IRC Logs for #crux-devel Tuesday, 2008-07-29

*** clb has joined #crux-devel13:09
*** clbot has quit IRC13:10
cptnI'm talking to Matt about it13:12
cptnmeanwhile, we can move on13:12
cptnmandir -> /usr/share/man13:12
*** thrice` has joined #crux-devel13:12
cptnthere was a discussion on that on the mailing list13:12
cptnI linked it from the Wiki13:12
cptnit appears no one was really in favour of keeping /usr/man13:12
jueTBH I like /usr/man more, but /usr/share/man is ok for me as well13:13
sepencontrary we need to rebuild a lot of ports13:13
aoni like /usr/man too13:13
cptnyeah, I like it more too13:13
cptnbut I hate patching all kinds of make files13:13
aonwhich one is the majority as default?13:14
cptnautoconf stuff is /usr/share13:14
aonnot all of it, though13:14
sepenas tilman said I think it not good to have a diferent autotools stuff than other distros13:15
sepenso /usr/share/man is the default in autoconf13:15
juehmm, we don't have13:15
sepenit was patched some time ago13:15
aonwhat was?13:15
namenlosabout patching makefiles: most of the distros install applications into /usr/local (just as an example13:15
juebut we have to put mandir=/usr/man often in our Pkfiles13:15
cptnnamenlos: they do?13:16
cptnwhich ones?13:16
namenlosas far as i know13:16
cptn"With Linux distributions like Red Hat, Debian, etc., a possible rule is that /usr is managed by the distribution's package system and /usr/local is not. "13:17
namenlossorry, i was mistaken:
namenlossorry, sorry.13:18
cptnbut it would be interesting what other distros do WRT to mandir13:18
cptnback in the days, I thought it would be best to have only /usr/man or /usr/share/man13:18
cptnbut nowadays I'm not sure13:18
mike_kbut what to do with all ports explicitly putting man to /usr/man? bumping and fixing all of them will take some time (even more for not autotools based ones). And a plenty of private ports would have them at the old location...13:18
aonyes, it's best to have only one of them13:18
cptnI actually prefer cleaner Packages13:18
aonsoon we'll have nls to drop --disable-nls from the pkgfiles ;)13:19
juemike_k: yeah, we can do such a change only at release boundary IMO13:20
aonbut fhs says share/man so i guess we should do that, unless there's some actual reason to keep it in /usr/man13:20
mike_kit could be possible to force pkgmk (stealth) moving man pages to /usr/share (breaking footprints =) )13:20
aonmike_k: yuck13:20
sepenI think we can do an effort for the 2.5 release13:21
treachmake everyone use prtverify, and make it tell everyone that /usr/man is "FATAL" or something. :>13:21
sepenits easy to cleanup Pkgfiles in this sense13:21
mike_kjue: I'd still like to have possiblity to cleany update from iso though. It would force bumping iso packages $release's a bit =)13:22
jueit's a big pile of useless work, but yes if we wnat that ....13:22
cptnhere's some numbers:13:22
cptnin opt, 225 ports install to /usr/man13:22
cptn90 have mandir13:22
sepenERROR gnome/ekiga .............. directory not allowed: usr/share/man/13:23
sepenthe only one from core, opt ,xorg ,contrib, gnome13:23
cptnIOW we'd probably have to patch more ports if we move to /usr/share/man13:24
aonthought so13:24
cptncore: 90/3913:24
aonso that's one nail in fhs's coffin ;)13:25
cptncontrib: 251/7813:25
cptnprobably makes sense to defer the change then13:25
sepenalso we can keep /usr/man and also symlink /usr/share/man -> /usr/man13:25
cptnor we could allow both13:26
cptnbut that's somewhat ugly13:26
aoni don't like allowing both at all13:26
sepenI don't mean to allow both of them, only links13:27
cptnmy numbers where probably inaccurate BTW :-)13:27
cptnrerunning the checks13:27
namenloswhy not allowing it for a certain time, so that all people can adjust their ports13:27
teKmost important is, that man foo works. no matter where those file is located?13:27
namenlosand pkgmk would rise a warning.13:27
namenlosi mean both directories13:28
aonnamenlos: imo it's more sane to do it so that e.g. 2.4 has /usr/man and 2.5 has /usr/share man, if we decide to switch13:28
aonthat's what branches are for13:28
sepen+1 to aon13:28
treachsounds like a lot of work, but broken down on a / maintainer basis, how bad is it?13:28
