IRC Logs for #crux Friday, 2012-06-15

*** nthwyatt has quit IRC00:26
*** ardo has joined #crux00:55
*** lasso|qt has joined #crux01:12
*** nthwyatt has joined #crux01:15
*** vaddi has joined #crux01:29
juegood morning02:18
*** mike_k has joined #crux02:44
Romstermorning all03:08
Romsteror evening.03:09
*** horrorStruck has joined #crux03:37
*** ardo has quit IRC03:46
Romstertesting chromium-19.0.1085.0.tar.bz2 once it's downloaded.05:08
teK_where?05:12
Romsterhttp://gsdview.appspot.com/chromium-browser-official/chromium-19.0.1085.0.tar.bz205:12
Romsteri got another 15 minutes to download it05:13
Romster51/174mb05:13
Romsterneed faster internet.05:13
teK_so mylocal cache seems tobe b0rked as I get a 404..05:16
Romstertry from here it redirected me to this one05:17
Romsterhttp://o-o.preferred.nuq04t11.v1.cache1.c.bigcache.googleapis.com/chromium-browser-official/chromium-19.0.1085.0.tar.bz205:17
teK_yep05:18
teK_thanks05:18
RomsterHTTP/1.0 405 Method Not Allowed from surl looking at file header05:19
Romstercurl*05:19
jueah, that sound good:05:20
teK_brb05:20
juethe reson why we moved glib to core was pkg-config, they removed the internal glib05:21
juewhich is not very nice because that leads to a circular dependency05:21
Romsteroh you thinking glib can go in opt again?05:21
juebut now I've read this -> http://lists.freedesktop.org/archives/pkg-config/2012-April/000824.html05:22
jueand the code is already in the official repo -> http://cgit.freedesktop.org/pkg-config/05:22
jueRomster: yeah, that would fix our problem, we can move glib back to opt05:23
Romstercould pkg-config be set to use the internal copy unless recompiled with opt/glib present? or is that a bad idea?05:25
juewhat do we gain with that?05:25
Romsteri'm not sure, but they used glib for a reason.... newer than the internal one.05:27
Romsterat some point in time.05:27
Romsteralso i don't know why glib defaults to it's own pcre05:28
juewhat I've read is: pkg-config uses only very few basic functions from glib, so that doesn'05:28
juet matter at all05:28
Romsterah carry on then.05:29
jueIMO such a low level tool like pkg-config shouldn't depend on huge libs like glib05:29
jueoops, glib should use system pcre05:29
Romsterthen it was silly of them to remove there cut down glib out of pkg-config causing so much pain for distro maintainers.05:29
jueyeah, indeed05:30
Romsterthank goodness they have re-added it. so you need to wait for next pkg-config version or is it out already05:30
juea 'finddeps glib' shows libpcre for me05:30
Romsteror just patch pkg-config for now.05:31
Romsterhmm i read the configure --help and saw this. --with-pcre=[internal/system]                          whether to use system PCRE [default=internal]05:31
jueRomster: sorry, but we have a  --with-pcre=system \05:38
juein our Pkgfile ;)05:39
Romsteroh i fail -_-06:11
teK_jue: build went fine06:13
Romsterjues regex looks better but it needs to be applied to mroe files than that single one teK_06:15
Romsterneed to grep the source for all of them06:16
teK_chromium builds with it. :-)06:16
teK_I will report this to sepen06:16
teK_building the newer chromium version right now06:16
Romsteri buit the newer one it failed.06:16
teK_I'll see :P06:17
teK_it has to work, this red border just sucks06:17
Romsteregrep -r '<glib/.*\.h>' /usr/ports/work/chromium/src/chromium-19.0.1085.0/06:17
Romsterton of files needs to be changed to glib.h06:18
teK_ah06:18
Romsterfind . -type f -exec sed -i 's|<glib/.*\.h>|glib.h|g' {} +06:22
Romsterugly but should work testing.06:22
teK_thought about that too but a single difffile will bemore  efficient06:22
teK_one second :)06:22
Romsterhmm06:22
Romstertrue but every version it'll change.06:23
teK_yeah06:23
teK_after all sepen has to decide06:23
Romstermistake i didn't do the sed right on the new string missing <>06:23
teK_+ a bug should be filed @chromium06:23
Romsteryeah06:24
Romsterbut why did glib have to break so badly than to only warn.06:24
Romsterto do it right they would have to ifdef it for the glib version06:24
Romsterbuilding06:25
jueI bet using glib includes direct is deprecated since a long time ;)06:25
Romsterprobably but we didn't move for a long time either but if that was the case chromium should cover this already06:26
