💾 Archived View for gemini.ctrl-c.club › ~phoebos › logs › freenode-kisslinux-2020-10-05.txt captured on 2022-07-17 at 03:18:24.
⬅️ Previous capture (2021-12-17)
-=-=-=-=-=-=-
2020-10-05T00:50:34 #kisslinux <micr0> im still confused about gobject-introspection 2020-10-05T00:51:13 #kisslinux <micr0> building fwupd errors out with a python script that checks for two dependencies: "missing pangocairo gobject introspection library" and "missing pango gobject introspection library" 2020-10-05T00:51:25 #kisslinux <micr0> is there a specific order I need to install pango and gobject-introspection 2020-10-05T00:53:56 #kisslinux <mcf> did you build pango with -Dintrospection=true ? maybe it autodetects false if you don't have gobject-introspection 2020-10-05T00:59:11 #kisslinux <micr0> welp that froze sway completely and my laptop hard-locked 2020-10-05T00:59:20 #kisslinux <micr0> yeah forking pango and setting introspection=true did it 2020-10-05T01:00:31 #kisslinux <micr0> now to find out gnu-efi headers TT 2020-10-05T04:24:47 #kisslinux <dylanaraps> dilynm: How big is your kernel package? How many files? 2020-10-05T04:24:54 #kisslinux <dylanaraps> Are you using gnugrep as /usr/bin/grep? 2020-10-05T05:21:07 #kisslinux <dilynm> dylanaraps: there seems to be a bug for alternatives. The to-be-replaced file isn't moved and the manifest isn't updated 2020-10-05T05:24:54 #kisslinux <dylanaraps> Thanks 2020-10-05T05:25:01 #kisslinux <dylanaraps> Will see if I can reproduce. 2020-10-05T05:25:29 #kisslinux <dilynm> If you can't I will be supremely worried (: 2020-10-05T05:25:36 #kisslinux <dylanaraps> Works here (: 2020-10-05T05:26:14 #kisslinux <dylanaraps> Any error message? 2020-10-05T05:26:41 #kisslinux <dylanaraps> Also, are we dealing with symlinks in your case? 2020-10-05T05:27:02 #kisslinux <dilynm> No I'm swapping netbsd-curses tic for ncurses tic 2020-10-05T05:27:27 #kisslinux <dylanaraps> Let me try the same packages 2020-10-05T05:27:30 #kisslinux <dilynm> No errors but it prints that "(was owned by )" 2020-10-05T05:27:41 #kisslinux <dilynm> When it swaps 2020-10-05T05:28:07 #kisslinux <dilynm> Rather; "/usr/bin/tic is now ncurses (was )" 2020-10-05T05:28:36 #kisslinux <dilynm> Oh. Hm. Okay, a swap doesn't happen and tic is still owned by ncurses 2020-10-05T05:28:45 #kisslinux <dilynm> Netbsd-curses*** 2020-10-05T05:28:54 #kisslinux <dylanaraps> Will fix 2020-10-05T05:28:58 #kisslinux <dilynm> So it fails at the cp part 2020-10-05T05:29:03 #kisslinux <dylanaraps> I'll also make it rollback any changes it makes 2020-10-05T05:29:13 #kisslinux <dylanaraps> (if error) 2020-10-05T05:29:54 #kisslinux <dilynm> Scratch that last part - the manifest isn't updated. The cp succeeds, the manifest update fails 2020-10-05T05:30:10 #kisslinux <dylanaraps> Thanks 2020-10-05T05:31:23 #kisslinux <dylanaraps> I can reproduce with these two packages 2020-10-05T05:31:30 #kisslinux <dylanaraps> Or rather... something broke over here too 2020-10-05T05:34:12 #kisslinux <dilynm> Yeah it's fine with busybox 2020-10-05T05:34:30 #kisslinux <dilynm> Although there is a separate issue if a user has acpid installed and swapped it seems 2020-10-05T05:35:14 #kisslinux <dilynm> Probably doesn't happen if I unlink the service tho huh xD 2020-10-05T05:35:35 #kisslinux <dylanaraps> Working on it 2020-10-05T05:37:10 #kisslinux <dilynm> (a similar issue to netbsd-curses/ncurses with openntpd/busybox) 2020-10-05T05:40:19 #kisslinux <dylanaraps> Thanks 2020-10-05T06:12:22 #kisslinux <dilynm> Yeah in fact even just attempting to install sbase or ubase will... Break the system 2020-10-05T06:16:01 #kisslinux <dylanaraps> dilynm: Should be fixed now. 2020-10-05T06:16:04 #kisslinux <dylanaraps> Just pushed a commit. 