IRC Logs for #crux Wednesday, 2017-04-12

*** moth has joined #crux00:03
*** timcowchip has left #crux ("http://quassel-irc.org - Chat comfortably. Anywhere.")00:21
cruxbot[opt.git/3.3]: bluez: updated signature00:34
*** moth has quit IRC00:37
*** moth has joined #crux00:44
*** e3d3 has joined #crux00:48
e3d3Is using Crux more simple than e.g. Arch Linux or derivate Manjaro ?00:50
joacimi don't know if it would be simpler, or as simple, but getting started with arch and such is probably easier00:53
joacimif you meant easy rather than simple00:54
e3d3I guess so (not a native English speaker)00:55
e3d3I'm mostly interested in Crux advertising 'KEEP it simple'. My distro's grow quickly complex after install, that why I solved some problems by using multiple distro's.00:59
e3d3Does Crux use systemd ?00:59
nogagplzno01:03
nogagplzsysvinit afaik01:03
e3d3thanks. I like to avoid systemd, but still don't have a preference for alternatives01:05
e3d3last question; is Crux light enough to run in virtualbox on my laptop with 2 GB ram ?01:07
jaegersure, you can run it that way... but keep in mind that crux is primarily a source-based distro, so lots of compiling01:09
e3d3Good to remind me, virtualbox allready is not my fastest program, and I am not very patience. Thanks.01:11
*** e3d3 has left #crux ("rcirc on GNU Emacs 25.1.1")01:14
joacimdepending on which version of OS X you run, 2GB is probably just barely enough for OS X and a few applications01:14
joacimoh...01:14
*** moth has quit IRC01:54
abenzindeed02:05
*** timcowchip has joined #crux02:14
*** _________mavric6 has quit IRC02:38
*** _________mavric6 has joined #crux02:39
*** emmett1 has joined #crux02:52
*** tilman_ has joined #crux04:02
*** emmett1 has quit IRC04:17
*** emmett1 has joined #crux04:25
*** abenz_ has joined #crux04:58
nwegood morning guys!05:00
*** tilman has joined #crux05:00
*** tilman_ has quit IRC05:03
cruxbot[opt.git/3.3]: wine: 2.4 -> 2.505:27
cruxbot[contrib.git/3.3]: python3-streamlink: 0.3.2 -> 0.5.005:28
*** casucci has joined #crux05:54
*** casucci has left #crux ()05:57
brian|lfswine is certainly compiling slower on 8 cores then 16 with a skylake I7 verses a 5960X with 16 cores06:31
frinnstnot very surprising06:53
abenzwhat?06:54
abenzslower on 8 vs 16, vs 5960X ?06:55
abenzyour england.. not very clear06:55
*** darthlukan has joined #crux06:56
*** erdic_ has joined #crux06:57
*** DaViruz has joined #crux06:59
emmett1he just want show off his new 16 cores machine, i understand that ;)07:06
*** timcowchip has left #crux ("http://quassel-irc.org - Chat comfortably. Anywhere.")07:31
Romsterjoacim> i rather see more emulator-like repositories  tbh <- feel free to contact me for access to the emulators repository08:18
cruxbot[compat-32.git/3.3]: libdrm-32: 2.4.75 -> 2.4.7908:27
cruxbot[compat-32.git/3.3]: sqlite3-32: 3.17.0 -> 3.18.008:27
cruxbot[compat-32.git/3.3]: dbus-32: 1.10.16 -> 1.10.1808:27
cruxbot[compat-32.git/3.3]: gdk-pixbuf-32: 2.36.5 -> 2.36.608:27
cruxbot[compat-32.git/3.3]: glib-32: 2.50.3 -> 2.52.108:27
cruxbot[compat-32.git/3.3]: mesa3d-32: 17.0.2 -> 17.0.308:27
cruxbot[compat-32.git/3.3]: pango-32: 1.40.3 -> 1.40.508:27
cruxbot[compat-32.git/3.3]: xorg-libpciaccess-32: 0.13.4 -> 0.13.508:27
Romstermore cores != faster, it depends on the processing power of each core.08:29
Romsterdocker -> ccache -> distcc -> on 8 computers, is a nice setup.08:33
just_funRomster, any number regarding the speed-up from ccache -> ccache+distcc*8 ?08:37
Romsterit's hard to tell really.08:38
Romsteras ccache skews the results.08:39