Romsterhmm that still did not work06:28
juebut yeah, a warning from glib would be sufficient06:28
Romsterfind . -type f -exec sed -i 's|<glib/.*\.h>|<glib.h>|g' {} +06:28
Romsterany ideas?06:28
juewhat's the problem?06:29
teK_jue: buildt some packages with the new pkg-config without problems06:31
jueteK_: good to hear, thanks06:32
Romsterhttp://pastebin.com/Ry4Lgb6F06:32
Romsterseems to be a link time error06:32
Romsteri'll test pkg-config in chroot shortly after this chromium06:33
teK_it's the same I get with 21.x06:33
teK_22:17 < teK_> do you enjoy it just asmuch as I do when chromium fails while linking the final binary right afterabout 30 minutes buildtime? :P06:33
teK_;)06:33
Romsteryeah lol06:33
Romsteronly it's less than 30 minutes now due to ccache06:34
Romster-fvisibility=hidden might have something todo with it.06:36
Romster-march=pentium406:36
Romsterthis thing overrides compiler options... damn thing also includes the kitchen sink06:37
Romsterhmm did someone say something about -lgdk_pixbuf-2.0 earlier missing.06:38
Romsterhmm no that's ok usr/lib/libgdk_pixbuf-2.0.so06:39
juethe missing functions are from nspr, I guess we need to add -lnspr4 or similar for the link call06:44
juebut actually I've no idea about chromium ;)06:45
juebbl06:45
Romsterhmm i thought those were glib functions06:47
Romsteri should of grepped the includes06:47
teK_jue was right06:49
teK_wow :)06:49
teK_http://dpaste.com/759950/ is the correct linker command :P06:49
teK_they see mescrollin..06:49
Romstertrust me to do a half assed job of finding the issue.06:50
teK_bla :)06:50
teK_still with red borders06:51
teK_yeah.06:51
Romster+ find . -type f -exec sed -i 's|<glib/.*\.h>|<glib.h>|g' '{}' +06:52
Romster+ export 'LDFLAGS= -lnspr4'06:52
Romsteris that like the flash bug with swapped red blue?06:52
Romsterusing vdpau?06:53
teK_http://www.abload.de/image.php?img=2012-06-14-204741_140e3gkp.png06:53
Romstermy gawd that's red06:53
teK_it's installed, yes06:53
teK_i rebuilt the whole chromium dependencies06:54
teK_maybe I shall revert the last nvidia update06:54
Romsterhmm i can test but i'm not up to date on everything yer06:54
Romsteryet06:54
Romsterreal3m39.127s06:55
Romsteruser5m27.735s06:55
Romstersys0m31.898s06:55
Romsterbuilt for me too.06:55
Romstergood call jue06:55
*** lasso|qt has quit IRC06:57
Romsterbuilding on multilib now06:59
Romsterthen i'll see if everything is red or not.06:59
teK_using the older nv-module won't help07:02
teK_dammit..07:03
Romster22:04:00 up 10 days,  4:14,  1 user,  load average: 6.91, 3.82, 2.0307:04
Romsterbit of load atm will be done shortly07:04
Romsterbuilding two things at the same time in different chroots while updating chromium on my system07:04
*** goodbox has quit IRC07:06
RomsterIt seems that the latest glib update (2.32.3) breaks emacs07:08
Romsteroh noes...07:08
Romsteron ML07:08
Romsteri never tested emacs07:08
frinnstyeah just tested also07:08
frinnstemacs 24.1 was released not that long ago07:10
Romsterworks ok?07:11
frinnstcompiling..07:12
Romsterah http://xkcd.com/303/07:13
frinnsthahha stfu07:13
Romsteri couldn't help it :D07:13
Romsterbesides you owed me one for that annoying thing you did ages ago so were even now.07:14
frinnstlol07:14
Romsterbumping xf86-input-synaptics shortly07:15
teK_http://code.google.com/p/chromium/issues/detail?id=27488 claims libpng is evil :o07:15
RomsterOk, I don't know then, deferring to... tony!07:17
Romsterhmm teK_ you did regenerate librsvg cache?07:17
*** lasso|qt has joined #crux07:18
teK_no I didnt07:18
frinnstseems like it works with emacs 24.107:18
Romstergdk-pixbuf-query-loaders > /usr/etc/gtk-2.0/gdk-pixbuf.loaders07:18
Romsterthen exit and start chromium07:19
Romstermight be something screwie there.07:19
Romsteri've hit issues with a stale librsvg cache before.07:19
teK_noe07:20
teK_nope07:20
Romsterdamn07:20
Romsteri'll know soon once this finishes compiling...07:20
Romstermore excuse to go and get another scotch07:20
teK_:o07:21
Romstercan't see why it cant' use system webkit than it's own.07:21
Romsternote to self must get distcc setup again07:23
Romstercan never have enough power07:23
