case for busybox's mountpoint issue; `touch foo bar; mount --bind foo bar; mountpoint bar` will probably make busybox's mountpoint false negative where ubase's or gnu's or hopefully yours will get it correct
2020-05-04T13:10:13 #kisslinux <dylan03> ubase's assumes a directory as well
2020-05-04T13:10:33 #kisslinux <paradigm> interesting
2020-05-04T13:11:57 #kisslinux <dylan03> Interestingly, sysvinit's mountpoint exits with 0 if /proc/mounts doesn't exist.
2020-05-04T13:12:02 #kisslinux <dylan03> https://fossies.org/linux/sysvinit/src/mountpoint.c
2020-05-04T13:12:09 #kisslinux <dylan03> (Line 80)
2020-05-04T13:12:27 #kisslinux <illiliti> util-linux fallbacks to non-proc method if /proc doesn't exist
2020-05-04T13:12:30 #kisslinux <dylan03> Nice. I need to also trim trailing '/' from the input.
2020-05-04T13:13:24 #kisslinux <dylan03> sysvinit's also checks to see if it is a directory.
2020-05-04T13:13:48 #kisslinux <paradigm> Ooh, file path normalization might be a point of concern; what if an item in the path is a symlink?
2020-05-04T13:14:51 #kisslinux <dylan03> paradigm: Our usage is very basic.
2020-05-04T13:14:52 #kisslinux <dylan03> https://termbin.com/1yx3j
2020-05-04T13:15:12 #kisslinux <dylan03> The field after 'mnt' is what we use as input to 'mountpoint -q'
2020-05-04T13:15:25 #kisslinux <paradigm> gotcha
2020-05-04T13:15:36 #kisslinux <dylan03> The init's implementation can be rather basic
2020-05-04T13:15:55 #kisslinux <paradigm> in that case, yes, I'd YAGNI the complexity of normalizing the file paths
2020-05-04T13:25:36 #kisslinux <dylan03> https://github.com/kisslinux/init/commit/3c1e74ec6569a62d0b379a5f026bff62324f01fc.
2020-05-04T13:25:46 #kisslinux <dylan03> This is the mountpoint implementation.
2020-05-04T13:25:54 #kisslinux <dylan03> https://github.com/kisslinux/init/commit/d6887c806ce2f7bf4cc6e050f810ffc1b9ffc79b
2020-05-04T13:26:07 #kisslinux <dylan03> This is the 'dd' change + a move to a function.
2020-05-04T13:26:11 #kisslinux <dylan03> https://github.com/kisslinux/init/commit/7561a7eb5b475136751856cd120cdd25753423f0
2020-05-04T13:26:19 #kisslinux <dylan03> This is the cryptsetup close change.
2020-05-04T13:26:45 #kisslinux <dylan03> illiliti: I'd like to hear your thoughts. Keep the suggestions coming!
2020-05-04T13:30:08 #kisslinux <illiliti> excellent work!
2020-05-04T13:38:19 #kisslinux <illiliti> regarding encryption related stuff. void-linux preserves /proc https://github.com/void-linux/void-runit/blob/master/3#L46 . but they uses not portable options to do it....
2020-05-04T13:40:37 #kisslinux <dylan03> We could iterate over /proc/mounts to exclude /proc.
2020-05-04T13:41:07 #kisslinux <dylan03> Though... How would we then handle '-l'?
2020-05-04T13:41:18 #kisslinux <dylan03> Can we simply remount /proc?
2020-05-04T13:42:23 #kisslinux <dylan03> illiliti: https://termbin.com/xlwhq
2020-05-04T13:42:23 #kisslinux <illiliti> need to check it
2020-05-04T13:52:38 #kisslinux <dylan03> brb
2020-05-04T13:59:50 #kisslinux <dylan03> brb
2020-05-04T14:00:19 #kisslinux <illiliti> by the way, `halt` doesn't halt. i'm using runit from community
2020-05-04T14:02:21 #kisslinux <dylan04> illiliti: https://github.com/kisslinux/init/commit/6bc2f6a5740dd40acaef743c8eaac2cb91a7a4c3
2020-05-04T14:02:36 #kisslinux <dylan04> > => Boot stage completed in 1s...