Romsterand that i have other things running on this desktop that has the main compile job on as well.08:39
Romsterwell I can time something bit like llvm08:39
Romsterbig*08:39
just_funI don't plan to use distcc, because I don't have 2 Crux machines on the same network, yet. I was just curious. No need to measure just for me.08:41
Romster24 real cpu cores at my disposal08:42
*** pekka10 has joined #crux09:05
Romsterjust_fun, ok i am using binutils to test as that's what the base line SBU is set to09:30
Romsterccache = real0m29.527s09:30
Romsteruser0m27.124s09:30
Romstersys0m8.919s09:30
Romsterdistcc = real0m28.273s09:32
Romsteruser0m25.943s09:32
Romstersys0m8.917s09:32
Romsterwell that's a bad test guess it needs to be bigger09:32
abenzdistcc09:33
abenzdo you use it to test all those ports ?09:33
Romsterlocal build of 4 make jobs09:35
Romsterreal0m27.114s09:35
Romsteruser0m26.793s09:35
Romstersys0m8.624s09:35
Romsterthat's for binutils09:35
Romsteri need to test it on something big now.09:35
Romsterin the past i was then i stopped and just used docker and ccache09:36
Romsterjust recently got my distcc all updated and going again.09:36
Romsteras you can see it makes zero difference on small stuff like binutils09:36
abenzi see09:37
Romsterbut on big builds it is faster09:37
Romstermaybe qt5 is a good test or llvm i am not sure what to test it against.09:37
Romstermaybe mesa3d as that's big but not humongous09:37
*** Introoter has joined #crux09:48
*** emmett1 has quit IRC10:08
just_funRomster, from what I've seen, ccache doesn't help much rebuiling the next release of most of the big programs, so distcc will help when building a new version.10:11
*** blueness has quit IRC10:16
*** onodera has joined #crux10:21
*** blueness has joined #crux10:21
*** onodera has quit IRC10:22
*** onodera has joined #crux10:22
*** chinarulezzz has joined #crux10:32
*** blueness has quit IRC10:41
*** blueness has joined #crux10:45
*** john_cephalopoda has joined #crux10:45
frinnstjust_fun: sure helps if something goes wroung tho10:49
frinnstRomster: what command were you timing?10:51
Romsteri waraped time { ./configure ... ; make ; make DESTDIR=$PKG install } around binutils, but  made a mistake i did 'CCACHE_DISABLE=true' instead of 'export CCACHE_DISABLE=true' so i'll have to do that again11:20
joacimRomster: I might when I have CRUX running on a desktop system =)11:23
joacimI just have the server right now, and I generally only want to maintain ports that I use.11:23
joacimoff to work because my coworker is "sick"11:25
*** druid_droid has quit IRC11:30
*** druid_droid has joined #crux11:32
Romsterhttps://gist.github.com/therealromster/aa694345269b37ecbfc72d43dfa4464411:37
Romsterbinutils and mesa3d11:38
Romsterbinutils is considered the 1 SBU calibration on any system11:38
Romsterthis does not include the time to unpack the source or pack the $PKG files.11:40
Romsterjust the configure and build and install phase11:41
frinnstreal0m28.246s11:41
frinnstodd11:41
frinnst-j1611:41
frinnstdidnt do a make install tho11:41
Romstertry with -j8 isn't it 8 real cores 16 HT?11:42
frinnstyup11:42
frinnstanyways, lunch over.. back to work11:42
Romsteri just stuck time { ./configure ; make ; make install }  in the Pkgfile11:42
Romsteri have 4 real cores in this AMD pc.11:42
Romsteroh and I compile in ram as well.11:43
Romstertmpfs11:43
Romsteri take a bit of a hit with -j24 when it's just hitting ccache11:44
Romsterjust_fun, ^11:45
Romsterjust_fun, as for the next release it depends how much of the code base changes, and how much of that has a different object file due to dependencies changing.11:46