teK_that's true, so I got myself a decent cpu and plenty of RAM07:29
Romsterthis box isn't bad but i can OC it and i haven't yet07:30
Romster8GB ram AMD Phenom(tm) II X4 965 Processor07:30
Romsterwas looking at a 6 core upgrade but i'll wait if they ever make a 12 core.07:31
Romsterstill compiling ugh07:33
frinnstteK_: only chromium that gives you troubles?07:36
teK_yes07:36
teK_though I do not use thatmany GUI-based apps07:37
teK_gimp works fine07:37
frinnstgaaah, nobody wants to maintain emacs?07:37
Romsternothing red on mine teK_ looks a sexy blue07:43
teK_-_-07:43
Romsterbut like i said my system is lagging on everything but leading on xorg07:43
Romsterso it's not the browser itselfs fault at least07:43
Romster(exe:9592): Gdk-WARNING **: XID collision, trouble ahead07:45
teK_it always told me that07:45
frinnstchromium is even a bigger behemoth than firefox07:46
Romsteryeah i'd say feels like driving a tank07:50
teK_a clean ~ won't help either07:55
teK_and it's not gtk related as switching to gtk mode resolves rendering problems forall  gtk parts (icons are shown) yet the tab-bar still is b0rked07:55
frinnstwow, udev is a dependency?!07:56
frinnstPackage libudev was not found in the pkg-config search path.07:57
frinnstfuck that shit07:57
frinnstyou are on your own tek :)07:57
RomsterteK_, if it helps i can list my dep versions07:57
teK_I will install windows07:58
cruxbot[opt.git/2.7]: emacs: updated to 24.108:02
RomsterteK_, http://pastebin.com/EsHitw2M08:03
Romsterif that helps or not i don't know but it works without any red anywhere with those versions.08:05
teK_there are some differences, yeah-_-08:23
Romsteryeah long task ahead08:26
frinnstnsfw but awesome: http://www.youtube.com/watch?v=33zPlnhymCU&feature=g-all-u08:31
RomsterThe uploader has not made this video available in your country.08:33
Romsterbiatch08:33
teK_I image frinnst to behave like that in reality08:33
teK_imagine08:33
teK_but not that friendly smilingfella inthe yellow shirt08:34
teK_fucking beeping08:34
Romsterhaha i hate the beeping08:35
teK_it's ridicilous08:35
teK_speaking of which.. http://news.slashdot.org/story/12/06/15/0139255/aussie-telco-lays-new-fiber-for-microsecond-trading-boost08:40
*** joe9 has joined #crux08:40
joe9just wanted to share this with the folks here: my system is so fast with these flags that I cannot believe it is the same system.08:41
joe9export CFLAGS="-O2 -pipe $(echo \"\" | gcc -march=native -v -E - 2>&1 | grep cc1 | sed -e's/^.* - //g')"08:41
joe9export CXXFLAGS="-O2 -pipe  -fvisibility-inlines-hidden $(echo \"\" | gcc -march=native -v -E - 2>&1 | grep cc1 | sed -e's/^.* - //g')"08:41
joe9if anyone is not using them, please think about it.08:41
joe9firfox is so much zappier/quicker now.08:41
teK_whyis this better than -march=native -mtune=native?08:43
joe9teK_: with distcc, that will not fly.08:44
joe9 as suggested by http://en.gentoo-wiki.com/wiki/Hardware_CFLAGS08:44
joe9#  http://gentoo-what-did-you-say.blogspot.com/2011/07/finding-cpu-flags-using-gcc.html08:44
joe9check the above urls.08:44
joe9i found them helpful.08:44
teK_ok08:44
joe9cannot say this enough. especially firefox, it is so much quicker now.08:46
Romsterjoe9, you've only touched the tip of the iceberg there is so much more than that that can be done.08:46
joe9romster, please share that with me. What else?08:47
rmullMy CPU is not recognized by -march=native :(08:47
joe9i am willing to spend more time and effort to learn.08:47
Romsteri'm pretty drunk right now :D08:47
Romsterusing that bintutils patch i posted to you for another.08:47
frinnstrmull: what cpu are you using?08:48
rmullfrinnst: Via Nano08:48
rmullgcc thinks it's a core2 but shit is out of wack if I use that setting08:48
frinnstobligatory cflags joke: http://funroll-loops.info/08:48
Romsteryeah seen it frinnst -fomg-optimise08:49
joe9frinnst: that link makes me want to go back to exherbo.08:52
frinnstwould be nice if firefox would work with avx08:53
Romsterjoe9, i had runit that was boosting loading speed08:53
frinnstdoubt it would make that much difference though08:54
Romsterparallel loading08:54
Romsterunless we got that already in current rc.08:54
joe9-fomit-frame-pointer -- anyone has this set in their default CFLAGS. any issues noted?08:54