2020-05-04T14:02:56 #kisslinux <dylan04> All boot messages will appear in 'dmesg' and boot time is counted in the cheapest way possible.
2020-05-04T14:06:04 #kisslinux <illiliti> nice!
2020-05-04T14:06:07 #kisslinux <illiliti> read the log please
2020-05-04T14:06:34 #kisslinux <illiliti> kernel shutted down(system halted message), but machine is still working
2020-05-04T14:07:11 #kisslinux <illiliti> poweroff command works fine
2020-05-04T14:11:07 #kisslinux <dylan04> Alright
2020-05-04T14:11:09 #kisslinux <dylan04> I'll look into it
2020-05-04T14:11:28 #kisslinux <dylan04> The issue is that runit (from community) 'halt' doesn't work?
2020-05-04T14:11:57 #kisslinux <dylan04> merakor:
2020-05-04T14:15:07 #kisslinux <dylan04> Ah
2020-05-04T14:15:27 #kisslinux <dylan04> illiliti: You're using busybox's 'halt' with (real) runit as init.
2020-05-04T14:15:49 #kisslinux <dylan04> runit doesn't actually come with 'halt'. Void Linux ship their own with their init stuff.
2020-05-04T14:15:52 #kisslinux <dylan04> https://github.com/void-linux/void-runit/blob/master/halt.c
2020-05-04T14:17:11 #kisslinux <dylan04> I think we simply remove 'halt' from KISS entirely. It's just a matter of dropping it from busybox. poweroff/reboot fill this gap just fine.
2020-05-04T14:18:01 #kisslinux <illiliti> just symlink it to poweroff. it should be enought
2020-05-04T14:18:38 #kisslinux <dylan04> They have different command-line usage no?
2020-05-04T14:23:30 #kisslinux <illiliti> idk
2020-05-04T14:28:34 #kisslinux <dylan04> They do
2020-05-04T14:28:39 #kisslinux <dylan04> https://linux.die.net/man/8/halt
2020-05-04T14:31:09 #kisslinux <illiliti> ok, nvm. i'll just create alias to poweroff
2020-05-04T14:31:22 #kisslinux <dylan04> Yeah
2020-05-04T14:31:29 #kisslinux <dylan04> I pushed the commit dropping it.
2020-05-04T14:31:53 #kisslinux <dylan04> The goal is for our init scripts is to be init agnostic.
2020-05-04T14:36:20 #kisslinux <illiliti> brb
2020-05-04T14:48:27 #kisslinux <konimex> (real) runit's halt-poweroff system is just "/sbin/runit-init 0"
2020-05-04T15:05:58 #kisslinux <dylan04> o/
2020-05-04T15:08:46 #kisslinux <illiliti> https://termbin.com/xlwhq < i applied this change, but this fails with mount: wrong fs type, bad option, bad superblock on /dev/sdb1, missing codepage or helper program, or other error
2020-05-04T15:09:03 #kisslinux <illiliti> s|/dev/sdb1/proc/
2020-05-04T15:10:16 #kisslinux <illiliti> https://github.com/kisslinux/init/blob/master/lib/init/rc.shutdown#L43 << this command is not working too
2020-05-04T15:12:53 #kisslinux <dylan04> I guess it's due to /proc being unmounted prior huh...
2020-05-04T15:13:20 #kisslinux <dylan04> The 'umount' command has no specification from what I can see.
2020-05-04T15:13:33 #kisslinux <dylan04> How portable is '-t'?
2020-05-04T15:14:04 #kisslinux <illiliti> it has https://refspecs.linuxfoundation.org/LSB_4.1.0/LSB-Core-generic/LSB-Core-generic/umount.html
2020-05-04T15:14:13 #kisslinux <dylan04> > A umount command appeared in Version 6 AT&T UNIX.
2020-05-04T15:14:16 #kisslinux <dylan04> Ah
2020-05-04T15:14:17 #kisslinux <dylan04> LSB
2020-05-04T15:15:02 #kisslinux <dylan04> I'll start using '-t'.
2020-05-04T15:15:08 #kisslinux <illiliti> no
2020-05-04T15:15:16 #kisslinux <dylan04> ?
2020-05-04T15:15:25 #kisslinux <dylan04> The standard is Linux only huh?