just_funI've tried with the linux kernel11:46
just_funwith direct and preprocess mode of ccache11:46
just_funccache is almost useless11:46
Romsterthat's actually not a bad test but that depends on what options are set11:46
just_funwith ccache I've used: direct (default), preprocess, preprocess+removing current dir11:47
just_fun(removing current dir from hashes)11:47
Romsterif you are just going up a patch level it should be fast if you go up a major version then i would expect the behaviour you speak of.11:47
just_funbase_dir and something with hash_dir, iirc11:47
*** BitPuffin|osx has joined #crux11:48
just_fun4.10.x11:48
just_funMy guess is that some version is included in some header which is included everywere11:48
Romsteri didn't play around much with the pre-processing on ccache11:48
Romsterwhich then invalidates the entire cache11:49
Romstercould be11:49
just_funis 2 times slower, from what I've tested11:49
Romsteri find with firefox it's fast if it's a patch level change but slower on a major version change.11:49
just_funI've cleaned the cache before running the tes11:49
just_funThat is what I was thinking: it depends on the project11:50
Romsterthe first ru you take a few seconds cost to store the objects in ccache11:50
Romsterrun*11:50
just_funI've made 3 runs: first, second and next version11:51
just_funmeasuring time and cache hits11:51
Romstersometimes (often) a patch will fail or i do something stupid with ln or rm or anything else and then it fails and i have to buuild over (again), ccache gets me to where it failed last time fast.11:51
just_funah!11:51
Romsteror I just need to rebuild because ABI changed on a library.11:52
Romsteron a large package but that is only one small library of the entire package11:52
Romsterso the small part that changes goes slow but the rest is sped up. but I always have the linking overhead on the host.11:53
Romsteron every build.11:53
Romsterthis distcc is a mix of 1 xeon, 1 intel q6600, 2 phenom II's and 4 x intel core2 duos.11:54
Romsterbit of a mixed bag i know.11:55
Romsterover gigabit lan11:55
just_funccache speeds a lot, but can't help thinking about direct mode,11:55
just_fun include <xx.h>11:56
just_funif xx.h is created in the current dir,11:56
just_funccache won't notice11:56
Romsterah11:57
just_funis not a common case, but still11:57
*** moth has joined #crux11:57
just_funthis is why I've tested with preprocessor mode11:57
just_funwhich can ignore spaces too11:57
Romsterdoes that do much?11:58
just_funno11:58
just_funonly on specific projects, probably11:58
Romsteri'd be worried about false errors it could introduce11:58
just_funonly playing with base_dir and hash_dir11:59
just_funonly when ...11:59
just_funotherwise, is safer and slower than direct mode11:59
just_fun:D11:59
RomsterCCACHE_COMPILERCHECK="%compiler% -dumpversion; crux"11:59
Romsteri set that so i don't clobber the cache, on different gcc versions and crux versions.12:00
just_funI use something like: CCACHE_COMPILERCHECK="string:$(%compiler% -dumpversion; crux)"12:01
Romsteri probably could just rely on the gcc version alone but in the past i've done other builds with the same gcc versions12:01
just_funto avoid spawning processes12:01
Romsterstring:$()12:02
Romsterokay that is different doesn't it default to string?12:02
just_funnot from what I've read12:03
just_fundefault is shell command12:03
just_fundefault is mtime of gcc12:03
just_funbut if set, is a command, from what I've understood12:03
just_funthe difference seems to be too little12:04
