IRC Logs for #crux-devel Wednesday, 2012-06-27

jueRomster: could you please explain why cmake depends on shared-mime-info?01:29
jueRomster: cmake builds and runs just fine for me without having shared-mime-info installed01:31
*** sepen has joined #crux-devel01:32
juesepen: you've read my comment wrt cmake/shared-mime-info?01:40
sepenjue: no, yesterday?01:57
*** jue_ has joined #crux-devel02:05
*** jue has quit IRC02:06
*** Romster has quit IRC02:06
sepenjue_: on 24Jun?02:12
jue_sepen: no, today, some minutes ago02:18
sepenoops, can't find that in the backlog for today02:19
sepenmaybe a freenode issue? or log02:19
*** Romster has joined #crux-devel02:19
sepenahh to Romster02:19
jue_yeah, there was a netsplit, it's here
jue_yes, it was added through his patch02:21
sepenjue_: my bad, I trusted in Romster changes02:21
sepenhmm, I didn't notice because I've it installed02:27
jue_maybe I've overlooked something, let's wait for him02:31
sepenjue_: 94% here02:35
sepenand now ... build result ok02:35
sepenmaybe he refers to runtime deps02:36
sepencoffee time02:41
jue_same here :)02:41
Romsteri just got home but i got tons of stuff todo whats up?02:46
Romster i got the deps from here.02:48
Romsterdepends=('curl' 'libarchive' 'shared-mime-info')02:48
jue_oops, we shouldn't blindly trust in arch IMO ;)03:03
Romsteri figured they should know what there doing03:04
jue_dunno, but it's our responibility03:07
jue_and obviously, the dep looks a bit strange to me03:09
Romsterafk sorry busy03:10
sepenpfff I adjust the deps again03:20
sepencurl libarchive expat ncurses03:26
*** jue_ is now known as jue03:29
pitilloyes frinnst, 6e the power supply03:32
*** mike_k_ has joined #crux-devel03:48
Romsternothing is going right i don't have any feeler gauges neather has my friend so i need to buy some before i can install these spark plugs, i don't don't have a long enough extension for my socket set to remove a battery in another car.... *grumbles*04:20
Romsterso i got pizza04:20
Romsternow i need to test my ports that use cmake again...04:21
Romsterjue, what happened to jur/rc port i'm still using that on my system now i have no source to that. mind emailing me a copy of that? or you got some pubic git repo that i can grab it from04:23
mike_k_do we have some strategy on moving config files on port updates? old would just be deleted without a prompt (04:50
Romsternot if there in pkgadd.conf to not overwrite04:56
Romsterbut your talking on moving..04:57
mike_k_I have no good idea, but sending notify email on the ML04:59
mike_k_or updating on a next CRUX release and including such a note to release notes05:00
Romsterwhat port and what file from where to where?05:01
mike_k_I am talking about opt/quota port. it's sources do not support changing config file location with -DBLAH_BLAH_CONFIG_PATH=05:01
mike_k_I don't remember the exact reason to put them in /etc/quota back in that time05:01
Romsternot everything is configurble in cmake, depends on the cmakelists on who writes it.05:01
Romsterif it's a daemon it should be in /etc/05:02
Romsteranything else not a daemon or in core goes in /usr/etc/05:02
mike_k_./configure itself is ok. just the Makefile and every .c file do not support such a change05:02
mike_k_they search in /etc, which is kindof hardcoded05:03
mike_k_jue: any thoughts?05:03
mike_k_I am about to prepare a diff05:03
Romstersymlink the directory to the location it searches for?05:04
Romsterif that'll work.05:04
Romsterif it's hard coded in one location in some define it'll be easy to sed it before compiling mike_k_05:05
mike_k_it lookes exactly for /etc/warnquota.conf, but we put it in /etc/quota/warnquota.conf. I don't want three more silly symlinks in etc, nor to patch or sed the sources05:06
Romstersome projects even do it in CFLAGS="-DFOOCONFIGPATH='/etc/foo'"05:06
mike_k_every guide on the net expects them in /etc05:06
Romsterso move it but you'll need to warn them to move there conf file.05:07
mike_k_Romster: yep, but it's sources do not check like: #ifndef. they just #define05:07
Romstersed the #define in the source then.05:07
mike_k_I'd prefer to move them to /etc actually )05:07
Romsterjust do that and and, test if the config file in /etc/quota/warnquota.conf doens't get removed when you install to /etc/warnquota.conf on the same port name, i'm pretty sure it'll keep the old conf around.05:09
Romsterbut wouldn't hurt to test.05:09
mike_k_Romster: I am pretty sure old configs would get _removed_05:11
mike_k_as they are not overwritten by anything05:11
mike_k_we can supply dummy files to force someone a rejmerge review, but that is odd and does not guarantee anything05:12