teKand what would be won?13:29
sepenwe can start 2.5 branches for all git port repos13:29
treachFHS compliance. I'm sure there's a sticker somewhere for it. ;)13:29
sepenand people could have time for patching their pkgfiles13:29
jueIMO we gain nothing from that13:30
cptnokay, ran the check again13:30
aoni don't think we need to do if yet13:30
cptnin every repository, we have more "prefix" changes than mandir13:30
aon_if_ the numbers show that more would need to be patched :)13:30
cptnwhich settles it for me13:30
teKtreach: If I find one thing, that's not compliant, will it be changed, no matter how hard it will be?13:30
cptnlet's keep /usr/man as long as /usr/share is the minority13:30
mike_kif you ever decide to revert that... pkgmk already compress pages, moving them could be done in background. only footprint update is needed.13:31
mike_kway better than editing all Pkgfiles around13:31
sepenyeah bad idea, pkgmk could break your port system13:32
mike_kpoor me =)13:33
aonpkgmk shouldn't be the place for such ad-hoc hacks13:33
cptnso, next issue: the git ports driver13:34
cptnnamenlos posted that a while ago, and I did some hacking13:35
cptnI think it would work fairly well13:35
cptnbut I know no one except namelos who uses it13:35
namenlosi don't use it atm. i only thought it would be neat and save people hosting their repositories in git some tiem.13:35
sepenI can do some test in the next days I think13:35
namenlosbut i would like to use it (maybe others, too, who don't have some space to host their repos...)13:36
aonyou don't need space to have a git repo? :)13:37
aonah right13:37
namenlosmine. e.g.:
cptnit would also allow use to avoid the git -> rsync export13:38
cptnand since git repos can be checked out via http, the httpup exports13:38
cptnthat said, I don't know if it's possible to do partial checkouts over http13:38
juebut the user has the full history on his disk ..13:39
namenlosno, he can clone with --depth 113:39
namenlosand then he only checks out the diffs13:39
jueok, but that's not in your driver13:39
namenlosit is afaik already implemented that way.13:39
cptnI added a couple of these things13:40
cptnand we discussed about other optimizations13:40
cptnbut there was no real interest from anyone else, so I stopped13:40
cptnmaybe we could actually do a poll on crux@ for one :-)13:41
cptnso see whether our users would like to move their repos to git13:41
sepenI suggest to write a wiki document like or
sepento howtogitrepo13:42
jueavoiding the git -> rsync exports sounds like a good thing for me, so isn't the question what we want to do on
namenlosthe problem is the root question: do we want to have a git driver included in the distro.13:43
sepenwhy not?13:43
cptnthere might be locking issues13:43
cptnI'm not sure if i.e. a commit to git would lock an update13:44
cptnI'd expect not, but I haven't tested13:44
aonthat would seem silly13:44
cptnI think it's what subversion did earlier13:45
cptnanyway, I guess people are more interested than I thought13:45
cptnso maybe we can prepare an updated driver13:45
cptnand then we can test it for a while13:45
cptnand when it pays off we can replace rsync with git for 2.513:45
juesound like a plan :-)13:45
namenlosi like it13:46
cptnregarding gmane:13:46
cptn(posting to crux@ via gmane)13:47
cptnI contacted the gmane guys13:47
cptnand got an answer that posting should now work13:47
cptnI haven't tried myself, but will ask Colin on crux@ to do some testing13:47
cptnif you see something suspicious please let me know :-)13:47
aonshould it be tested? :)13:50
cptnI think so13:50
cptndo you have an nntp client handy? :-)13:51
cptnso, according to my flyspray search we have no new unassigned bugs13:55
cptnI guess that's it for the regular items then13:55
aondidn't work13:56
aonArticle rejected: 441 gmane.linux.distributions.crux.general is unidirectional13:56
cptnokay, only got the mail today13:57
cptnmaybe it'll be updated in a nightly cron13:57
cptnthanks for testing13:57
aonyeah, could be13:57
aoni'll test it tomorrow :)13:57
cptn also says "posting is't allowed"13:57
cptnany urgend issues we didn't discuss yet?14:01