Romsterhmm i think i am doing that wrong and should be how you got it. looking in man now.12:04
Romstera command string12:05
Romsterthat's the bit i read12:05
RomsterHash the standard output and standard error output of the specified command.12:05
frinnstthe kernel is very good at telling what needs a recompile and what needs to be rebuilt12:05
Romsteryeah it's MAkefile is well done.12:06
RomsterMakefile*12:06
Romster Several commands can be specified with semicolon as separator.12:06
Romsterstring: is more to get the environment variable12:07
Romsterjust_fun, i'd expect calling a $() sub shell to run those commands on every compile would be more expensive that what i am doing.12:08
Romsterand i was probably going over board when i used a ASM compiled "crux" command than the shell script "crux" command provided by crux. http://crux.ster.zone/projects/crux/12:10
just_funI set CCACHE_COMPILERCHECK before starting the build (contained)12:10
Romstertan executes faster than having to invoke shell every time. if i am doing that  could just set the string value in the command directly. and be even faster...12:10
just_funI don't have the script handy12:11
just_funbut I think I do something like:12:11
just_funoutput=...; export CCACHE_COMPILERCHECK="string:$output"12:11
just_funand start the build12:11
just_funbut, again, is just OCD ... :) because it doesn't seems to matter much12:13
just_funat least on my tests on a very slow machine12:13
Romsterwell CCACHE_COMPILERCHECK="string:$output" is valid and better than calling a sub shell every single time.12:13
RomsterCCACHE_COMPILERCHECK needs to be fast.12:13
Romsterand i should be doing your method...12:13
Romsterbe faster than calling gcc and crux command over and over.12:14
Romsternever thought of that -_- thanks for poinging that out just_fun12:14
just_funthank you for pointing me to ccache12:15
Romsteri'll do it that way with string:$CCACHE_CRUX set12:15
just_funbtw, not sure how the ENTER is expanded in $output12:15
just_funwhen doing "gcc -dumpversion; crux"12:16
Romsterit would be \n in the string12:16
Romstershould return as "6.3.0\nCRUX version 3.3"12:19
Romsterthen that gets hashed.12:19
Romster; seems to concentrate the string on the same line when it's invoked in a sub-shell12:20
Romsterfoo=$(gcc -dumpversion; crux)12:21
Romsterecho $foo12:21
Romster6.3.0 CRUX version 3.312:21
Romsterone other problem i see i set the PATH for ccache as PATH="/usr/lib/ccache/:$PATH"12:26
Romsterso in my call i use /usr/bin/gcc ...12:26
Romsterexport OS_CCACHE_COMPILERCHECK="$(/usr/bin/gcc -dumpversion; crux)"12:26
Romsterexport CCACHE_COMPILERCHECK="string:$OS_CCACHE_COMPILERCHECK"12:26
Romster#export CCACHE_COMPILERCHECK="%compiler% -dumpversion; crux"12:26
Romsterto replace my slower command with a environment variable12:27
just_funI think ccache will run gcc with -dumpversion12:27
Romsterso it wont try to use /usr/lib/ccache/gcc that is a symlink to /usr/bin/distcc12:28
just_funah!12:28
Romsteri'd rather set the path to gcc directly.12:28
Romsterto save any possible issues.12:28
Romsterit will probably work but i'd rather be safe than sorry12:29
Romsteralso when i wanna use ccache with distcc just one line export CCACHE_PREFIX="distcc"12:30
Romsterexport DISTCC_HOSTS="--randomize 192.168.1.10/4 192.168.1.11/4 192.168.1.30/4 192.168.1.31/4 192.168.1.51/2 192.168.1.52/2 192.168.1.53/2 192.168.1.54/2"12:30
Romsterit's all DHCP now but i set a ton of DHCP reservations so they wont change ip12:31
Romster#export MAKEFLAGS="-j$(nproc)"12:31
Romsterexport MAKEFLAGS="-j$(/usr/bin/distcc -j 2> /dev/null)"12:31
