💾 Archived View for gemini.ctrl-c.club › ~phoebos › logs › freenode-kisslinux-2020-09-27.txt captured on 2024-05-26 at 16:22:01.
⬅️ Previous capture (2021-12-17)
-=-=-=-=-=-=-
2020-09-27T01:57:12 #kisslinux <muevoid> What are peoples thought's on statically linking vs dynamically linking.
2020-09-27T02:00:52 #kisslinux <muevoid> I think I learn more towards statically linking but want other peoples thoughts.
2020-09-27T02:50:19 #kisslinux <dilyn> afaik static linking is preferable in particular contexts but is in general not the most important thing
2020-09-27T02:50:25 #kisslinux <dilyn> especially if what you value is speed
2020-09-27T02:52:40 #kisslinux <muevoid> Afaik can't static binaries be just as fast if not faster? Due to not having to look up dynamic libraries while starting?
2020-09-27T02:56:17 #kisslinux <muevoid> With my fork of kiss I am setting up a file system like sta.li file system. Right now most things are still dynamically linked however I am considering static linking.
2020-09-27T02:56:41 #kisslinux <konimex> things like qt5/gtk/mesa probably won't be able to be statically linked
2020-09-27T02:56:52 #kisslinux <mcpcpc[m]> muevoid: what terminal emulator were using when you were using `kirc` earlier?
2020-09-27T03:02:01 #kisslinux <muevoid> foot
2020-09-27T03:02:20 #kisslinux <muevoid> and yes I'm aware some things can't be statically linked konimex
2020-09-27T03:04:05 #kisslinux <muevoid> what is <net/route.h> provided by?
2020-09-27T03:04:11 #kisslinux <dilyn> statically linked files theoretically perform faster but afaik tests have shown little to no improvement
2020-09-27T03:04:18 #kisslinux <dilyn> there are benchmarks on musl's website iirc
2020-09-27T03:04:38 #kisslinux <dilyn> muevoid: musl
2020-09-27T03:04:43 #kisslinux <muevoid> I think I will most likely stick with dynamic linking as of right now.
2020-09-27T03:04:52 #kisslinux <muevoid> Ah kk
2020-09-27T03:05:57 #kisslinux <muevoid> For some reason my musl build isn't installing it :<
2020-09-27T03:06:10 #kisslinux <muevoid> Nevermind fixed it
2020-09-27T03:06:34 #kisslinux <muevoid> Had to do PREFIX=/
2020-09-27T03:13:53 #kisslinux <muevoid> Now gcc is failing :)
2020-09-27T03:20:32 #kisslinux <muevoid> #include <new>?
2020-09-27T03:29:03 #kisslinux <dilyn> ?
2020-09-27T03:29:55 #kisslinux <muevoid> Gcc is failing cause it can't find #include <new> I am trying to figure it out
2020-09-27T03:30:08 #kisslinux <dilyn> what step is it failing at?
2020-09-27T03:30:10 #kisslinux <muevoid> The only thing I changed is removed all the /usr
2020-09-27T03:30:13 #kisslinux <muevoid> gmp
2020-09-27T03:34:37 #kisslinux <muevoid> dilyn is that provided by musl probably and I scuffed my musl build could you do kiss-manifest musl dilynm?
2020-09-27T03:34:48 #kisslinux <dilyn> is gmp provided by musl?
2020-09-27T03:34:56 #kisslinux <dilyn> no it's required by gcc and should be built alongside it
2020-09-27T03:35:16 #kisslinux <dilyn> see KISS' gcc build script; it's comprehensive and basically the minimum for a full gcc
2020-09-27T03:35:23 #kisslinux <dilyn> if you just want a bootstrap or something refer to lfs
2020-09-27T03:35:47 #kisslinux <muevoid> I am basically doing the same as KISS' gcc build script just with all the /usr's replaced
2020-09-27T03:36:21 #kisslinux <muevoid> https://termbin.com/vstf
2020-09-27T03:39:39 #kisslinux <dilyn> ooooffff
2020-09-27T03:39:40 #kisslinux <dilyn> hmm
2020-09-27T03:39:51 #kisslinux <muevoid> Yeah lmao
2020-09-27T03:40:34 #kisslinux <dilyn> well, you're probably going to run into issues with your default include path
2020-09-27T03:41:23 #kisslinux <muevoid> The weird thing is I have a symlink from /usr to / so I would think it wouldn't be an issue
2020-09-27T03:42:16 #kisslinux <muevoid> Is #include <new> just like in the gcc sources or do you have it in /usr/include?
2020-09-27T03:43:17 #kisslinux <dilyn> i have a /usr/include/c++/v1/new
2020-09-27T03:43:19 #kisslinux <dilyn> owned by... nothing
2020-09-27T03:43:27 #kisslinux <muevoid> hmmm
2020-09-27T03:43:29 #kisslinux <dilyn> presumably owned by llvm
2020-09-27T03:43:41 #kisslinux <dilyn> the comments suggest as much lol
2020-09-27T03:43:51 #kisslinux <muevoid> Anyone on here have gcc that can check?