2020-10-05T06:16:20 #kisslinux <dylanaraps> Oh 2020-10-05T06:16:24 #kisslinux <dylanaraps> I see what you mean 2020-10-05T06:16:29 #kisslinux <dilynm> Lol 2020-10-05T06:16:30 #kisslinux <dylanaraps> Just broke mine :P 2020-10-05T06:16:36 #kisslinux <dylanaraps> > /bin/sh: rm: not found 2020-10-05T06:17:07 #kisslinux <dilynm> Mmhmm 2020-10-05T06:17:09 #kisslinux <dilynm> Lol 2020-10-05T06:17:37 #kisslinux <dylanaraps> This is why I made a pre-release ;) 2020-10-05T06:17:42 #kisslinux <dilynm> And so I fixed it, but now I'm having efivars issues so I have to find my usb with arch to fix it 2020-10-05T06:17:49 #kisslinux <dilynm> Yes good call haha. Glad I'm testing xD 2020-10-05T06:18:21 #kisslinux <dylanaraps> My kiss-static idea is looking more intriguing now 2020-10-05T06:18:32 #kisslinux <dilynm> I think these double ipas are causing me to suffer temporary blindness and I can't find it xD bed time 2020-10-05T06:18:40 #kisslinux <dilynm> Static kiss: the next way to bork my system 2020-10-05T06:19:34 #kisslinux <dylanaraps> basically custom busybox compiled with only what the package manager needs. Package manager is then embedded in the busybox binary and the busybox binary is renamed to 'kiss'. 2020-10-05T06:19:46 #kisslinux <dylanaraps> In the busybox config, it is set to prefer internal applets to external ones always 2020-10-05T06:19:57 #kisslinux <dylanaraps> Comes in at 300-400kb 2020-10-05T06:24:04 #kisslinux <dylanaraps> AFixed. 2020-10-05T06:24:07 #kisslinux <dylanaraps> Fixed* 2020-10-05T06:25:54 #kisslinux <dylanaraps> I'm creating a test suite of sorts now. 2020-10-05T06:50:28 #kisslinux <dilynm> Fun fun! I'll test the latest kiss master in the morning; I'm gonna have to figure out this efivars issue -.- 2020-10-05T06:50:35 #kisslinux <dilynm> Gngn o/ 2020-10-05T07:02:27 #kisslinux <muevoid> o/ 2020-10-05T07:03:52 #kisslinux <lpy> o/ 2020-10-05T07:05:40 #kisslinux <muevoid> How goes it? 2020-10-05T07:06:04 #kisslinux <lpy> good, you? 2020-10-05T07:06:41 #kisslinux <muevoid> Pretty good, working on a game framework. 2020-10-05T07:06:59 #kisslinux <lpy> o 2020-10-05T07:07:01 #kisslinux <lpy> nice 2020-10-05T12:24:04 #kisslinux <Rio6> how do you all update your kernel source. git? patch files? 2020-10-05T12:24:57 #kisslinux <dylanaraps> I do it manually 2020-10-05T12:25:29 #kisslinux <dylanaraps> I copy my .config over and run `make; make install 2020-10-05T12:26:09 #kisslinux <Rio6> do you re download the source tarball every time 2020-10-05T12:27:22 #kisslinux <Rio6> i was using the patch files from kernel.org then messed up the versions 2020-10-05T12:27:40 #kisslinux <dylanaraps> Yes 2020-10-05T12:27:58 #kisslinux <dylanaraps> I keep full sources for each version I choose to use. 2020-10-05T12:28:14 #kisslinux <dylanaraps> I store any patches I use in the source directories 2020-10-05T12:28:25 #kisslinux <dylanaraps> And I then copy my .config around 2020-10-05T12:28:42 #kisslinux <Rio6> ok 2020-10-05T12:29:13 #kisslinux <dylanaraps> I know other users have packages set up for their kernel 2020-10-05T12:29:20 #kisslinux <dylanaraps> I've never felt the need to do this though 2020-10-05T12:29:34 #kisslinux <Rio6> I was thinking if I reuse the same directory to build it might be faster to upgrade 2020-10-05T12:29:46 #kisslinux <Rio6> not sure if thatsbgood idea 2020-10-05T12:29:53 #kisslinux <dylanaraps> ccache is handy 2020-10-05T12:30:06 #kisslinux <dylanaraps> Having a small .config is also good 2020-10-05T12:30:15 #kisslinux <dylanaraps> My kernel takes 5 minutes regardless of ccache for instance 2020-10-05T12:32:12 #kisslinux <Rio6> i started with localconfig from another repo, might trim it down later 2020-10-05T12:33:30 #kisslinux <Rio6> havent used ccache before, is it much faster? 