Romstertry testing with a dummy Pkgfile port touch a few files echo some contents.05:13
Romsterhmm now seems a good time to reboot on newer kernel and nvidia at the same time.05:14
mike_k_prooflink, if someone was not sure:
frinnstdtach cve05:30
frinnstoh, i thought it was a opt/contrib port. guess not05:31
Romstermike_k_, i guess touch the old old locations until next release so then those files don't get removed. so users get the chance to mv them.05:55
mike_k_Romster: yes, that could save them some time. not sure if we did that before. anyway notify email should inform those, who care06:00
jaegerWhat's the result of the shared-mime-info conversation?07:30
Romsterapparently redundant dependency07:46
jaegerah, just noticed the cmake commit07:47
jaegerI guess I should have waited for that discussion before building another updated ISO07:47
Romsterwont affect the built package i guess07:49
jaegerI don't think it will in the case of shared-mime-info but on the other hand I should add parted to opt and update the other new ports anyway08:00
jaegerpciutils and xterm got updated08:00
sepenjaeger: what do you think about FS#69408:12
jaegersepen: I have no objections to that. I would like to remove grub and add syslinux and grub2 but I didn't comment there as it's already marked as due in 2.808:15
sepenah ok, 2.808:16
jaegerwe're still generally leaving core ports out of Depends on:, right?08:22
Romsterjaeger, there is some other deps in xorg that need fixing too.08:23
Romsterwant me to pastebin what i've found so you can fix them?08:24
jaegersure, I can take a look08:24
sepenjaeger: irrc the rule is that deps from core don't require to be listed08:25
jaegerThat's what I was thinking, just wanted to make sure since it's been a long time since I worked in opt or core much08:25
Romsteri always thought that deps in core should only be listed if it's not the toolchain gccglibc binutils but listed if it builds agenst it like zlib if it uses that and ncurses etc.08:28
Romsterbut nothing like perl as that's not a built in dep. in so files.08:28
Romsteror binaries08:28
*** acrux has quit IRC08:29
*** acrux has joined #crux-devel08:30
*** acrux has joined #crux-devel08:30
Romsterjaeger, i haven't done alot of testing on what fixes those.08:31
Romstersince i've got that all sorted in the updated xorg R7.708:31
Romsterthat patch for gccmakedep fixes if you set say CC='ccache gcc' which i kept hitting.08:32
jaegersome of those ports should be removed08:33
jaegerfor example there's no reason for us to have any sun video drivers in crux08:33
jaegerThe list is bigger than that but I can't reach currently08:36
sepenimho I always prefered to list all deps (like we did in crux-arm), or at least is a way to help other stuff like prt-get or even setup.dependencies for the iso08:41
sepenjaeger: could you take a look of FS#83408:42
jaegerI've seen it08:44
jaegerIt's on my todo list but have been troubleshooting ISO stuff08:44
sepenare you the man doing massive commits to xfonts? ;D08:44
jaegerI did some recently, yes08:45
sepenFS#791 is related too, see acrux's comments08:46
jaegeryeah, saw that as well08:47
jaegerI don't see any reason to put the font packages in opt but I did suggest standardizing the name if you recall08:47
juejaeger: wrt core deps, see
jaegerah, didn't think to look in the wiki, was looking for it in the handbook08:50
*** horrorSt1uck has quit IRC08:51
jaegerI'll have to add a dependencies line for parted, then08:51
Romsterif it's in finddeps and isn't glibc gcc binutils it should be listed on depends on line IMO.08:52
juebasically you have to list all core libs, but no programs like sed, awk and the like08:52
*** horrorStruck has joined #crux-devel08:52
jueexcept gcc an dglibc08:52
Romsternothing if it's not linking to it.08:52
sepenjue: see automake deps: gawk, perl, autoconf08:53
juethat's not necessary08:53
Romsteri don't think core ports follow the same rules?08:53
jaegerin the case of parted it'll be libdevmapper, readline, udev, util-linux-ng08:53
sepenyeah, just I noted that is not 100% applied08:54
sepenjaeger: I built parted without udev installed (on my mdev box)08:57
jueI think our PortGuidelines is quite clear and applies to all ports08:57
jaegerThat's fine but on a normal crux install it will link to udev08:57
sepenah ok08:58
jueRomster: indeed, finddeps is a nice tool for finding deps08:58
Romsteri use that all the time :)08:59
sepennot for cmake lol08:59
* Romster facepalms08:59
Romsternope i didn't there.08:59
Romsteri wont rely on arch ever again...08:59
sepenjaeger: noticed that?
