IRC Logs for #crux-devel Thursday, 2007-03-01

*** pitillo has joined #crux-devel00:07
*** Romster has quit IRC00:53
*** Romster has joined #crux-devel01:00
*** Roomster has joined #crux-devel02:21
*** Romster has quit IRC02:37
*** sip has joined #crux-devel06:04
*** jaeger has joined #crux-devel08:19
*** Roomster has quit IRC08:20
jaegersip: got your email, checking the deps now08:22
*** Romster has joined #crux-devel08:24
jaegersip: deps updated for glitz, gtk, and pango - look ok?08:35
sipoh, finally found the source of that sound :)08:36
sipdidn't remember I had an open irc client somewhere08:36
* jaeger grumbles at buildroot08:38
siphow's the livecd going?08:39
jaegermostly fine, busybox is the problem at the moment08:39
jaegerrecent versions warn against building it statically with glibc, so I'm attempting to build a static uClibc busybox instead08:40
jaegerwhich isn't going too well so far... the first attempt segfaults08:40
jaegertrying again with some different settings08:40
jaegerif all else fails I'll put a dynamic one in the initramfs and move on, but I'd prefer not to08:40
sipwell, having to maintain a uclibc environment is a pita for sure08:41
jaegerI'm letting the buildroot maintainers do it for me, since it won't need to be updated often, I would think08:42
jaegerthe only reason I've updated it at all, in fact, is that the switch_root and pass to new init in 1.1.3 seems to have problems08:42
sipdid the same with a crux lookalike I've been working on recently08:42
jaegerthought it might be fixed in 1.4.1 since there were a lot of bugs fixed08:42
jaegerdid you use switch_root, by chance?08:43
sipplain old initramfs, dedicated partition for config files08:43
sipusb stick, for what that matters08:44
jaegerthe problem I have is that directly after switch_root, the kernel panics with the typical "attempted to kill init" message08:45
jaegerwhich I'm having trouble figuring out since the init in the new root should be perfectly fine08:45
sipgtk deps looks fine to me08:46
jaegerok, good08:47
sipbulding the iso...08:48
jaegeryay, shark!08:50
sipluckily recompilation is not needed. my notebook is melting08:51
jaegerhrmm... the new busybox segfaults as well09:00
sipon a buildroot setup?09:00
jaegerI haven't tried to boot the buildroot fs itself but it came from buildroot, yeah09:01
jaegeris there some reason a statically linked uClibc executable wouldn't run on a normal glibc system?09:01
sipit's the other way around09:02
siplemme check my version09:03
sipbb 1.2.2 uclibc 0.9.8 works here09:05
jaegerdo you have a static bb executable already?09:06
jaegermind sending me a copy to test?09:06
sipI can send you a ext2 compressed fs09:06
jaegerthat would be fine09:06
sipI remember having one with bb 1.4.x but cannot find it now09:10
jaeger <-- this mean anything to you?09:15
sipugh, no09:15
jaegerfigured it wouldn't hurt to ask :)09:16
*** j^2 has joined #crux-devel09:17
j^2hey all09:18
siphey j^209:18
j^2how's it going sip?09:18
sipeverything's fine, thanks. and you?09:18
aonhi j^209:18
aonand others, too :)09:19
j^2so fraking excited...i'm getting the insulin pump ;)09:19
j^2yo aon09:19
j^2hey can anyone see...09:19
j^ ?09:19
aonsounds scary09:19
sipit may contain random crap09:20
aonj^2: seems ok09:20
j^2aon, what about09:20
j^ ?09:20
jaegersip: thanks09:20
sipnp. I suspect a newer version is buried somewhere at the office09:21
aonmy notebook's value is $0.0009:21
j^2aon you can see bestbuy?09:21
aoni didn't check yet09:21
aonbestbuy doesn't seem to open09:22
sipbestbuy does not open here09:22
j^2what about09:22
aonresolves ok, though09:22
j^2(getting a pattern eh? :P)09:22
aonthat xdoesn't work either09:22
j^2stupid iis09:23
j^2i dont understand it for the life of me09:23
jaegersip: which version of gcc did you build in that setup?09:24
sipheh, you're asking too much09:24
sipvery likely 4.0.209:25
jaegerwish I understood why the app segfaults right after a build09:25
jaegermight help, heh09:26
sipoh, wait, maybe I found the original buildroot image09:26
sipit's 4.1.109:27
jaegerI tried 4.0.2 and 4.1.209:27
sipI can upload the full image, but it's really nothing different than a stock buildroot09:29
jaegerno big deal, but thanks09:29
jaegerdid you build it on a crux system?09:29
jaegerinteresting... when I try the one you uploaded, it can't exec /init09:30
sipoh, btw, generally speaking not all buildroot releases are eually stable09:30
jaegerI assumed that :)09:30
jaegerjust trying to figure out one that works09:30
jaegermore segfaults09:41
* jaeger sighs09:41
sipdid you over-customize the default buildroot settings?09:42
jaegerI don't think so, but perhaps09:43