2020-05-04T15:15:57 #kisslinux <illiliti> ubase umount doesn't have '-t'
2020-05-04T15:16:33 #kisslinux <dylan04> > ubase is a collection of unportable tools, similar in spirit to util-linux but much simpler.
2020-05-04T15:16:38 #kisslinux <dylan04> Says so right on the tin. :P
2020-05-04T15:16:42 #kisslinux <dylan04> Unportable tools
2020-05-04T15:18:28 #kisslinux <dylan04> FreeBSD supports it: https://www.freebsd.org/cgi/man.cgi?query=umount&sektion=8
2020-05-04T15:18:45 #kisslinux <dylan04> As does OpenBSD: https://man.openbsd.org/umount.8
2020-05-04T15:18:59 #kisslinux <dylan04> As does NetBSD: https://netbsd.gw.com/cgi-bin/man-cgi?umount+8+NetBSD-current
2020-05-04T15:19:16 #kisslinux <dylan04> As does Toybox: https://github.com/gfto/toybox/blob/master/toys/lsb/umount.c
2020-05-04T15:23:46 #kisslinux <dylan05> At some point you have to question: Am I unportable for using <flag> or is <implementation> unportable for not supporting <flag>.
2020-05-04T15:24:20 #kisslinux <illiliti> so... we will drop umount as we did it for stat, su, ... or implement '-t' using /proc/mounts?
2020-05-04T15:26:20 #kisslinux <dylan05> Well. How will we handle '-l', ordering, etc with our own implementation?
2020-05-04T15:26:52 #kisslinux <dylan05> It's getting rather absurd.
2020-05-04T15:29:07 #kisslinux <dylan05> mcf: How do you feel about adding '-t' to (your) ubase?
2020-05-04T15:29:45 #kisslinux <dylan05> I'm unaware of any other (usable on Linux) implementations of umount without '-t'.
2020-05-04T15:32:01 #kisslinux <dylan05> ubase 'umount -a' should skip '/proc' from what I understand. https://github.com/michaelforney/ubase/blob/master/umount.c#L24
2020-05-04T15:32:16 #kisslinux <dylan05> Do other 'umount' implementations match this behavior?
2020-05-04T15:33:35 #kisslinux <dylan05> Busybox's doesn't match from a quick glance at its source.
2020-05-04T15:43:59 #kisslinux <illiliti> i realized that '-l' flag intercepts '-a' which lead to /proc being unmounted
2020-05-04T15:44:29 #kisslinux <illiliti> at least in util-linux
2020-05-04T15:56:22 #kisslinux <dylan05> We're going to use '-t'. It's in the LSB and ubase only supports Linux...
2020-05-04T15:56:47 #kisslinux <dylan05> Every umount implementation I can find for Linux (and even BSDs) has '-t'.
2020-05-04T15:56:56 #kisslinux <dylan05> https://github.com/michaelforney/ubase/commit/97cbad7eefeb1cf2b2b1f52bc42b18ea017c52b9
2020-05-04T15:57:02 #kisslinux <dylan05> > This is becoming a nightmare. Just support Linux.
2020-05-04T15:57:19 #kisslinux <dylan05> https://github.com/michaelforney/ubase
2020-05-04T15:57:25 #kisslinux <dylan05> > ubase - suckless linux base utils
2020-05-04T16:06:15 #kisslinux <dylan05> https://github.com/kisslinux/init/commit/5975576ebded1bcf9103d1f5d21780d71b30f1ca
2020-05-04T16:27:03 #kisslinux <illiliti> dylan05: this change doesn't fix the problem. lazy option intercepts '-t', read the log
2020-05-04T16:28:12 #kisslinux <dylan05> In which implementation of umount?
2020-05-04T16:28:19 #kisslinux <illiliti> util-linux
2020-05-04T16:28:26 #kisslinux <illiliti> '-l' also unmounts '/', so this https://github.com/kisslinux/init/blob/master/lib/init/rc.shutdown#L56 is useless too
2020-05-04T16:29:43 #kisslinux <dylan05> Wonderful
2020-05-04T16:30:06 #kisslinux <illiliti> :)
2020-05-04T16:30:51 #kisslinux <illiliti> maybe it's bug, idk
2020-05-04T16:42:52 #kisslinux <dylan05> Regarding busybox umount's -a being broken: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=257625
2020-05-04T16:42:55 #kisslinux <dylan05> I don't think it is?