2020-09-27T03:44:19 #kisslinux <muevoid> I'm very confused lmao
2020-09-27T03:44:38 #kisslinux <dilyn> it's owned by gcc on KISS
2020-09-27T03:45:10 #kisslinux <muevoid> Hmmmm
2020-09-27T03:45:52 #kisslinux <muevoid> I'm trying with --prefix=/usr just to see if that is indeed the issue
2020-09-27T03:47:42 #kisslinux <dilyn> are you building gmp in the gcc source tree?
2020-09-27T03:47:48 #kisslinux <dilyn> in like gcc-10.2.0/gmp?
2020-09-27T03:48:14 #kisslinux <muevoid> I think it may be due to one of the other core packages as using the default gc I think it may be due to one of the other core packages as using the exact gcc build from kiss doesn't help
2020-09-27T03:49:29 #kisslinux <muevoid> Sorry for the repeat kirc is being weird
2020-09-27T03:49:53 #kisslinux <mcpcpc[m]> :S
2020-09-27T03:50:19 #kisslinux <mcpcpc[m]> muevoid: same issue as before?
2020-09-27T03:50:25 #kisslinux <muevoid> :P didn't mean anything bad by it just the bug with the two lines lol
2020-09-27T03:50:27 #kisslinux <muevoid> Yes
2020-09-27T03:51:55 #kisslinux <mcpcpc[m]> no worries. i will look into it.
2020-09-27T03:52:19 #kisslinux <muevoid> kk
2020-09-27T03:56:50 #kisslinux <dilyn> gngn o/
2020-09-27T03:56:59 #kisslinux <muevoid> o/
2020-09-27T04:01:04 #kisslinux <muevoid> periish since I know you are using non standard filesystem structure did you have any problems with g
2020-09-27T04:01:18 #kisslinux <muevoid> gcc*
2020-09-27T04:02:58 #kisslinux <mcpcpc[m]> muevoid: feel free to add any other context. https://github.com/mcpcpc/kirc/issues/39
2020-09-27T04:03:42 #kisslinux <muevoid> Will do alater working on getting this working
2020-09-27T04:38:43 #kisslinux <muevoid> I figured out the issue but not sure how to fix it for some reason gcc isn't seeing the libstdc++ headers
2020-09-27T04:56:39 #kisslinux <muevoid> Bleck now more issues :)
2020-09-27T05:36:19 #kisslinux <muevoid> Hey dylanaraps o/
2020-09-27T05:36:29 #kisslinux <muevoid> my gcc build is segfaulting now :<
2020-09-27T05:58:07 #kisslinux <dylanaraps> Why?
2020-09-27T05:58:23 #kisslinux <dylanaraps> Gives you no other information?
2020-09-27T05:58:33 #kisslinux <dylanaraps> You could do a debug build
2020-09-27T06:00:39 #kisslinux <muevoid> Seg faults during Makefile:2089: specs
2020-09-27T06:00:46 #kisslinux <muevoid> How would I do a debug build?
2020-09-27T06:01:06 #kisslinux <muevoid> if you want to see the build it's in github.com/muevoid/mue-linux in core
2020-09-27T06:01:16 #kisslinux <muevoid> It's basically the same as kiss without usr
2020-09-27T06:05:36 #kisslinux <dylanaraps> Right
2020-09-27T06:05:54 #kisslinux <dylanaraps> What does Makefile line 2089 contain?
2020-09-27T06:07:00 #kisslinux <dylanaraps> Oh yeah. New feature for post-install/pre-remove package hooks. If the file is /not/ executable, the package manager will simply display its contents to the screen. Allows packages only needing to print messages to avoid executing code for this purpose.
2020-09-27T06:07:39 #kisslinux <muevoid> Not sure give me one min
2020-09-27T06:16:22 #kisslinux <muevoid> I have no clue where it is failing. Somewhere in the cpp build I think it may be due to other parts of my core files but I have no clue where.
2020-09-27T06:45:31 #kisslinux <muevoid> It is not the other core builds cause now it builds with the kiss build...
2020-09-27T06:46:40 #kisslinux <dylanaraps> diff both build files
2020-09-27T06:49:14 #kisslinux <muevoid> I'm just gonna slowly start changing the kiss build till it either fails again or works
2020-09-27T06:50:41 #kisslinux <dylanaraps> Try just changing --prefix to /
2020-09-27T06:51:07 #kisslinux <muevoid> That's the first change I am making
2020-09-27T06:54:24 #kisslinux <dylanaraps> be on later
2020-09-27T06:54:27 #kisslinux <dylanaraps> back to work
2020-09-27T06:54:28 #kisslinux <dylanaraps> o/
2020-09-27T06:54:30 #kisslinux <muevoid> kk o/
2020-09-27T06:58:14 #kisslinux <muevoid> It is indeed the prefix being set to / that causes the segfault...
