💾 Archived View for gemini.ctrl-c.club › ~phoebos › logs › kisslinux-2022-12-30.txt captured on 2023-01-29 at 03:43:30.
-=-=-=-=-=-=-
[2022-12-30T05:52:14Z] <wael_> Hi
[2022-12-30T05:57:16Z] <niceguy5000[m]> HI
[2022-12-30T05:58:57Z] <niceguy5000[m]> Is Unix dead?
[2022-12-30T06:57:34Z] <wael_> Probably, but BSD isnt
[2022-12-30T08:13:21Z] <testuser[m]> Hi
[2022-12-30T14:54:09Z] <testuser[m]> illiliti: is there an accurate way to determine if disk has enough space for pkg installation?
[2022-12-30T14:54:16Z] <testuser[m]> for adding a check
[2022-12-30T14:54:34Z] <testuser[m]> also i need to look into pipewire hotplug, forgot about it
[2022-12-30T16:07:58Z] <illiliti> perhaps you can parse output of df "${KISS_ROOT:-/}"
[2022-12-30T16:09:09Z] <illiliti> not sure if it's accurate way though
[2022-12-30T16:12:29Z] <illiliti> imagine user has separate /usr
[2022-12-30T16:13:42Z] <illiliti> to get accurate size, you have to iterate over df output and take into the account these separate mountpoints
[2022-12-30T16:14:01Z] <illiliti> i believe this is the only way
[2022-12-30T16:15:42Z] <illiliti> oh
[2022-12-30T16:15:54Z] <illiliti> also pseudo filesystems
[2022-12-30T16:16:10Z] <illiliti> you need to ignore them somehow
[2022-12-30T16:31:46Z] <testuser[m]> hmm
[2022-12-30T16:31:49Z] <testuser[m]> how do current pkg managers do it
[2022-12-30T16:33:16Z] <Ellowee[m]> Your'd have to store size of build tree in a file and compare against free on root
[2022-12-30T16:33:48Z] <Ellowee[m]> * Your'd have to store size of build tree in a file and compare against free on root
[2022-12-30T16:33:48Z] <Ellowee[m]> (A post build hook?)
[2022-12-30T16:34:34Z] <illiliti> not that simple because mountpoints
[2022-12-30T16:48:37Z] <illiliti> idea: iterate over package's manifest, get size of each top-level directory and run df -P on that directory in KISS_ROOT to get available size, compare
[2022-12-30T16:49:49Z] <illiliti> but as you see it is not perfect because we don't handle nested mountpoints
[2022-12-30T16:51:06Z] <illiliti> i wish df have a flag to print nested mountpoints
[2022-12-30T16:51:25Z] <illiliti> that would simplify things
[2022-12-30T16:54:38Z] <illiliti> you can still handle nested mountpoints though
[2022-12-30T16:54:51Z] <illiliti> parse parse parse df -P output
[2022-12-30T17:31:09Z] <phoebos> hei illiliti does the opengroup have plans to require make to set a PWD macro in 202x?
[2022-12-30T17:37:21Z] <phoebos> hmm https://www.austingroupbugs.net/view.php?id=338
[2022-12-30T18:05:42Z] <illiliti> no clue
[2022-12-30T18:05:49Z] <illiliti> why do you ask
[2022-12-30T18:06:55Z] <phoebos> would be useful and BSD and GNU have incompatible syntaxes for it
[2022-12-30T18:12:54Z] <illiliti> i see
[2022-12-30T18:13:13Z] <illiliti> smake also has CURDIR
[2022-12-30T18:19:48Z] <illiliti> i'll send a proposal when i get free time
[2022-12-30T18:32:47Z] <phoebos> nice
[2022-12-30T18:32:50Z] <phoebos> let me know if i can help
[2022-12-30T19:15:42Z] <sad_plan> is there a floating wayland wm thats reasonably good? I only know of spkp, which it seems its really working :p
[2022-12-30T19:52:20Z] <testuser[m]> I don't understand your question. Monero doesn't need any wrappers to use LMDB. Nor does LMDB have any memory management bugs, so a language like rust would only bloat it and slow it down.
[2022-12-30T20:02:25Z] <phoebos> sad_plan: wayfire?
[2022-12-30T20:02:53Z] <sad_plan> phoebos: hm, isnt that what dilyn used earlier? Ill check it out
[2022-12-30T20:04:29Z] <sad_plan> *3D*. fancy stuff
[2022-12-30T20:13:53Z] <sad_plan> hikari was the one which I thought about, but dilyn did use wayfire
[2022-12-30T20:15:52Z] <illiliti> phoebos: wait, what make should put into CURDIR if run with -f path/to/Makefile
[2022-12-30T20:18:49Z] <illiliti> implementations seem to not handle this case
[2022-12-30T21:23:54Z] <phoebos> illiliti: make -C is an explicit chdir, -f is just using a different makefile, so I think CURDIR should not be changed by -f but should be by -C
[2022-12-30T21:24:02Z] <phoebos> bmake does this
[2022-12-30T21:24:39Z] <phoebos> gnu make distinguishes PWD and CURDIR when doing -C
[2022-12-30T21:26:28Z] <phoebos> in that case, PWD is where you run it and CURDIR is where make is running (-C CURDIR)
[2022-12-30T21:26:30Z] <phoebos> weird
[2022-12-30T21:27:01Z] <phoebos> oh no i forgot gmake doesn't set PWD
[2022-12-30T21:29:17Z] <illiliti> -C is no problem
[2022-12-30T21:29:44Z] <illiliti> i afraid -f will break CURDIR
[2022-12-30T21:30:33Z] <illiliti> cd /tmp && make -f /tmp/project/Makefile
[2022-12-30T21:30:58Z] <phoebos> why? because a makefile using CURDIR will expect CURDIR to be the same dir as it's written in?
[2022-12-30T21:31:14Z] <illiliti> yes2
[2022-12-30T21:31:56Z] <phoebos> then there should be a note that in that case it is necessary to do cd dir; make rather than make -f dir/Makefile
[2022-12-30T21:34:57Z] <phoebos> > in particular note that if you include files from other directories the value of CURDIR does not change
[2022-12-30T21:35:51Z] <phoebos> is PWD a better name because of the incompatibility with CURDIR/.CURDIR
[2022-12-30T21:37:49Z] <illiliti> PWD would be an invention; not suitable for addition
[2022-12-30T21:42:24Z] <illiliti> which incompatibility btw? i only see that bmake need leading dot, but that's minor
[2022-12-30T21:47:31Z] <illiliti> oh wait, PWD is documented in bmake. then it would not be invention
[2022-12-30T21:50:26Z] <phoebos> yeah i just meant the leading dot