2020-05-04T16:42:59 #kisslinux <dylan05> "Fixed in 2004"
2020-05-04T16:45:29 #kisslinux <himmalerin> python dependencies are a mess. I'm trying to package a single thing (magic-wormhole) and I've already got four python packages and I'm not even a quarter of the way done
2020-05-04T16:45:47 #kisslinux <himmalerin> I wonder if the way to go is through pip instead of making actual packages for everything
2020-05-04T16:46:32 #kisslinux <dylan05> himmalerin: Yeah. I'm really hesitant to allow python-* packages in Community for this reason. It's a mess.
2020-05-04T16:51:30 #kisslinux <ominous_anonymou> I wouldn't allow them, agree with you
2020-05-04T17:01:06 #kisslinux <ominous_anonymou> @hammerin, I was able to `pip install --user magic-wormhole` and it installed fine. I think through pip is the way to go except for one-offs where this isn't possible (ex. scipy might need some extra system packages outside of Python that would need to be built and installed)
2020-05-04T17:06:12 #kisslinux <himmalerin> ominous_anonymou: Where do you get pip from? I don't see it in community
2020-05-04T17:06:40 #kisslinux <ominous_anonymou> https://pip.pypa.io/en/stable/installing/ you can curl the install file
2020-05-04T17:06:47 #kisslinux <dilynm> It's bundled with python
2020-05-04T17:07:06 #kisslinux <ominous_anonymou> dilynm not always
2020-05-04T17:07:14 #kisslinux <dilynm> In our case
2020-05-04T17:07:18 #kisslinux <ominous_anonymou> ah ok
2020-05-04T17:08:28 #kisslinux <dylan05> Yeah
2020-05-04T17:08:49 #kisslinux <dylan05> I bundle pip with Python to match the behavior of node, rust, etc.
2020-05-04T17:09:53 #kisslinux <ominous_anonymou> has it always been bundled?
2020-05-04T17:10:09 #kisslinux <ominous_anonymou> maybe i was thinking of a different VM where i had to install it
2020-05-04T17:10:46 #kisslinux <dylan05> Yes
2020-05-04T17:11:48 #kisslinux <dylan05> This behavior requires a simple patch as Python 3 snoops around / and disables installing pip if it's already available... This is what happens when you make your software too "smart".
2020-05-04T17:11:51 #kisslinux <dylan05> https://termbin.com/qksq
2020-05-04T17:13:25 #kisslinux <ominous_anonymou> yeah i've been a little disappointed with the direction of Python recently. The PEPs that are getting pushed are, in my opinion, bloat and counter to Python's original goals
2020-05-04T17:14:48 #kisslinux <dylan05> Well, the BDFL is gone (The term was created in relation to him!).
2020-05-04T17:14:57 #kisslinux <dylan05> It makes sense that the direction has changed.
2020-05-04T17:19:51 #kisslinux <dylan05> brb
2020-05-04T18:27:30 #kisslinux <himmalerin> ominous_anonymou: I ran the pip install command, where does pip put the executable file? It dumped a whole bunch of stuff in ~/.local, but no magic-wormhole executable as far as I can tell
2020-05-04T18:27:56 #kisslinux <mcf> dylan05: sure, feel free to send a patch
2020-05-04T18:28:22 #kisslinux <ominous_anonymou> is there anything under ~/.local/bin/wormhole?
2020-05-04T18:29:12 #kisslinux <himmalerin> oh that's it, I was searching for magic-wormhole. thanks!
2020-05-04T18:30:46 #kisslinux <ominous_anonymou> sure!
2020-05-04T19:44:44 #kisslinux <dylan01> https://k1ss.org/community
2020-05-04T19:45:02 #kisslinux <dylan01> Created a page on the website dedicated to projects, repositories, forks, etc by users of KISS.
2020-05-04T19:45:04 #kisslinux <dylan01> :)
2020-05-04T19:45:14 #kisslinux <dylan01> Please let me know if I'm missing anything!