2020-09-27T07:00:01 #kisslinux <muevoid> I'm trying to get rid of the --prefix option and instead when doing make install do PREFIX=/
2020-09-27T07:55:27 #kisslinux <muevoid> I got it working by setting prefix=""
2020-09-27T08:44:31 #kisslinux <muevoid> Hey merakor! How are you doing?
2020-09-27T11:26:27 #kisslinux <dylanaraps> o/
2020-09-27T11:27:23 #kisslinux <icyphox> Hi
2020-09-27T11:27:44 #kisslinux <icyphox> Man, my wpm has tanked hard.
2020-09-27T11:28:00 #kisslinux <dylanaraps> I type like an old person
2020-09-27T11:28:06 #kisslinux <dylanaraps> I wouldn't worry about it
2020-09-27T11:28:26 #kisslinux <dylanaraps> :P
2020-09-27T11:29:41 #kisslinux <icyphox> Nah, I've switched to a different layout.
2020-09-27T11:30:00 #kisslinux <icyphox> I've been using Workman for the past 10 days.
2020-09-27T11:30:42 #kisslinux <dylanaraps> Ah
2020-09-27T11:31:06 #kisslinux <icyphox> I used to be about 100+ wpm, it's now at 50 ish.
2020-09-27T11:31:56 #kisslinux <dylanaraps> That's my normal wpm from when I last did one of those tests
2020-09-27T11:31:57 #kisslinux <dylanaraps> ;)
2020-09-27T11:32:11 #kisslinux <icyphox> Woah
2020-09-27T11:32:32 #kisslinux <icyphox> I actually expected you to be some god tier 160+ fella.
2020-09-27T11:32:38 #kisslinux <dylanaraps> I type with two fingers + thumb on spacebar usually lol
2020-09-27T11:32:56 #kisslinux <icyphox> No way ahaha.
2020-09-27T11:33:46 #kisslinux <dylanaraps> Yeah
2020-09-27T11:33:57 #kisslinux <dylanaraps> Dunno
2020-09-27T11:34:04 #kisslinux <dylanaraps> Never been a hindrance to my work or anything
2020-09-27T11:36:01 #kisslinux <icyphox> I suppose not. Nobody writes code that fast anyway.
2020-09-27T11:36:16 #kisslinux <icyphox> wpm is all just e-dick waving
2020-09-27T11:36:31 #kisslinux <icyphox> I mostly switched for ergonomics
2020-09-27T11:37:00 #kisslinux <icyphox> High frequency letters on the home row.
2020-09-27T11:37:09 #kisslinux <icyphox> Feels so much nicer to type on, heh.
2020-09-27T11:37:57 #kisslinux <bsl> is it better than dvorak?
2020-09-27T11:39:42 #kisslinux <bsl> hi btw...i've been lurking here hoping to find someone to help me figure out why ath10k module is not finding the firmware even though file is there
2020-09-27T11:40:11 #kisslinux <bsl> tried to use strace, but is not showing that the module tries to open the file it complains is not finding
2020-09-27T11:40:13 #kisslinux <bsl> any ideas?
2020-09-27T11:41:46 #kisslinux <sh4rm4^bnc> bsl: strace
2020-09-27T11:41:59 #kisslinux <sh4rm4^bnc> strace -f modprobe ...
2020-09-27T11:42:07 #kisslinux <sh4rm4^bnc> it will show you which paths it opens
2020-09-27T11:42:22 #kisslinux <sh4rm4^bnc> look for open and ENOENT
2020-09-27T11:42:33 #kisslinux <bsl> sh4rm4^bnc, hey man o/
2020-09-27T11:42:50 #kisslinux <bsl> let me paste strace output somewhere
2020-09-27T11:46:20 #kisslinux <sh4rm4^bnc> bsl: https://github.com/sabotage-linux/sabotage/blob/master/KEEP/bin/pastebin
2020-09-27T11:46:43 #kisslinux <sh4rm4^bnc> strace -f modprobe ... 2>&1 | pastebin
2020-09-27T11:46:49 #kisslinux <bsl> https://dpaste.com/7T94AEPAL
2020-09-27T11:47:10 #kisslinux <bsl> wgetpaste ships with gentoo livecd :)
2020-09-27T11:48:19 #kisslinux <sh4rm4^bnc> oh, cool. +++ exited with 0 +++ -> seems to work ?
2020-09-27T11:49:33 #kisslinux <sh4rm4^bnc> it seems the kernel looks for the firmware itself, so strace won't help, though dmesg could
2020-09-27T11:50:49 #kisslinux <bsl> sh4rm4^bnc, modprobe stdouts: failed to find firmware-N.bin (N between 2 and 6) from ath10k/QCA9377/hw1.0: -2
2020-09-27T11:51:06 #kisslinux <bsl> dmesg | grep ath shows nothing since it's a module
2020-09-27T11:54:59 #kisslinux <sh4rm4^bnc> odd, there is nothing in your trace showing that it's writing an error message
2020-09-27T11:55:10 #kisslinux <bsl> sh4rm4^bnc, kernel boots directly from UEFI, there is no System.map file, could that be a problem?