jaegeryes, though it didn't seem to cause any problems that I could see09:01
juebtw, the reason why I didn't do the update to nvidia unil now isn't laziness or something like that, but because I got a really bad 3D performance with it.09:01
sepenok, just I build the package but I don't try it so far09:02
jaegersepen: I've been using it for a while on the multilib ISO due to the need for GPT/EFI support, no problems so far09:02
jaegerjue: with which chipset?09:02
jueat least the values with glxgears are around a tenth of the old version09:02
sepenjaeger: I'll try it on my efikamx (crux-arm) ;)09:03
jaegeron my quadro fx 580 (G96) I get much better performance in diablo 3 in wine than I did with the previous driver09:03
jueit's a old GeForce 660009:03
jaegerhrmm, interesting09:03
Romsterhaven't tested new video driver in wine yet09:04
jueI'm talking about 295.59 resp. 302.1709:05
jaegerI've never really considered glxgears a consistent test so I didn't think to compare the two09:05
jueyeah, indeed, but I have nothing better ;)09:06
jaegerfair enough09:06
jaegerI wonder if the phoronix test suite or unigine would be good benchmarks09:07
jaegerperhaps it doesn't work with cards that don't support DX11, though, not sure09:09
jaegerI'll test it on my quadro, it's a dx10 card09:10
Romsteris that a test suite or a game O_O09:13
jaegerIt's a game engine but the heaven part is a benchmark using it09:15
jaegerIt does work on my quadro fx 580, it just uses opengl09:15
jaeger4-5 FPS, heh09:15
sepenI considered quake3 or true-combat-elite as the best performance test ever ;D09:17
juejaeger: .oO you forgot the deps in parted09:19
jaegerI did but they're being pushed right now09:20
juesorry, I was a bit too fast :)09:21
jaegerNo worries09:21
*** joe9 has joined #crux-devel09:21
jueRomster: rc-ng ->
Romsterso you've ditched it in favor of what we got now i take it.09:27
Romsterwasn't someone working on rc-ng?09:27
juemike_k_: to get it right, our --sysconfdir=/etc/quota doesn't work in practice?09:28
sepenRomster: read the ML about that09:28
sepenMichal Soltys is the guy behind rc-ng
sepenalso I pointed out some ideas and modifications
jueRomster: the reactions I got are all more or less negative, so I thought it makes no sense to use it any longer09:31
Romsteryeah i read all that ages ago sepen but jsut recently jue removed it outof his collection.09:31
Romsteri thought the good bits were going ot be pulled out to put in the current rc?09:31
Romsterwasn't mike_k_ or someone assigned to that?09:31
jueit's teK_09:32
Romsterok teK_09:32
sepenrelated to brtf09:34
sepenofftopic: I'm really comfortable with mdev so far09:37
Romsteri'm not game to try it on my mdadm/lvm setup yet.09:41
*** joe9 has quit IRC10:45
*** joe9 has joined #crux-devel11:03
*** sepen has quit IRC11:13
jueFWIW, ... the release of glibc 2.16 will be made on the first week of July11:22
jaegerI will be out of town from the 4th to 9th of july, also11:24
mike_k_jue: yes, --sysconfdir=/etc/quota does not work except it places configs there. nor .c sources support a way of overriding them with "-DSOME_PATH=foo"11:26
jueme too next week, but from 2th to 7th11:26
mike_k_I'd just use /etc and notify a ML11:26
juemike_k_: ok, thanks11:27
mike_k_jue: I'll post the diff(s) now11:27
mike_k_man-pages needs to be updated too11:27
mike_k_* as I've updated quota to 4.00 and it does not provide quotaqtl man page (which we remove from core/man-pages)11:29
jueno problem to leave it in man-pages11:30
*** mike_k_ is now known as mike_k11:31
mike_kI've added quotagrpadmins, as it had soem dummy rule. and adjusted the warnquota.conf's path to sendmail to work out of the box11:32
juemike_k: thanks, will commit the stuff later11:33
mike_kvery pleased, how it works with msmtp, aliases and remote SMTP server, btw11:33
mike_kjue: thanks a lot11:34
*** sepen has joined #crux-devel12:42
*** c0x has quit IRC13:05
*** horrorStruck has quit IRC15:39
*** horrorStruck has joined #crux-devel15:41
*** mike_k has quit IRC17:08
jaegerreceived the CPU for my ESXi server a day early :P weird18:12
*** joe9 has quit IRC18:52
*** joe9 has joined #crux-devel18:59
*** cvibo has joined #crux-devel20:21
*** cvibo has left #crux-devel20:22
*** sepen has quit IRC20:35
*** c0x has joined #crux-devel21:47
*** _____________ma has quit IRC21:48
*** _____________ma has joined #crux-devel21:49
*** horrorStruck has quit IRC23:04
*** horrorStruck has joined #crux-devel23:09
*** deus_ex has quit IRC23:14
*** horrorStruck has quit IRC23:22
*** horrorStruck has joined #crux-devel23:27

Generated by 2.11.0 by Marius Gedminas - find it at!