2020-05-04T19:47:59 #kisslinux <illiliti> wow! tinyramfs! thank you so much
2020-05-04T19:57:52 #kisslinux <illiliti> dylan01: https://github.com/cemkeylan/sysmgr << init written in posix shell by cemkeylan
2020-05-04T20:01:32 #kisslinux <dylan01> illiliti: No problem!
2020-05-04T20:03:26 #kisslinux <dylan01> Also this: https://github.com/cemkeylan/shinit
2020-05-04T20:03:29 #kisslinux <dylan01> Really neat.
2020-05-04T20:05:21 #kisslinux <dylan01> I've also added a page to list ports of KISS to different architectures: https://k1ss.org/ports
2020-05-04T20:06:00 #kisslinux <dylan01> Next up on my list is a "Post Installation" page.
2020-05-04T20:06:13 #kisslinux <dylan01> https://github.com/kisslinux/website/issues/50
2020-05-04T20:26:17 #kisslinux <illiliti> dylan: btw, https://github.com/dylanaraps/pure-sh-bible#arithmetic << are you sure that `plus-equal` and similars is defined by posix? i can't find any references
2020-05-04T20:33:33 #kisslinux <E5ten> I'm pretty sure they are
2020-05-04T20:35:05 #kisslinux <E5ten> illiliti: POSIX says anything in C arithmetic expansion aside from {in,de}crementation operators, and sizeof should work
2020-05-04T20:35:09 #kisslinux <E5ten> that includes +=
2020-05-04T20:35:37 #kisslinux <E5ten> (obviously only the arithmetic +=, not like var+=string to append)
2020-05-04T20:38:13 #kisslinux <illiliti> ok! thanks
2020-05-04T20:44:31 #kisslinux <dylan02> illiliti: Let me check if it works in mrsh.
2020-05-04T20:46:11 #kisslinux <E5ten> in fairness if it doesn't it means mrsh is broken because POSIX does say it should be supported
2020-05-04T20:59:47 #kisslinux <dylan01> It works in mrsh.
2020-05-04T20:59:53 #kisslinux <dylan01> My net is cutting out every 2 seconds.
2020-05-04T21:00:00 #kisslinux <dylan01> Seems to be OK now.
2020-05-04T21:10:41 #kisslinux <onodera> evening
2020-05-04T21:10:48 #kisslinux <dylan01> ello
2020-05-04T21:14:13 #kisslinux <dylan01> illiliti: Thanks for the comment re post-installation
2020-05-04T21:14:43 #kisslinux <onodera> i bought a receiver from the 70's today
2020-05-04T21:15:26 #kisslinux <dylan01> Nice
2020-05-04T21:18:30 #kisslinux <dylan02> My net is horrible right now.
2020-05-04T21:18:45 #kisslinux <dylan02> onodera: I love old stuff like that. I have a camera from 1953 which I still use.
2020-05-04T21:19:11 #kisslinux <humaid_01> Hi
2020-05-04T21:19:17 #kisslinux <dylan02> Hello
2020-05-04T21:19:30 #kisslinux <hechos> my net too so much streaming now.
2020-05-04T21:19:37 #kisslinux <humaid_01> How are you people?
2020-05-04T21:19:39 #kisslinux <hechos> 4AM work fine!
2020-05-04T21:20:05 #kisslinux <hechos> fine playing kiss
2020-05-04T21:20:31 #kisslinux <humaid_01> Yeah me too :)
2020-05-04T21:46:01 #kisslinux <himmalerin> what normally supplies the `file` command? It seems I'm missing it
2020-05-04T21:55:24 #kisslinux <iwq> dylan: who do the digits in your nickname change every few messages?
2020-05-04T21:55:28 #kisslinux <iwq> why*
2020-05-04T21:56:42 #kisslinux <himmalerin> dylan02: iwq asked "why do the digits in your nickname change every few messages?"
2020-05-04T21:57:33 #kisslinux <hechos> strange...