Romsternone but it makes it harder to debug without a pointer08:55
joe9ha, ok. does that speed up things?08:55
Romsterif you don't need to debug it's fine to use08:55
Romsteryes a little08:55
joe9romster, can you please any more clues to speed up things a lot more. i am referring to your comment here: "you've only touched the tip of the iceberg there is so much more than that that can be done."08:56
Romsterif you want a big speed up and are prepared to patch a few Pkgfiles install elfutils recompile gcc then set -flto in cc cxx and ldflags08:56
Romsterthere is more flags you can use after that but then it gets risky at breaking stuff, your using visability=hidden in CXXFLAGS now right?08:57
Romsterthe more you optimise the more chance of breaking stuff past that point.08:57
Romster-flto on LDFLAGS and CCFLAGS CXXFLAGS for the majority of ports will work those that don't you have to strip them off.08:58
joe9romster " your using visability=hidden in CXXFLAGS now right?" -- yes.08:59
Romsteri did hack pkgutils in hvk to have variables in pkgmk.conf to hack flags per a port name. but it was ugly.08:59
Romsteryes08:59
Romsterit's pretty safe to do.08:59
joe9there was one port that had issues with the flags08:59
joe9and glibc had issues too.08:59
Romsterworst that can happen is some port says missing symbol at compile time.08:59
joe9other than that, everything worked/compiled fine.08:59
Romsteryeah that's not using C++ though09:00
Romsterthat's a CSSFLAGS option only09:00
RomsterCXXFLAGS sorry09:00
acruxRomster: like we do with ilenia; we can se a specific pkgmk.connf per port09:01
Romsterforces g++ to resolve symbols at linking time, longer time to compile faster to load and run c++ apps.09:01
Romsteroh you got that acrux09:01
acruxe.g. http://cruxppc.org/Ilenia09:01
acruxbtw, also cloog,ppl is a good speed bump09:01
RomsterPKGMK_WINE_CFLAGS='-O0 -g -fno-omit-frame-pointer -funit-at-a-time -Wfatal-errors'09:02
Romstereach port made it's own variable names.09:02
Romsteryeah i tired that but cloog,ppl is not very well tested yet has a few bugs.09:02
Romsterunless that's changed the past few months09:02
joe9acrux: how is ilenia different from what exherbo tries to do?09:03
acruxilenia it's a port manager (used on cruxppc)09:04
Romsternest time i try to optimize i'll patch pkgmk to use a directory for environment vars for each Pkgfile much like how ldconf.d works for libs09:04
Romsterif i ever get unlazy todo that09:04
acruxlol09:04
Romsteri got enough things to do as is.09:04
Romsterhonestly how do i do it?09:04
joe9romster, is -flto a shortform for something else.09:10
joe9I cannot seem to find much info for it.09:10
acruxhttp://gcc.gnu.org/wiki/LinkTimeOptimization09:10
joe9acrux, thanks.09:11
Romsternope but it's really really fast for firefox and stuff but not everything works with that LTO09:11
Romsteri haven't used gold linker yet not sure how mature that is yet09:12
Romsterbest way to learn is to try and expect to break your system a few times. i know i have :D09:12
joe9do I have to do anything to install the gold linker?09:13
joe9or, is it part of the gcc port.09:13
Romstereasy enuugth to keep a backup of good known built packages.09:13
Romsterit's a part of binutils you need to enable it at configure time.09:14
Romsteri haven't tested that so no experience yet with that unless acrux has already?09:14
Romsterhttp://twitpic.com/9wnuvx pizza anyone09:16
acruxno experience with gold09:17
Romsterhows your week been acrux you still working friday or tjat09:19
Romsterthat's just over?09:19
Romsterdamn typing while drinking is super bad for me.09:19
acruxnot yet, 16:1709:19
acruxend of game is 18:0009:20
Romsteroh man hanging out 18 damn that's a long day09:20
Romsteri get off at 17:3009:20
acruxbtw, after i need to study... -.-09:20
Romsteri left work got food and booze went home now drunk, poor you.09:20
acrux:D09:21
Romsteri'm only now starting to slow down my brain.... at 0:2009:21
Romstermustitasking so much all week makes it hard to relax. and mellow out.09:21
acruxagree09:22
acruxwell, coffe and cigarette time break09:22
Romsteronly way to keep my sanity is to just let go of the madness and relax, but with studying you have to keep alert.09:22