2020-09-27T11:55:23 #kisslinux <bsl> although it didn't have any problem with gentoo, it booted the same way
2020-09-27T11:55:26 #kisslinux <sh4rm4^bnc> System.map should only be needed for debug symbols
2020-09-27T11:56:11 #kisslinux <sh4rm4^bnc> btw there's a kernel build option to bake the firmware into the kernel
2020-09-27T11:56:41 #kisslinux <sh4rm4^bnc> might be preferable if it has issues accesing the path to /lib/firmware
2020-09-27T11:57:21 #kisslinux <icyphox> bsl: Well, the site claims it's an improvement over Dvorak and Colemak.
2020-09-27T11:57:32 #kisslinux <icyphox> There are some comparisons too.
2020-09-27T11:57:51 #kisslinux <icyphox> Dvorak is a litte heavier on the right-hand, AFIAK.
2020-09-27T11:58:45 #kisslinux <bsl> sh4rm4^bnc, i'd like to find the culprit tbh, why is doing this...
2020-09-27T11:59:48 #kisslinux <sh4rm4^bnc> bsl: i don't understand how you can tell me that modprobe prints "failed to find" yet this can't be seen in its trace
2020-09-27T11:59:53 #kisslinux <sh4rm4^bnc> was that from another run?
2020-09-27T12:00:17 #kisslinux <bsl> icyphox, yes, now I remember that workman is last iteration of better kbd layout -- saw it in HN some months ago :)
2020-09-27T12:02:26 #kisslinux <bsl> sh4rm4^bnc, https://i.imgur.com/ntVZry7.png
2020-09-27T12:02:53 #kisslinux <icyphox> Yeah. There have been some improvements /on/ Workman too -- there's the Norman layout.
2020-09-27T12:03:50 #kisslinux <bsl> icyphox, very similar to qwerty :D
2020-09-27T12:04:37 #kisslinux <sh4rm4^bnc> bsl, i don't have these files in my ath10k dir
2020-09-27T12:05:39 #kisslinux <bsl> what do you have?
2020-09-27T12:08:30 #kisslinux <sh4rm4^bnc> http://ix.io/2yUY
2020-09-27T12:09:15 #kisslinux <sh4rm4^bnc> actually i didn't notice hw1.0 was a dir
2020-09-27T12:10:05 #kisslinux <bsl> looks same to me, only you have links to firmware dir
2020-09-27T12:11:27 #kisslinux <sh4rm4^bnc> so there's only firmware5
2020-09-27T12:13:35 #kisslinux <bsl> ok, deleted firmware-6 modprobe -r * ...but same thing happens
2020-09-27T12:17:50 #kisslinux <sh4rm4^bnc> "In case the driver is built into the kernel (<*>) instead of as a module (<M>), then the firmware needs to be built into the kernel as well. "
2020-09-27T12:19:23 #kisslinux <bsl> which driver? ath10k? it's <M> atm
2020-09-27T12:20:20 #kisslinux <sh4rm4^bnc> seems latest drivers have a different structure https://github.com/kvalo/ath10k-firmware/tree/master/QCA9377/hw1.0/WLAN.TF.2.1
2020-09-27T12:22:21 #kisslinux <bsl> ok, let me try that
2020-09-27T12:25:16 #kisslinux <bsl> sh4rm4^bnc, same :(
2020-09-27T12:25:52 #kisslinux <bsl> the error msg is "Failed to find firmware-N.bin (N between 2 and 6) from ath10k/...
2020-09-27T12:26:11 #kisslinux <bsl> so it must be looking for firmware-[2..6]
2020-09-27T12:26:13 #kisslinux <sh4rm4^bnc> you previously already had firmware-6.bin ? i didn't
2020-09-27T12:26:26 #kisslinux <sh4rm4^bnc> in my version there's only 5
2020-09-27T12:26:28 #kisslinux <bsl> yes, it came with linux-firmware.tar.gz from kernel.org
2020-09-27T12:27:12 #kisslinux <sh4rm4^bnc> i guess it's time to read what the code in the driver actually does
2020-09-27T12:27:55 #kisslinux <sh4rm4^bnc> or maybe you just use a more recent kernel, e.g. 5.4 LTS
2020-09-27T12:28:07 #kisslinux <sh4rm4^bnc> according to your paste i assume you use 5.0
2020-09-27T12:28:29 #kisslinux <bsl> 5.8.11
2020-09-27T12:29:08 #kisslinux <bsl> i tried looking at ath10k_pci.mod.c but could not find any open()
2020-09-27T12:29:24 #kisslinux <sh4rm4^bnc> i guess you could search for the firmware-N error message
2020-09-27T12:29:31 #kisslinux <sh4rm4^bnc> and add some kprintf's
2020-09-27T12:33:07 #kisslinux <bsl> ok, found something in core.c ar->hw_params.fw.dir -- let's see what it is holding, brb
2020-09-27T12:33:23 #kisslinux <soliwilos> bsl: https://forums.gentoo.org/viewtopic-t-1088792-start-0.html
2020-09-27T12:34:12 #kisslinux <soliwilos> If you scroll down to the last couple posts there, that person made it work by removing some extra firmware files.