2020-10-05T12:33:59 #kisslinux <dylanaraps> The first build is slower but all rebuilds will be faster (so long as the cache is hit) 2020-10-05T12:34:34 #kisslinux <dylanaraps> It works even when package versions differ (as source files can be identical between releases) 2020-10-05T12:35:36 #kisslinux <dylanaraps> Here's my stats from the past 3 or so days: https://termbin.com/hhdcr 2020-10-05T12:36:17 #kisslinux <dylanaraps> Setting it up is really easy. kiss b ccache && kiss i ccache && export PATH=/usr/lib/ccache/bin:$PATH 2020-10-05T12:36:23 #kisslinux <dylanaraps> That's it 2020-10-05T12:36:46 #kisslinux <Rio6> I'll try it 2020-10-05T12:36:58 #kisslinux <Rio6> why is the bin not in normal path 2020-10-05T12:37:15 #kisslinux <dylanaraps> It contains wrappers for gcc, cc, c++, etc. 2020-10-05T12:37:38 #kisslinux <Rio6> ah k 2020-10-05T12:38:02 #kisslinux <dylanaraps> And it must be at the start of $PATH 2020-10-05T12:38:32 #kisslinux <Rio6> I think I've seen people use CC CXX variables instead 2020-10-05T12:38:45 #kisslinux <Rio6> but path sounds easier 2020-10-05T12:39:10 #kisslinux <dylanaraps> Yes 2020-10-05T12:39:16 #kisslinux <dylanaraps> It also works on the command-line 2020-10-05T12:39:58 #kisslinux <dylanaraps> Also when build systems don't listen to CC/CXX (and they don't use absolute paths to call compilers) 2020-10-05T12:40:10 #kisslinux <dylanaraps> It's set and forget really 2020-10-05T12:40:57 #kisslinux <dylanaraps> I'm using the latest master which is backwards incompatible to the current release. It uses zstd over zlib, BLAKE3 over BLAKE2, plus many more changes. 2020-10-05T13:45:44 #kisslinux <dilynm> dylanaraps: didn't see your message; kernel package is 15000 files; I'm using BusyBox grep and will test with gnugrep in a moment 2020-10-05T13:46:02 #kisslinux <dilynm> Latest kiss master seems to be working quite well although it's hard to find all the broken things 2020-10-05T13:47:15 #kisslinux <dilynm> Jkjk. Big issue 2020-10-05T13:48:57 #kisslinux <dilynm> With dash, busybox, and util-linux installed, util-linux and dash are in use (/bin/sh, etc), swapping everything for busybox yields a list of files swapped, but they say things like "/usr/bin/setsid is now provided by busybox (was )", and worse: "/usr/bin/swapon was provided by busybox (was dash)" 2020-10-05T13:53:14 #kisslinux <dilynm> Kiss a correctly reports that, for instance, hummingbird's init is in kiss/choices (and that the one in use is busybox's), but swapping it does not put busybox init in kiss/choices and the manifest reports that in fact both are in /usr/bin. It is, however, now hummingbird's init in bin 2020-10-05T14:05:50 #kisslinux <dylanaraps> dilynm: Thanks. Will try to reproduce. 