Romsteractually wish pkgmk.conf by default used export MAKEFLAGS="-j$(nproc)"12:32
just_funsee the last lines in crux-devel about that :D12:36
Romsterabout random races failing a compile?12:42
Romsteri'd rather set -j1 where necessary than to have the entire system run at -j112:42
onodera^, I think the *vast majority* of CRUX users use -j>112:43
onoderaso a default MAKEFLAGS with nproc would make ports have less bugs because they would probably get detected quicker12:44
Romsterexactly12:45
Romsterbuilding thunderbird (wonder if i should do stats on that and add it to my gist or something more mainstream like qt5)?12:46
just_funonodera, using -j2.. by default adds bugs to the ports12:47
Romsteri've been using -j4 on all my builds and other times -j2412:48
Romsterif anyone is going to find a problem i would12:48
just_fun:)12:48
just_funRomster, you've missed the one in gnu-efi12:49
just_funAlso, I've seen bc port switching to -j112:49
john_cephalopodaI set makeflags to "-j8".12:50
RomsterI have no hardware yet that has EFI12:51
john_cephalopoda(In .bashrc and .bash_profile)12:51
Romsterso that's why i didn't even look at that.12:51
Romsteri also spent a ton of time fixing "depends on:" so things just work from opt core compat-32 and xorg12:52
Romsteri don't catch everything, but i do find a few.13:03
*** druid_droid has quit IRC13:07
*** druid_droid has joined #crux13:13
*** druid_droid has quit IRC13:18
*** druid_droid has joined #crux13:19
*** blueness has quit IRC13:21
*** blueness has joined #crux13:25
*** chinarulezzz has quit IRC13:26
*** chinarulezzz has joined #crux13:27
*** druid_droid has quit IRC13:32
*** emmett1 has joined #crux13:40
*** druid_droid has joined #crux13:44
cruxbot[opt.git/3.3]: gtksourceview: 3.18.1 -> 3.24.114:27
*** moth has quit IRC14:35
*** moth has joined #crux14:35
*** onodera has quit IRC14:51
cruxbot[opt.git/3.3]: dovecot: update to 2.2.29.114:54
cruxbot[opt.git/3.3]: fuse3: update to 3.0.114:54
cruxbot[opt.git/3.3]: mupdf: update to 1.1114:54
cruxbot[opt.git/3.3]: nano: update to 2.8.114:54
*** moth has quit IRC14:57
cruxbot[opt.git/3.3]: fontconfig: update to 2.12.115:00
cruxbot[opt.git/3.3]: libpcre2: update to 10.2315:07
*** moth has joined #crux15:20
*** blueness has quit IRC15:26
*** onodera has joined #crux15:45
*** emmett1 has quit IRC16:12
*** druid_droid has quit IRC16:16
*** Guest2312 has joined #crux16:59
*** chinarulezzz has quit IRC17:05
*** blueness has joined #crux17:27
*** Guest2312 has quit IRC17:54
*** john_cephalopoda has quit IRC18:12
*** john_cephalopoda has joined #crux18:13
*** druid_droid has joined #crux18:27
*** tsaop has joined #crux18:54
*** tsaop has quit IRC19:13
pedjafrinnst, what's your CPU usage with x265 encoded, totally legit, media files :) ?19:18
pedjadoes your graphics card has h.265 hardware-decoding?19:19
pedjamplayer build doesn't pick up x265 libs. hm.19:26
pedjaaah, it's HEVC.19:31
pedjait's been a while since I saw 'your system is too slow to play this'19:32
pedjaCrux Docker image, using latest updated ISO (yes, jaeger is cool like that) is ~470Mb.not bad.19:42
jaegernice19:59
jaegerMy plan is to set up an automatic git update for the docker image when the ISO is updated, just haven't gotten there yet19:59
jaegerupdating docker/runc/containerd/libnetwork currently, though19:59
jaegerseems like it has a lot fewer man pages now, hrmm20:04
pedjadid you use 'make man-pages' from the root dir, or?20:13
pedjamanpages*20:14
pedjathey haven't moved all of them from 'the legacy way of building' yet, iirc :)20:15