2020-09-27T12:35:16 #kisslinux <bsl> soliwilos, linking hw1.0 to hw 1.1 didn't help
2020-09-27T12:35:41 #kisslinux <bsl> i don't know how, but on gentoo live cd, which has the same structure, works without any problems
2020-09-27T12:38:45 #kisslinux <soliwilos> bsl: Not referring to the symlinking mentioned there, but making sure to only have firmware-5.bin, notice_ath10k_firmware.txt, board-2.bin and board.bin files in the hw1.0 directory.
2020-09-27T12:40:30 #kisslinux <soliwilos> s/notice_ath10k_firmware.txt/notice_ath10k_firmware-5.txt/
2020-09-27T12:40:52 #kisslinux <bsl> soliwilos, tried that, few moments ago, didn't help
2020-09-27T12:41:35 #kisslinux <soliwilos> Tried an older firmware release?
2020-09-27T12:42:30 #kisslinux <soliwilos> Guessing the firmware files on the Gentoo livecd you're using are older than the ones you've downloaded?
2020-09-27T12:43:32 #kisslinux <bsl> soliwilos, I tried to copy them from livecd, again, same thing
2020-09-27T12:43:50 #kisslinux <bsl> looks like problem is in the kernel module, not firmware
2020-09-27T12:44:37 #kisslinux <soliwilos> 5.8.12 is out, could try the newer kernel.
2020-09-27T12:53:42 #kisslinux <muevoid> What does kiss-manifest show for ncurses on KISS?
2020-09-27T12:54:40 #kisslinux <muevoid> Because something with my build makes it where things can't find /include/ncurses.h
2020-09-27T12:54:55 #kisslinux <muevoid> It is installed just under /include/ncursesw/ncurses.h
2020-09-27T12:57:49 #kisslinux <bsl> muevoid, /usr/include/ncurses.h
2020-09-27T12:59:05 #kisslinux <muevoid> Yeah I know that is where it is usually at but for some reason when setting prefix to / it doesn't show up in /include/ncurses.h only /include/ncursesw/ncurses.h. I will try to figure this out
2020-09-27T13:01:32 #kisslinux <muevoid> Does it have a subdirectory for /ncursesw on normal KISS?
2020-09-27T13:01:46 #kisslinux <sh4rm4^bnc> ncurses has a ton of configure options. are you sure you didn't change any?
2020-09-27T13:02:02 #kisslinux <muevoid> All I changed was the prefix
2020-09-27T15:57:30 #kisslinux <dilyn> somebody wrote a patch for adding innetgr to musl's include/netdb.h that merely returns null
2020-09-27T15:57:35 #kisslinux <dilyn> eight years ago xD
2020-09-27T15:57:46 #kisslinux <dilyn> so hopefully, it works, and I can finally have js-less polkit
2020-09-27T16:03:15 #kisslinux <dilyn> dylanaraps: if you're interested https://www.openwall.com/lists/musl/2012/05/16/2/2
2020-09-27T16:03:22 #kisslinux <dilyn> otherwise I can just carry it around
2020-09-27T17:02:48 #kisslinux <E5ten> dylanaraps: shouldn't the KISS dir variables aside from prefix not have a leading /?
2020-09-27T17:21:03 #kisslinux <dylanaraps> E5ten: This is more to mimic how they're set by autotools/etc
2020-09-27T17:21:07 #kisslinux <dylanaraps> It doesn't really matter
2020-09-27T17:21:12 #kisslinux <dylanaraps> Works all the same
2020-09-27T17:21:33 #kisslinux <dylanaraps> KISS_PREFIX used in ./configure must have leading / no?
2020-09-27T17:21:53 #kisslinux <dylanaraps> Or would it always be written as --prefix="/$KISS_PREFIX" ?
2020-09-27T17:22:05 #kisslinux <dylanaraps> Rather than --prefix="$KISS_PREFIX"
2020-09-27T17:22:29 #kisslinux <dylanaraps> I could also force a leading slash if we need to require one...