Romsterenjoy09:22
acruxRomster: uff...this year i _must_ close my degree09:23
Romsterneed good expresso coffee to get ready for studying.09:23
joe9anyone using the gold linker on crux?09:23
acruxsee you later09:23
* acrux away09:23
joe9prt-get fsearch gold does not find anything.09:23
Romsteroh i hate having to rush the stress of it.09:23
Romsterlater acrux09:23
Romsterjoe9, .... i may need to headdesk at that.09:24
horrorStruckjoe9: i used to09:24
horrorStruckyou have to rebuild binutils and gcc09:24
joe9horrorStruck: how did you install it?09:24
Romsterlook at the binutils port ./configure --help09:24
joe9horrorStruck: ok, thanks.09:24
joe9Romster: thanks.09:24
Romster--enable-gold at a guess for binutils09:25
Romsterrecompile.09:25
horrorStruckjoe9: for binutils, add --enable-gold --enable-plugins09:25
Romsterbut i'm not sure how it workls though.09:25
horrorStruckjoe9: for gcc, add the same options and --with-plugin-ld=ld.gold09:26
Romsteri've seen what M$ is doing to win8 they are butchering the visual effects09:26
RomsterhorrorStruck, conclusion of gold any quirks to be aware of and is there a fall back method if it breaks something?09:27
horrorStruckjoe9: then add  -fuse-linker-plugin flag09:27
Romsteri think i read somewhere you can choose at compile tome what linker to use.09:28
Romsterlike that one ^09:28
horrorStruckRomster: i didnt use it a lot, most ports would build with it but didnt notice any real benefit09:28
Romsterall i could imagine is resolving symbools quicker at link time.09:29
joe9Romster: that means the builds will be quicker.09:29
joe9and loading will be faster too.09:30
Romsteryes you are right09:30
Romsterit depends on how complex the port is really.09:30
Romsterthings like firefox would benefit the most.09:30
horrorStruckIIRC it fails to build09:30
joe9the -fuse-linker-plugin -- does this belong in LDFLAGS?09:31
joe9or, in the gcc configure?09:31
Romsterqt webkit anything with tons of stuff to resolve.09:31
horrorStruckjoe9: CFLAGS/CXXFLAGS09:31
joe9horrorStruck: ok, thanks.09:31
joe9horrorStruck: and -flto to CFLAGS/CXXFLAGS/LDFLAGS09:31
RomsterhorrorStruck, is it like LTO has tobe in CFLAGS/CXXFLAGS and LDFLAGSto work?09:32
Romsterjoe9, i'd try one on it's own first adding both at the same time would make it hard to know what broke something.09:33
Romsteryou really want to be a ricer don't you :D09:33
Romsteroh i forgot the most obvious speed up was too drunk to think.... now i remeber09:34
joe9i am sick of firefox being so slow. with the speed increase in firefox, it opened my eyes to what can be possible.09:34
horrorStrucki never put anything in LDFLAGS (-flto or -fuse-linker-plugin)09:34
horrorStrucki have this however export LDFLAGS="-Wl,-O1 -Wl,--hash-style=gnu -Wl,--as-needed -Wl,--sort-common"09:35
joe9horrorStruck: how do those flags help?09:35
joe9if you do not mind answering.09:35
Romsterhmm that's about the same as doing this patch http://hvlinux.net/ports/core/binutils/enable-relro-new-dtags-and-optimise-by-default.patch09:36
Romsteronly you can't rely on all ports to respect system LDFLAGS09:36
horrorStruckjoe9: O1 helps for speed (like -O2 for example)09:37
joe9horrorStruck: shouldn't that be "on" by default?09:37
Romsteronly O1 is the only optimization or linker level09:37
Romsteryeah that patch does that ^09:38
joe9romster, I could not find the patch is the binutils port directory.09:38
joe9Romster:  is that patch a part of binutils?09:38
Romstersee the url above..09:38
Romsterno it's applies to binutils09:38
Romsterit*09:39
horrorStruck--sort-common: quoting: prevent gaps between symbols due to alignment constraints, presumably increasing efficiency layout.09:39
Romsterhttp://hvlinux.net/ports/core/binutils/Pkgfile09:39
joe9 Romster: cool thanks.09:39
horrorStruck--as-needed helps to link only to needed libs09:40
joe9i do not see as-needed in the patch pointed to by romster.09:40
Romsterjoe9, i think you need to read more and test it's probably the only way to learn.09:41
horrorStruckjoe9: you don't *have* to use a patch, just edit pkgmk.conf09:41
joe9horrorStruck: i like the patch. it seems to good to have it as part of the source.09:41