2020-10-05T14:09:07 #kisslinux <dilynm> As for install speeds on packages >10k files, here's kernel 5.8.13, 11,597 lines in the manifest: 2020-10-05T14:09:10 #kisslinux <dilynm> linux, kissv5, bb grep: 2020-10-05T14:09:10 #kisslinux <dilynm> 55.93s real, 50.49s user, 9.43s sys 2020-10-05T14:09:10 #kisslinux <dilynm> linux, kissv6, bb grep: 2020-10-05T14:09:10 #kisslinux <dilynm> 33.9s real, 30.08s user, 7.03s sys 2020-10-05T14:09:11 #kisslinux <dilynm> linux, kissv5, gnugrep: 2020-10-05T14:09:11 #kisslinux <dilynm> 24.07s real, 18.92s user, 9.18s sys 2020-10-05T14:09:12 #kisslinux <dilynm> linux, kissv6, gnugrep: 2020-10-05T14:09:12 #kisslinux <dilynm> 15.04s real, 11.w8s user, 6.77s sys 2020-10-05T14:10:04 #kisslinux <dylanaraps> Nice 2020-10-05T14:10:57 #kisslinux <dilynm> Yeah, talk about 'sizeable speedups' with gnugrep xD 2020-10-05T14:14:17 #kisslinux <dylanaraps> I should push it more 2020-10-05T14:14:26 #kisslinux <dylanaraps> busybox grep is sloooow 2020-10-05T14:15:21 #kisslinux <dilynm> It's almost painful seeing those numbers 2020-10-05T14:15:57 #kisslinux <E5ten> dilynm: any chance you could try with toybox grep? curious how it compares here 2020-10-05T14:16:59 #kisslinux <dilynm> Sure! 2020-10-05T14:22:27 #kisslinux <sh4rm4^bnc> dilynm, what's the diff in gnugrep v5 vs v6? different default optimization flags to the compiler? 2020-10-05T14:23:50 #kisslinux <dilynm> kissv5: 41.08s real, 35.81 user, 9.23s sys 2020-10-05T14:23:50 #kisslinux <dilynm> kissv6: 33.46s real, 29.69s user, 6.82s sys 2020-10-05T14:24:38 #kisslinux <sh4rm4^bnc> i mean how the two builds differ 2020-10-05T14:25:00 #kisslinux <dylanaraps> v5 / v6 are kiss pkg manager versions 2020-10-05T14:25:12 #kisslinux <dilynm> Sh4rm4^bnc: these are all just single runs of kiss i linux (if we want 'real' results we should average multiple runs), identical CFLAGS for gnugrep in both 2020-10-05T14:25:25 #kisslinux <dilynm> Although busybox is built with empty CFLAGS because it doesn't cooperate with clang 2020-10-05T14:25:34 #kisslinux <dilynm> So that could heavily skew bb grep 2020-10-05T14:25:42 #kisslinux <E5ten> dilynm: are those new numbers the toybox ones? 2020-10-05T14:26:06 #kisslinux <dilynm> Actually, toy box and gnugrep were built with empty CFLAGS as well I just checked. So this is no optimizations 2020-10-05T14:26:10 #kisslinux <dilynm> Yes they are e5ten 2020-10-05T14:26:16 #kisslinux <E5ten> thanks 2020-10-05T14:27:24 #kisslinux <sh4rm4^bnc> gnu grep seems to be really optimized for speed: http://lh3lh3.users.sourceforge.net/reb.shtml 2020-10-05T14:28:16 #kisslinux <sh4rm4^bnc> busybox probably just uses the libc regex impl, which in musl libc's case is TRE (quite slow, but no quadratic pathological cases) 2020-10-05T14:28:16 #kisslinux <dilynm> Yeah it's like lightning 2020-10-05T14:32:31 #kisslinux <sh4rm4^bnc> it could then be made a lot faster by linking e.g. to oniguruma's posix api, but that would make the busybox binary a lot bigger 2020-10-05T14:43:30 #kisslinux <dilynm> That would be interesting... 2020-10-05T14:52:02 #kisslinux <sh4rm4^bnc> the only change needed to the code is #include <onigposix.h> instead of #include <regex.h> and adding -lonig to LDFLAGS 2020-10-05T14:52:11 #kisslinux <sh4rm4^bnc> however you gotta use git master of oniguruma 2020-10-05T14:58:27 #kisslinux <dilynm> I'll try it out later today and see what happens :o 2020-10-05T14:59:19 #kisslinux <sh4rm4^bnc> i tested it for tinyproxy's config parser and parsing 32K allow/deny statements took 0.3 sec vs 2.5 secs with musl's regexec 2020-10-05T15:00:59 #kisslinux <sh4rm4^bnc> neatlibc's regex.c used 0.8 secs, but i doubt it's up for the task of being used in grep 2020-10-05T18:19:37 #kisslinux <dilynm> Using onigurama, kissv5 bb grep: 1m07.43s real, 1m02.39s user, 7.90s sys; kissv6 bb grep: 1m00.31s real, 56.71s user, 5.91s system 2020-10-05T18:19:44 #kisslinux <dilynm> I feel like I've done something wrong tho