2020-09-27T17:22:43 #kisslinux <dylanaraps> This would make both 'usr' and '/usr' valid values
2020-09-27T17:23:08 #kisslinux <dylanaraps> I'm not too keen on this feature yet
2020-09-27T17:23:26 #kisslinux <dylanaraps> I need to spend some time thinking about it
2020-09-27T17:23:39 #kisslinux <dylanaraps> Just pushed this to get the ball rolling
2020-09-27T17:23:52 #kisslinux <dylanaraps> Issue is: https://github.com/kisslinux/kiss/issues/191
2020-09-27T17:24:14 #kisslinux <sh4rm4^bnc> if prefix is just / then you would run ./configure --prefix=
2020-09-27T17:24:17 #kisslinux <sh4rm4^bnc> (i.e. empty)
2020-09-27T17:24:41 #kisslinux <sh4rm4^bnc> if it is /usr, then --prefix=/usr
2020-09-27T17:24:44 #kisslinux <E5ten> dylanaraps: No I mean KISS_PREFIX would keep the leading /, the others wouldn't
2020-09-27T17:25:21 #kisslinux <dylanaraps> If you made the value '/' it'd just do //path and it'd work fine no? sh4rm4^bnc:
2020-09-27T17:25:25 #kisslinux <E5ten> Because the way you have it like $KISS_PREFIX/$KISS_LIBDIR it expands to /use//lib
2020-09-27T17:25:53 #kisslinux <sh4rm4^bnc> dylanaraps, it would work, but // has a special meaning in POSIX
2020-09-27T17:26:14 #kisslinux <E5ten> Imo it'd be preferable for the defaults to make a coherent path?
2020-09-27T17:26:34 #kisslinux <dylanaraps> E5ten: Yeah
2020-09-27T17:27:12 #kisslinux <sh4rm4^bnc> https://www.openwall.com/lists/musl/2020/09/08/5
2020-09-27T17:27:19 #kisslinux <sh4rm4^bnc> > - Initial // is treated as special, but //. and //.. resolve to /
2020-09-27T17:28:34 #kisslinux <dylanaraps> Yeah
2020-09-27T17:29:16 #kisslinux <sh4rm4^bnc> however i dont know what "special" means here, but iirc dalias once said something about NFS
2020-09-27T17:30:45 #kisslinux <dylanaraps> Was going to ask
2020-09-27T17:30:55 #kisslinux <dylanaraps> lol
2020-09-27T17:32:21 #kisslinux <dylanaraps> E5ten: ${KISS_PREFIX:-/usr} should probably just be ${KISS_PREFIX-/usr} as we want to allow to be empty for '/' right?
2020-09-27T17:48:21 #kisslinux <E5ten> That makes sense
2020-09-27T17:48:43 #kisslinux <E5ten> You also probably wanna do the trailing / strip from all of these variables
2020-09-27T17:49:22 #kisslinux <E5ten> So if someone does KISS_PREFIX=/ or /usr/ or something the path won't have duplicated slashes
2020-09-27T17:50:02 #kisslinux <E5ten> Especially for the KISS_PREFIX=/ to avoid the leading double slash
2020-09-27T17:56:39 #kisslinux <dylanaraps> Yup
2020-09-27T18:07:40 #kisslinux <dylanaraps> Done
2020-09-27T18:08:12 #kisslinux <E5ten> nice
2020-09-27T18:08:53 #kisslinux <dylanaraps> What's a six letter equivalent to DATADIR? lol
2020-09-27T18:09:04 #kisslinux <dylanaraps> Everything else is six long
2020-09-27T18:11:04 #kisslinux <E5ten> I don't think it exists lol
2020-09-27T18:11:41 #kisslinux <E5ten> anything else would be confusing cuz that's what everything uses (well also datarootdir but idk the difference really)
2020-09-27T18:12:08 #kisslinux <dylanaraps> datarootdir controls prefix for share, doc, info, etc
2020-09-27T18:12:32 #kisslinux <dylanaraps> ignore the share...
2020-09-27T18:12:44 #kisslinux <dylanaraps> datadir controls it independently?
2020-09-27T18:12:52 #kisslinux <dylanaraps> I just read this about an hour ago hang on
2020-09-27T18:13:21 #kisslinux <dylanaraps> > This is usually the same place as ‘datarootdir’, but we use the two separate variables so that you can move these program-specific files without altering the location for Info files, man pages, etc.
2020-09-27T18:13:25 #kisslinux <dylanaraps> (for datadir)
2020-09-27T18:13:41 #kisslinux <E5ten> I see
2020-09-27T18:13:47 #kisslinux <E5ten> so maybe kiss should be using DATAROOTDIR then?
2020-09-27T18:14:38 #kisslinux <dylanaraps> I need to hear from the various people with kiss forks using different hierarchies
2020-09-27T18:14:59 #kisslinux <E5ten> makes sense
2020-09-27T18:15:14 #kisslinux <dylanaraps> We may even end up setting both in the end
2020-09-27T18:15:25 #kisslinux <E5ten> fair enough
2020-09-27T18:15:31 #kisslinux <dylanaraps> INCDIR should maybe be INCLUDEDIR
2020-09-27T18:15:38 #kisslinux <dylanaraps> If we want these to map 1:1 that is
2020-09-27T18:15:38 #kisslinux <E5ten> yeah
2020-09-27T18:15:49 #kisslinux <dylanaraps> But 1:1 with autotools? or??