joe9but, I think I would need to add  -Wl,--as-needed -Wl,--sort-common to supplement the patch.09:42
Romsterthe patch is more reliable but you can set the LDFLAGS for the majority of ports to use.09:42
Romsterhmm dunno...09:42
horrorStruckjoe9: i didnt look at the patch but it's nice to be able to dsiable some options easliy with editing pkgmk.conf09:42
joe9horrorStruck: ok, thanks. makes sense.09:42
Romsteri'm too drunk to be of help right now i'll let horrorStruck take over.09:43
horrorStrucki wish i was drunk09:43
Romsterhehe09:43
joe9romster, this is the binutils Pkgfile I have. http://codepad.org/TemRnYjW09:43
joe9romster, the Pkgfile that you pasted. where is it from?09:44
joe9HVLinux system team -- ??09:44
horrorStruckjoe9: maybe good to add --enable-ld=default09:45
Romsterjoe9, a long long time ago i got disgrunted with crux i forked and called it hvl/hvlinux09:45
joe9romster, oh, ok.09:45
Romsterhad only like 3 devs then it got too much for me to maintain everything myself.09:46
Romsteri did learn alot of optimizing then.09:46
Romsternow i'm on crux multilib09:46
Romsterseems i go for experiential stuff.09:47
frinnstbreaking things is fun09:48
Romsterindeed09:48
jse...until you realize you have to get stuff done, too.09:48
Romsteri've learned if you don't break your system you never learn how it works.09:48
Romsterjse, ltns09:49
jseThat's a great point. Sometimes you have to break it to know how it works.09:49
frinnsthehe yeah. I lack a good lab environment at home09:50
Romsterforces you to learn how it works and what went wrong.09:50
frinnstso my desktop takes most of my experimentation09:50
horrorStruckand breaking stuff also showed me that crux can always get resurrected09:51
Romsteryeah it's easy to repair09:51
joe9 horrorStruck: you do not have --enable-new-dtags09:51
joe9http://codepad.org/P1oIoOXa09:52
horrorStruckjoe9: never heard about this09:52
Romsterhmm not seen that configure option but i know about dtags09:54
joe9 http://tk-blog.blogspot.com/2009/02/relro-not-so-well-known-memory.html09:55
Romster200909:56
Romsterack i'll read that when sover.09:56
horrorStruckjoe9: you could play with graphite in gcc too but you'll need to add some new deps to gcc09:56
Romstersober*09:56
joe9what is graphite?09:57
Romsteri played with that too.09:57
Romsterhttp://gcc.gnu.org/wiki/Graphite09:57
RomsterGraphite is a framework for high-level memory optimizations using the polyhedral model.09:58
joe9Romster: thanks.09:58
Romsterit's deep stuff.09:58
horrorStrucki use it all the time, never had any build issue09:58
Romsterthis whole plugins for gcc has be concerned.09:58
horrorStruckit was introduced in gcc 4.4, see release notes http://gcc.gnu.org/gcc-4.4/changes.html09:59
Romsteri sure hope no propriety modules happen to creep in09:59
joe9horrorStruck: can you show your Pkgfile gcc.10:00
joe9horrorStruck: did you have to put in any changes there for graphite build10:00
horrorStruckjoe9: sure but i didnt bother to add the new deps10:00
horrorStruckjoe9: that's 4.6.3 but it works with crux's version too https://gist.github.com/293688510:01
joe9horrorStruck: just wanted to doublecheck. you are not using gold anymore, correct/10:02
joe9s,/,?,10:02
horrorStruckjoe9: no, not anymore10:02
horrorStruckbut i will reenable it right now10:03
joe9http://forums.gentoo.org/viewtopic-t-850087.html10:03
joe9found this article about graphit.10:04
joe9i am on a single cpu, so, not sure if graphite will help me much.10:04
horrorStruckjoe9: you'll need cloog 0.15.10, ppl 0.11.2 and some other stuff too10:04
Romsterppl10:05
horrorStruckok let's see if current firefox likes gold10:05
Romsteroh you said that something recently got changed from ppl to something else i can only vaguely remember10:05
horrorStruckme?10:06
horrorStrucki just remember i had to use specific versions or it would fail with gcc 4.5, that's too old, can't remember. maybe isl related10:07
joe9on a realtime system, I get some kernel oops once in a while about an invalid value in EIP. the reste of the machine works fine after the oops. on that machine, if I use -fomit-frame-pointer, will it crap out the kernel?10:10
joe9Romster: horrorStruck: any suggestions, please?10:11
horrorStruck-fomit-frame-pointer is known to be safe10:12