2020-05-04T21:58:21 #kisslinux <micro_O> I wonder what would be needed for repology to understand there is kiss main, kiss community, and some other kiss repos, they are all for one distro
2020-05-04T21:58:57 #kisslinux <himmalerin> If I want to custom build a package from repo/extra and put the modified version in my personal kiss repo, which one will it build? My kiss repo appears after repo/main in $KISS_PATH
2020-05-04T21:59:20 #kisslinux <micro_O> like i appreciate seeing main and community separate for some tracking purposes, but it would be cool to see kiss-universe stats as well
2020-05-04T21:59:22 #kisslinux <dylan02> The first match will be used. himmalerin:
2020-05-04T21:59:56 #kisslinux <himmalerin> So I should move my kiss repo to appear before repo/main in $KISS_PATH
2020-05-04T22:00:45 #kisslinux <dylan02> micro_O: 'kiss outdated' will check all of your installed packages against Repology's database and report if any are out of date. This works regardless of whether or not Repology knows about your repository.
2020-05-04T22:00:58 #kisslinux <dylan02> himmalerin: Yeah. It works exactly like '$PATH'.
2020-05-04T22:01:54 #kisslinux <himmalerin> I thought $PATH preferred the last match? For /usr/bin/local
2020-05-04T22:02:02 #kisslinux <himmalerin> maybe I'm wrong
2020-05-04T22:02:15 #kisslinux <dylan02> micro_O: Still. It'd be nice for Repology to understand that there are many KISS repositories out there. Repology doesn't support "3rd-party" repositories (and I doubt an exception can be made). :(
2020-05-04T22:02:47 #kisslinux <hechos> dylan02: you are araps
2020-05-04T22:04:17 #kisslinux <dylan02> hechos: I am.
2020-05-04T22:04:39 #kisslinux <hechos> okey
2020-05-04T22:04:45 #kisslinux <micro_O> dylan02: i am reading their format and such. I might be able to scrape together some kinda `kiss-universe` repo, that just has submodules for all the well known (aka, linked from k1ss) repos
2020-05-04T22:05:03 #kisslinux <hechos> i implemented mount efivars on kiss-chroot
2020-05-04T22:05:17 #kisslinux <dylan02> hechos: As did I!
2020-05-04T22:05:39 #kisslinux <hechos> on the last repo?
2020-05-04T22:07:02 #kisslinux <dylan02> himmalerin: > When a command name is specified by the user or an exec call is made from a program, the system searches through $PATH, examining each directory from left to right in the list, looking for a filename that matches the command name. Once found, the program is executed as a child process of the command shell or program that issued the command.
2020-05-04T22:07:07 #kisslinux <dylan02> https://en.wikipedia.org/wiki/PATH_(variable)
2020-05-04T22:07:09 #kisslinux <hechos> where can see the code?
2020-05-04T22:07:32 #kisslinux <dylan02> https://github.com/kisslinux/kiss/blob/master/contrib/kiss-chroot
2020-05-04T22:07:47 #kisslinux <hechos> alright thanks.
2020-05-04T22:07:54 #kisslinux <dylan02> The commit which added support for efivars: https://github.com/kisslinux/kiss/commit/a11d48d5959861f973ec5dcf7beb95ee9583f763
2020-05-04T22:15:14 #kisslinux <hechos> package manager dont resolve deps right
2020-05-04T22:15:16 #kisslinux <hechos> ?
2020-05-04T22:21:48 #kisslinux <dylan02> I need more information
2020-05-04T22:22:08 #kisslinux <hechos> i dont try to offend.
2020-05-04T22:22:14 #kisslinux <dylan02> I know, I know
2020-05-04T22:22:18 #kisslinux <dylan02> :P
2020-05-04T22:22:19 #kisslinux <hechos> just want to know.
2020-05-04T22:22:37 #kisslinux <hechos> because when instal sowm can resolve
2020-05-04T22:22:43 #kisslinux <hechos> have to install manually.
2020-05-04T22:23:51 #kisslinux <dylan02> Got a log of the terminal output?
2020-05-04T22:24:00 #kisslinux <dylan02> Or a set of commands I can use to reproduce the issue?
2020-05-04T22:27:15 #kisslinux <hechos> i dont know
2020-05-04T22:27:25 #kisslinux <hechos> just run kiss b sowm
2020-05-04T22:27:43 #kisslinux <dylan02> What's the issue?
2020-05-04T22:28:00 #kisslinux <hechos> sowm to compile just need libX11
2020-05-04T22:28:02 #kisslinux <hechos> right?