2020-09-27T18:15:50 #kisslinux <E5ten> I think that'd probably be preferable
2020-09-27T18:15:55 #kisslinux <E5ten> cmake also uses includedir
2020-09-27T18:16:05 #kisslinux <E5ten> same for meson
2020-09-27T18:16:10 #kisslinux <dylanaraps> Neato
2020-09-27T18:16:50 #kisslinux <E5ten> cmake has datarootdir and datadir, meson only has the latter
2020-09-27T18:17:26 #kisslinux <dylanaraps> datadir defaults to 'share' with meson
2020-09-27T18:17:44 #kisslinux <E5ten> datadir defaults to datarootdir in cmake, datarootdir defaults to share
2020-09-27T18:18:01 #kisslinux <dylanaraps> mandir, etc are set to 'share/man', ...
2020-09-27T18:18:31 #kisslinux <dylanaraps> so it looks like it isn't $datadir/$mandir with meson?
2020-09-27T18:18:43 #kisslinux <dylanaraps> ie, there's no datarootdir
2020-09-27T18:19:03 #kisslinux <dylanaraps> https://github.com/mesonbuild/meson/issues/7616
2020-09-27T18:19:03 #kisslinux <E5ten> I think all of them are usually set to relative to prefix except for the special ones
2020-09-27T18:19:25 #kisslinux <E5ten> like in cmake mandir defaults to datarootdir/man
2020-09-27T18:19:33 #kisslinux <dylanaraps> yeah
2020-09-27T18:20:03 #kisslinux <E5ten> because the whole thing with mandir is that it's usually either prefix/man or prefix/share/man
2020-09-27T18:20:32 #kisslinux <E5ten> so if the build system didn't make it relative to prefix you'd have the person using the build system having to figure out which place mandir should go
2020-09-27T18:20:40 #kisslinux <E5ten> which kinda defeats the point of mandir lol
2020-09-27T18:20:42 #kisslinux <dylanaraps> Yup
2020-09-27T18:21:00 #kisslinux <dylanaraps> So many build files will need updating
2020-09-27T18:21:09 #kisslinux <dylanaraps> I'm sure we'll be sending bugs upstream too
2020-09-27T18:22:00 #kisslinux <dylanaraps> Ever changed your system hierarchy on-the-fly before? I wonder how feasible it is.
2020-09-27T18:22:10 #kisslinux <E5ten> I haven't lol
2020-09-27T18:23:32 #kisslinux <merakor> That sounds like a recipe for disaster
2020-09-27T18:23:43 #kisslinux <dylanaraps> Oh yeah
2020-09-27T18:24:27 #kisslinux <dylanaraps> Could probably do it via CFLAGS/CXXFLAGS + -L, -I, PATH additions, etc
2020-09-27T18:25:03 #kisslinux <dylanaraps> Then you'd rebuild everything
2020-09-27T18:25:10 #kisslinux <dylanaraps> And pray
2020-09-27T18:25:25 #kisslinux <merakor> > And pray
2020-09-27T18:25:28 #kisslinux <merakor> True lol
2020-09-27T18:27:19 #kisslinux <merakor> Honestly, linking everything to /usr is the most practical hierarchy solution
2020-09-27T18:27:34 #kisslinux <dylanaraps> Yes
2020-09-27T18:27:58 #kisslinux <dylanaraps> Path of least resistance
2020-09-27T18:28:07 #kisslinux <merakor> Can't go wrong with /usr and /usr/local
2020-09-27T18:30:17 #kisslinux <dylanaraps> Another solution is to shuffle the directories into the set hierarchy post-build and just forget the build system mess
2020-09-27T18:30:25 #kisslinux <dylanaraps> Unsure of how doable this is
2020-09-27T18:31:42 #kisslinux <dylanaraps> But you'd break pkgconf and many other things
2020-09-27T18:31:45 #kisslinux <dylanaraps> Forget it
2020-09-27T18:31:46 #kisslinux <dylanaraps> :P
2020-09-27T18:32:06 #kisslinux <E5ten> lol yeah does not seem doable
2020-09-27T18:34:15 #kisslinux <dylanaraps> We may have to create wrappers for the build systems after all
2020-09-27T18:35:00 #kisslinux <dylanaraps> Having everyone set 10 (or whatever the number is) configure flags for each package will become tiresome
2020-09-27T18:36:12 #kisslinux <dylanaraps> But then we add more reliance on shell for build scripts (unless we want to force this restriction)
2020-09-27T18:37:27 #kisslinux <dylanaraps> Question ultimately is whether we want to open this can of worms or just stick with /usr (path of least resistance)
2020-09-27T18:42:17 #kisslinux <E5ten> I kinda feel like this is almost impossible to completely achieve, and there will be issues that come up, imo sticking with /usr is better
2020-09-27T18:43:09 #kisslinux <dylanaraps> This is why it was chosen to begin with :P
2020-09-27T18:43:30 #kisslinux <dylanaraps> Thought I'd at least attempt the feature
2020-09-27T18:43:35 #kisslinux <sh4rm4^bnc> that's why sabotage symlinks /usr to /
2020-09-27T18:43:57 #kisslinux <sh4rm4^bnc> there's too much hardcoded /usr everywhere
2020-09-27T18:44:54 #kisslinux <sh4rm4^bnc> the most prominent being /usr/bin/env
2020-09-27T18:45:39 #kisslinux <dylanaraps> Yup
2020-09-27T18:45:42 #kisslinux <E5ten> shebangs have to have hardcoding of some kind though? what would be preferable?