joe9ok, thanks.10:12
Romstersafe but hard to debug if you need -g10:15
Romsterwhich is usually rare10:15
joe9this is my development system. so, I might need -g on a particular app or something like that.10:16
joe9looks like it is part of -O210:18
joe9http://codepad.org/N5yVpYZm10:18
horrorStruckgold is broken with 4.6.3 on my current system, time to do some upgrade10:26
joe9horrorStruck: ok, thanks. will stick with just your LDFLAGS for the time being.10:26
joe9no gold or graphite.10:26
joe9for me, now.10:26
horrorStruckif yo uuse 4.5.y and the versions i mentioned above, it should work10:29
joe9i am on 4.5.3, the crux standard gcc version.10:30
cruxbot[opt.git/2.7]: [notify] flash-player-plugin: updated to 11.2.202.23610:40
joe9gcc -c  -g -fkeep-inline-functions -DIN_GCC   -W -Wall -Wwrite-strings -Wcast-qual -Wstrict-prototypes -Wmissing-prototypes -Wmissing-format-attribute -pedantic -Wno-long-long -Wno-variadic-macros -Wno-overlength-strings -Wold-style-definition -Wc++-compat -fno-common  -DHAVE_CONFIG_H -I. -I. -I../../gcc-4.5.3/gcc -I../../gcc-4.5.3/gcc/. -I../../gcc-4.5.3/gcc/../include -I../../gcc-4.5.3/gcc/../libcpp/include  ...10:47
joe9... -I../../gcc-4.5.3/gcc/../libdecnumber -I../../gcc-4.5.3/gcc/../libdecnumber/bid -I../libdecnumber   -I/usr/include/libelf  ../../gcc-4.5.3/gcc/dce.c -o dce.o10:47
joe9i notice that gcc build not use the CFLAGS or CXXFLAGS or LDFLAGS from pkgmk.conf10:47
joe9i am talking about the standard  /usr/ports/core/gcc build.10:48
joe9horrorStruck: i notice that your Pkgfile does not do anything different w.r.t the CFLAGS or CXXFLAGS or LDFLAGS10:50
joe9horrorStruck: does the build still use the flags from pkgmk.conf?10:51
*** lasso|qt has quit IRC10:53
horrorStruckjoe9: not anytime during the build, i guess gcc devs did that on purpose11:21
joe9ok, thanks.11:21
horrorStruckjust a guess11:21
horrorStrucki may be totally wrong of course11:21
horrorStruck:P11:21
horrorStruckjust fucked up my toolchain thanks to you :D11:22
joe9oh, no. sorry about that.11:22
horrorStruckjoking11:22
*** joacim has quit IRC11:39
joe9horrorStruck: were you able to fix your toolchain?12:18
joe9horrorStruck: were you able to compile firefox with gold?12:53
horrorStruckactually i really fucked it :P still building in a chroot now13:03
joe9ever tried -ffast-math?13:07
horrorStruckno, apart from i mentioned earlier, i use -fexcess-precision=fast and that's all13:11
*** hcaulfield57 has joined #crux13:20
hcaulfield57I can't seem to figure out how to add another repository for ports, I've added the line in prt-get.conf and added the rsync file under /usr/ports and ran ports -u but nothing happens13:21
pitillohcaulfield57: rsync file should be in /etc/ports/13:22
joe9horrorStruck: quick question, when building gcc, do I need make benchmark, when I have gcc already installed on my system?13:22
jse/usr/ports only contains the actual ports. You want /etc/ports instead.13:22
hcaulfield57pitillo: Oh stupid me, thank you so much. Thanks too jse13:23
jse:)13:23
joe9sorry, bootstrap13:23
joe9not benchmark13:23
pitillohcaulfield57: np :)13:23
hcaulfield57That's what the handbook says too, I guess I just saw /usr/ports by default haha13:23
joe9horrorStruck: checkout these flags, they seem to help too  -fweb -frename-registers -ftracer -funroll-loops13:26
horrorStruckjoe9: normally no, you have to rebuild libtool13:26
*** hcaulfield57 has quit IRC13:28
horrorStruckjoe9: thanks, but now going back to mostly default as i screwed up something and cant find what so far13:29
joe9do not use -funroll-loops, it breaks binutils port.13:30
joe9is there a file that stores the list of installed ports on a system?13:57
horrorStruckprt-get listinst > file13:59
joe9prt-get quickdep $( pkginfo -i | cut -d ' ' -f 1 ) | tr ' ' '\n' \14:01
joe9is what I am currently using.14:01
joe9not sure if there was anything better...14:02
joe9prt-get quickdep $(prt-get listinst) -- little less parsing, based on your idea -- horrorStruck14:03
joe9i want to put the list of ports installed on a system in version control. I could do the above, pipe it to a file and version it.14:06
joe9or, I was wondering if there was such a file already.14:06