sepenfor your holidays14:01
cptnheh, I like the right one14:02
sepenwell DE discussion14:02
cptnwe could consider opening a cafepress store or something14:02
sepenbut I think should be in the future14:02
cptnbut I won't do it14:02
* treach pictures some lardass in the t-shirt to the right. :>14:02
cptnwow, I really like that logo on the right14:02
treachyeah. looks good :)14:03
cptnokay, gonna have dinner then, ttyl :-)14:05
*** Rotwang has joined #crux-devel14:08
aoni don't want a shirt with "i686-optimized" on it :)14:09
aonotherwise they're pretty nice14:10
mike_kI've put this together inspired by
mike_kjust to know that the only sane online shop I found can't print transparent images on black =(14:11
aonthat, otoh, doesn't have much to do with (#)crux14:12
* mike_k kicks himself out of the irc14:14
*** mike_k has quit IRC14:14
* rehabdoll slaps tilman 14:23
rehabdollgit perms :>14:23
*** j^2 has quit IRC14:31
nipuLcaffeination sequence complete14:34
nipuLone thing i noticed missng from the agenda, the contrib deps discussion on -devel14:36
nipuLone problem i see with the DE specific ports issue is the situation where someone might want more than one DE on their system14:45
jueIMO we should first of all check how big the problem really is, meaning which ports are dups within the DE repos or to opt, and second is it possible to unify these dups. If this is solved it's easy to move them all into contrib14:59
jueanyway, have to leave now, night all15:00
thrice`night jue :-)15:00
nipuLheh, seems the meeting was over before i even woke up15:01
*** thrice` has left #crux-devel15:19
Rotwangweekly meetings are probably too casual15:37
Rotwangmaybe bimonthly would be better ;]15:37
treachworkload isn't constant.15:38
treachsometimes there could be material for (at least) two meetings a week...15:39
treachand sometimes some people have to go up really early for nothing. :>15:39
treachmea culpa, sorry for the swenglish. :>15:43
RotwangnipuL: in my countr there is a saying15:45
Rotwangwell I can't translate it15:46
nipuL"but you get the point"15:46
treachthere's a saying in polish.15:46
Rotwangbut the moral is that it's basically good thing to get up early15:46
treachinteresting. :>15:46
treach"the early bird catches the worm"15:46
nipuLi was about to suggest that15:46
Rotwangsomething like that ;]15:46
treachnot that I'd want a worm though. :P15:47
nipuLtequilla worm?15:47
Rotwangnor you are a bird15:47
Rotwangsorry for my ponglish15:47
treachin swedish we say "morning time have gold in mouth" or somthing similar. Just as silly. :>15:47
nipuLnot many birds catching worms at 5am here15:47
nipuLespecially in the middle of winter15:48
treachI'm sitting here with a fan directed at me. :>15:48
nipuLthe birds are lookgin at me thinking "go back to bed you fool!"15:48
treachit's 28 C inside.15:48
nipuLi grew up in the desert, i hate winter here15:49
Rotwangim sweating all day15:49
treachit's a bit too hot for me. especially since the air has been a bit humid today.15:49
Rotwangonly fan in this flat is directed at my dog :{15:49
nipuLit doesn't snow much, but it's often much colder than in the places where it does15:50
nipuLbut we get frost so thick it might as well have snowed15:50
treachwell, deserts should at least have dry air.15:51
nipuLid' rather it be humid15:51
treachhumidity is the worst. :/15:51
treachoh, no, you don't..15:51
nipuLtry living in a salt flat15:51
nipuLthere's dry, then there's DRY15:51
treachcold is much, much worse when the air is humid than when it's dry.15:51
treachsame with heat.15:52
treachok, I don't know anything about salt deserts. :)15:52
treachbut generally speaking15:52
nipuLover 50C in summer15:52
treachsounds like a bit too much. :)15:53
nipuLit sends most people mad15:53
treachI don't think it has ever been that hot here. -50 have been recorded though. :>15:54
Rotwanghuh, it's nothing for russians!@15:55
treachwell, good for us that we don't live there.15:56
*** treach has quit IRC17:01
*** namenlos has quit IRC17:04
*** Rotwang has quit IRC17:27
*** pedja has joined #crux-devel19:24
*** fetid has joined #crux-devel20:31

Generated by 2.11.0 by Marius Gedminas - find it at!