2020-09-27T18:46:02 #kisslinux <dylanaraps> Example: https://termbin.com/a22s
2020-09-27T18:46:02 #kisslinux <mcf> i've had pretty good look with just a /usr/bin -> /bin symlink
2020-09-27T18:46:23 #kisslinux <mcf> luck*
2020-09-27T18:46:27 #kisslinux <dylanaraps> (package manager would ensure these are set)
2020-09-27T18:47:13 #kisslinux <sh4rm4^bnc> what is oldincludedir ?
2020-09-27T18:47:51 #kisslinux <dylanaraps> Not a meson flag that's for sure
2020-09-27T18:48:12 #kisslinux <dylanaraps> For headers not intended for the default compiler
2020-09-27T18:48:26 #kisslinux <dylanaraps> Usually set to /usr/include nowadays
2020-09-27T18:48:43 #kisslinux <dylanaraps> https://www.gnu.org/prep/standards/html_node/Directory-Variables.html
2020-09-27T18:48:47 #kisslinux <dylanaraps> search for it here
2020-09-27T18:49:46 #kisslinux <E5ten> doesn't seem like a var worth supporting even if you do decide to stick with this and don't revert back to /usr
2020-09-27T18:50:36 #kisslinux <dylanaraps> Yeah
2020-09-27T18:50:43 #kisslinux <dylanaraps> Doubt it's used at all
2020-09-27T18:50:50 #kisslinux <dylanaraps> Can just set it to INCLUDEDIR
2020-09-27T18:51:16 #kisslinux <sh4rm4^bnc> sounds really crufty
2020-09-27T18:51:41 #kisslinux <sh4rm4^bnc> also gcc looking by default in /usr/local seems like a bad idea
2020-09-27T18:52:15 #kisslinux <sh4rm4^bnc> one would typically add it to -L LDFLAGS if so desired
2020-09-27T19:01:12 #kisslinux <dylanaraps> https://github.com/kisslinux/kiss/blob/master/kiss#L1515-L1526
2020-09-27T19:01:20 #kisslinux <dylanaraps> Does anything use SHAREDSTATEDIR?
2020-09-27T19:01:38 #kisslinux <dylanaraps> I don't have this directory
2020-09-27T19:01:54 #kisslinux <dylanaraps> /com ?
2020-09-27T19:02:21 #kisslinux <dylanaraps> Fulton: We're discussion your feature request
2020-09-27T19:02:26 #kisslinux <dylanaraps> discussing*
2020-09-27T19:03:04 #kisslinux <dylanaraps> Let me know if you have anything to add
2020-09-27T19:04:05 #kisslinux <dylanaraps> How much patching (to work around build systems) has been required in your fork thus far?
2020-09-27T19:21:06 #kisslinux <Fulton> dylanaraps: generally just removing /usr from things. My main concern is compatibility with third party repos.
2020-09-27T19:22:56 #kisslinux <dylanaraps> Yes. This will break packages not using wrappers. I could do some trickery with PATH from package manager side so that (for example) 'meson' points to our wrapper which executes real meson with flags.
2020-09-27T19:23:04 #kisslinux <dylanaraps> not using the envars*
2020-09-27T19:23:31 #kisslinux <dylanaraps> (assuming we also use wrappers for build systems)
2020-09-27T19:24:31 #kisslinux <dylanaraps> I don't think we'll go ahead with this...
2020-09-27T20:19:21 #kisslinux <merakor> Something like kiss-meson, kiss-configure could work
2020-09-27T20:19:34 #kisslinux <merakor> But idk
2020-09-27T20:20:52 #kisslinux <merakor> Like a wrapper script which is basically `./configure --prefix=KISS_PREFIX ... "$@"`
2020-09-27T22:04:31 #kisslinux <miskatonic> the site says that the installation of kiss requires an arbitrary linux live medium. What can a liuve mediun achieve that is not possiblr to achieve from an already running linux system?
2020-09-27T22:10:21 #kisslinux <sh4rm4^bnc> a umounted HDD, i suppose
2020-09-27T22:32:04 #kisslinux <Fulton> miskatonic: I would recommend using Gentoo's live usb