joe9does anyone store the /var/pkgmk/packages/ in version control. not sure if that might be too much?14:06
*** lasso has joined #crux14:22
*** lasso has quit IRC14:36
*** mike_k has quit IRC15:06
*** horrorSt1uck has joined #crux15:07
*** horrorStruck has quit IRC15:10
*** jdolan has joined #crux15:51
*** ChanServ sets mode: +o jdolan15:51
cruxbot[opt.git/2.7]: emacs: strip {c,e}tags from port16:03
frinnstfuck england :(16:09
frinnstjoe9: check /var/lib/pkg/db16:12
joe9frinnst: thanks.16:12
frinnstaaaw man, now i'm depressed16:14
tilmanfrinnst: why does your country bother with football?16:19
frinnstoh fuck off!16:23
frinnst!"&ยค!")#16:23
frinnst:)16:23
rmulltilman: Came out of retirement to say that? :P16:23
tilmanreally not sure what's ambigous about my email16:24
tilmanbut w/e16:24
rmullI'm just joking around, I'm too new to know anything about you. No offense I hope16:24
tilmanfrinnst: didnt mean to tease/insult really. kinda depressing to see them do bad/mediocre in every tourney though16:25
frinnsthahahhahah fuuuuuuuuuuuuuuuu16:25
tilmanuh-oh16:25
frinnst:D16:25
tilmani see i'm not helping my case16:25
tilman;)16:25
frinnstbronze -94!16:25
frinnstwe still live on that16:26
tilmanfrinnst: wow, didnt know ;)16:27
*** vaddi has quit IRC17:00
*** vee has joined #crux17:20
*** vee has quit IRC17:25
*** Rotwang has joined #crux17:51
*** vee has joined #crux18:08
*** Rotwang has quit IRC18:22
*** tilman has quit IRC18:35
*** tilman has joined #crux18:36
*** ChanServ sets mode: +o tilman18:36
veeanyone on a laptop thats running ip tables wanna share your settings? too lazy to do so xD18:48
veenot sure how safe this is... http://easyfwgen.morizot.net/gen/18:54
jaegerIt just generates a script, you can go through the script after it's generated and figure out what it does18:55
frinnstvee: i recognize that site. i've used it way back and it's pretty basic but sane19:23
Romsterreminds me i need to clean up my firewall rules one day.20:11
*** joacim has joined #crux21:30
*** mavrick61 has quit IRC21:48
*** mavrick61 has joined #crux21:50
joe9need some help, please: before the login prompt comes up, my system goes into a reboot automatically22:29
joe9I can see the "Adding swap.." message come up.22:29
joe9without the init, I can do swapon and the swap devices work fine.22:29
joe9is there anything that I could be missing.22:29
joe9I had some customization in /etc/rc.local, but, I removed that too.22:29
joe9i am not using startx or X. just the plain console prompt.22:30
joe9it reboots at the point just before the console login prompt shows up after the kernel boot messages are cleared. it gives a small machine tick noise and reboots.22:31
joe9i tried with different kernels, some, old ones which were working before. I do not think it is the kernel issue. I  used the rescue shell of initramfs to mount the drives and all that.22:32
joe9when I do the switch_root to /sbin/init, it does everything else, but just after the "Adding swap..." reboots.22:32
joe9nothing new in /var/log either. even the /var/log/kernel file is not being updated.22:33
joe9only the /var/log/wtmp file is getting updated with the timestamp of the reboot.22:33
joe9removed the services from rc.conf except for net crond and sshd22:34
rmulljoe9: Yikes, how did this begin to happen?22:35
joe9i rebuilt a kernel and made some /etc changes. not sure what caused it to happen though.22:36
joe9is there anyway I could go through the list of things that init does manually?22:36
rmullSometimes you can run an interactive init22:38
rmullI've seen it in other distros22:38
rmullI don't know how to trigger it or if it's supported in crux though22:38
rmullgood luck, off to bed22:43
*** joe9 has quit IRC22:43
veeessentialy gunna use that, and block access to chinese and russian sites. sorry mother russia23:45
*** vee has quit IRC23:53
*** joe9 has joined #crux23:59
joe9with startx and latest libdrm, I get failed to load module nouveau.23:59
joe9with startx and latest libdrm, I get failed to load module nouveau. I can see that libdrm created the nouveau module as libdrm_nouveau.so. Is there anthing i should do to get  startx see that module?23:59

Generated by irclog2html.py 2.11.0 by Marius Gedminas - find it at mg.pov.lt!