pedja(md2man-all.sh from the man subdir)20:16
pedjaso I just 'make manpages' and let it do its magic20:17
jaegerThey did break the legacy method20:17
jaegerbut the docker machine method seems to still work20:18
pedjadocker machine?20:20
jaeger"make manpages"20:22
jaegerit runs in docker20:22
pedjaI am thinking of packaging docker-compose, it doesn't seem to have *that* many dependencies20:23
jaegerinstead of just the md2man-all script which worked before but now that the man pages dir is reorganized it hasn't been updated/fixed20:23
pedjayeah, it's on their TODO list :)20:23
Introoterhey, i know no one likes it in here, but is there a tree with a makefile for systemd if i wish to use it?20:24
jaegerIntrooter: I think the only person who's messed with it much in crux was horrorstruck but I've no idea where his repo was if he had one20:25
jaegerthough if someone else has done it, too, they're welcome to speak up20:25
Introotermeh ill just build from src20:26
pedjabuilding it it's just the start, though.20:26
Introotereh?20:26
pedjaintegrating it, so the system actually boots :)20:27
Introoteri think i can do that, i've switched enough times between inits on Gentoo, this would be a little more manual work i guess20:28
pedjashare your experience here after20:29
Introotersure20:29
Introoterill have some time at the end of the week20:29
pedjawill you use a VM to test it?20:30
Introoterim thinking of testing it on a VM first20:30
Introoteryes20:31
pedjasmart :)20:31
Introoterim also looking to do a minimal build20:31
Introoterill leave out networkd, resolved, nspawn, machined, etc20:31
cruxbot[contrib.git/3.3]: go: updated to version 1.8.120:31
cruxbot[contrib.git/3.3]: containerd: updated to version 422e31c20:31
cruxbot[contrib.git/3.3]: runc: updated to version 9c2d8d120:31
cruxbot[contrib.git/3.3]: docker: updated to version 17.04.0-ce20:31
brian|lfswow cruxbot has been busy today20:41
john_cephalopodaUnfortunately I only got 2 GB of mobile data :รพ20:54
john_cephalopodaElse I'd update.20:54
joacimconfiguring a system with uefi is quite a bit more annoying than bios21:17
joacimi miss lilo already21:17
jaegerpretty easy after you've done it a few times21:25
joacimgrub2 understands symlinks?21:26
joacimif i could make a symlink to the kernel image make install produce21:27
joacimthe default config for grub-mkconfig looks for vmlinuz-*, and make install just names it vmlinuz21:27
joacimcould modify the scripts so that the image is named vmlinuz-version tho21:28
jaegerno idea there21:34
jaegerhaven't tried with symlinks21:34
pedjaanybody using rEFInd as a bootloader?22:12
pedjaI've seen it mentioned a few times22:12
jaegernot I22:27
*** timcowchip has joined #crux22:44
joacim"If the PC uses an AMD Carrizo DDR4 processor, installing this update will block downloading and installing future Windows updates"22:47
joacim"Microsoft is working on a resolution and will provide an update in an upcoming release"22:47
joacimI wonder why they're working on a solution for that =)22:47
*** druid_droid has quit IRC23:00
*** onodera has quit IRC23:11
*** druid_droid has joined #crux23:16
joacimi don't get it. turboboost on this laptop is a lot more lively than it is on my desktop23:17
joacimi rarely see frequencies higher than 3.5 GHz on my desktop, even tho it should boost up to 4GHz23:17
joacimthis laptop boosts up to 3.3GHz all the time23:17
*** john_cephalopoda has left #crux ("Trees are hiding until the final day.")23:23
*** Introoter has quit IRC23:28
*** moth has quit IRC23:42

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!