2020-05-04T22:28:40 #kisslinux <dylan02> Yes. Also 'make' though this is an implicit dependency.
2020-05-04T22:31:51 #kisslinux <hechos> aha...
2020-05-04T22:34:10 #kisslinux <himmalerin> dylan02: would you accept a pr to make ffmpeg build with gnutls or is that a no since gnutls is in community? I'd like ffmpeg to be able to use input files via https://.
2020-05-04T22:34:21 #kisslinux <himmalerin> It can work libressl instead of gnutls, but then you also need to flip the --enable-nonfree option which I'm not sure you want
2020-05-04T22:38:37 #kisslinux <dylan02> I'll enable the nonfree option. It's very badly named. It doesn't mean proprietary code. It really just means "non-redistributable" (due to licensing issues).
2020-05-04T22:38:52 #kisslinux <dylan02> For a source based distribution (like KISS) it's fine.
2020-05-04T22:38:56 #kisslinux <hechos> lol!!
2020-05-04T22:39:42 #kisslinux <hechos> i like nonfree explanation!
2020-05-04T22:40:46 #kisslinux <himmalerin> pr or something else? If you just want to do it all it needs is --enable-openssl and --enable-nonfree in the build file, and probably(? not all dependencies are in there either way) libressl in the depends file
2020-05-04T22:40:53 #kisslinux <hechos> dylan02: have compiled packages how can index to pkgmanager
2020-05-04T22:45:56 #kisslinux <dylan02> himmalerin: Pushed to repositories (kiss u + kiss b ffmpeg + kiss i ffmpeg)
2020-05-04T22:46:12 #kisslinux <dylan02> hechos: What do you mean?
2020-05-04T22:47:05 #kisslinux <hechos> okey got all work on my desktop and try to move to laptop
2020-05-04T22:47:18 #kisslinux <hechos> what need to restore system
2020-05-04T22:47:48 #kisslinux <hechos> i mind pre-compiled packages and database for kiss read alredy have compile.
2020-05-04T22:47:53 #kisslinux <dylan02> Ah. Installing the binaries to another machine.
2020-05-04T22:48:03 #kisslinux <hechos> exactly.
2020-05-04T22:48:09 #kisslinux <dylan02> You could simply put them in ~/.cache/kiss/bin
2020-05-04T22:48:24 #kisslinux <dylan02> Or you can install them via 'kiss i /path/to/tarball.tar.gz'
2020-05-04T22:48:48 #kisslinux <hechos> ohh okey.
2020-05-04T22:49:01 #kisslinux <hechos> should be wokk kiss i /path/file?
2020-05-04T22:49:13 #kisslinux <hechos> nice!
2020-05-04T22:49:32 #kisslinux <hechos> im looking how can automate some parts of deploy.
2020-05-04T22:49:48 #kisslinux <dylan02> Yeah
2020-05-04T22:50:39 #kisslinux <himmalerin> dylan02: awesome, thanks!
2020-05-04T22:50:44 #kisslinux <himmalerin> no rev bump thought?
2020-05-04T22:50:51 #kisslinux <himmalerin> s/thought/though
2020-05-04T22:51:21 #kisslinux <dylan02> Well, it doesn't always make sense to force a rebuild onto every user.
2020-05-04T22:51:59 #kisslinux <himmalerin> fair enough
2020-05-04T22:52:48 #kisslinux <dylan02> Users will receive these changes when ffmpeg updates next. Those wanting them now can simply rebuild ffmpeg.
2020-05-04T22:53:23 #kisslinux <dylan02> This works really well.
2020-05-04T23:17:52 #kisslinux <hechos> can be a problem to due for example, kiss b package1,pakcage2.... and so..?
2020-05-04T23:20:57 #kisslinux <dylan02> 'kiss b pkg pkg pkg' is the usage.
2020-05-04T23:21:22 #kisslinux <micro_O> coooool so i have a local copy of repology running
2020-05-04T23:21:45 #kisslinux <hechos> great im still compiling...
2020-05-04T23:21:48 #kisslinux <micro_O> which means i can experiment with combining repos
2020-05-04T23:33:26 #kisslinux <hechos> Zond block device support is just for RAID0 1 10 DUP ?