sipOT: some vounteer to bribe the domain owner? It would be nice to have too09:43
sipseems it's a professor at WMU09:43
sipand he uses a new domain09:44
jaegerthings I changed: kernel headers -> 2.6.20, uClibc ->, gcc 4.0.409:45
jaegerthe rest is default09:45
jaeger(and busybox 1.4.1)09:46
sipis the segfaulting bb the one obtained by br or a custom one?09:48
jaegerI only pass it a config file09:48
siphere lies the problem --------^09:50
sipYou should try the default bb configuration firts.09:50
sipIf even that segfaults, go outside for a walk ;)09:50
jaegerI'm actually building it on 2 different machines now, with the default config and my config09:54
jaegerof course, the default doesn't produce a static bb, but whatever09:54
sipoh, go default and then build a static one on the resulting image.09:56
sipthat worked for me09:56
jaegerhrmm, could certainly try that09:57
jaegerthink chrooting would be sufficient?09:57
sipworst case you need to mount dev/proc09:58
sipI worked with loopmounted ext2 images so far without major issues09:59
jaegerI'll give it a shot09:59
*** Romster has quit IRC10:06
*** Romster has joined #crux-devel10:23
*** bd2 has joined #crux-devel10:24
bd2hey all10:26
bd2tilman, sip.. I guess you're in most charge about crux now.. didn't you thought about openness contrib access yet?10:26
tilmanit's not that hard to get access10:27
sipwell, I do not have a strong opinion on that. I think our task is more to provide the platform, other than that contrib should auto-regulate10:28
tilmananonymous write access is scary imo10:29
sipvery scary10:30
bd2not an anonymous. but just give access to *any* person who asked, plus invite portsdb members.10:30
bd2and don't ask them to do other jobs than maintaining their ports, like doing administrative tasks (review other's members ports)10:31
sipalmost-anonymous access ;)10:31
bd2sip, you can call it post-moderated access.10:32
*** j^2 has quit IRC10:35
*** j^2 has joined #crux-devel10:35
bd2port maintainers already will do huge job for crux - they will maintain ports. why you force them to be also police men? I'm not against police men, but they should be volunteers, not forced to do this in change to be members.10:37
bd2btw, why this scheme is scary? contrib repo already unofficial, unsupported, and I guess it's still disabled by default in /etc/ports/.10:40
bd2currently, I don't see much difference between opt/ and contrib/, both are restricted to "proven" members. You could move proven contrib members to opt, and open contrib.10:42
sipwell, one of the objectives of the new contrib was to improve the quality of ports, not sure if this can be reached with casual contributors10:43
sipbut apart from this, my only concern is security: I'm not very fond on giving access to unknown people10:46
bd2sip, I guess you're not reading #crux. you could read my sarcastic maths from there
tilmanwe love to read sarcastic stuff from people who think we're idiots10:48
tilmanand i mean 'i' when i say 'we' ;p10:48
bd2sip, security of what? of git repo? c'mon, it's revision control system. things could always fixed10:48
tilmanbd2: hohohohohoh10:48
sipunless randomguy adds a malicius line to a pkgfile10:48
sipmeanwhile, randomvictim updates the port ans he's screwed10:49
aonperhaps the randomguys also fix the contrib as others break it10:49
bd2tilman, I'm sure you're not idiots, but contrib issue seems a bit dumb10:49
aoni mean, isn't that how it works elsewhere, too?10:49
bd2sip, hey. contrib repo is unofficial and unsupported. Say it /etc/ports before user uncomment it. Move proven people to opt.10:50
sippersonally that would change contrib from "mildly useful" to "do not touch"10:51
sipmaybe that's just me.10:51
bd2I'm talking about common workplace for crux ports maintainers.10:51
sipIf other contrib members share that view, could you please collect ideas and suggestions and create a page on the public wiki?10:52
sipi think this would greatly help us taking some decision10:52
bd2Okay. probably I'll ask somebody else to write article. ;-) I'm unsure if people will understand my english. Otherwise I'll write it anyway.10:54
aonuhm, are you even a contrib member? :)10:55
aon(regarding other contrib members...)10:55
aon(especially the other part)10:55
jaegersip: did you have a problem with ncurses_dll.h missing in your ucroot?10:56
sipnot that I remember10:56
jaegerbusybox's menuconfig wants it10:57
bd2aon, no. for many reasons before contrib gitorization, and now I just don't encouraged to be police men by force.10:57
jaegerI'll use oldconfig for now and figure it out later, I guess10:58
sipI think I used a config file from the glibc one10:58
sipexactly ;)10:58
pitillobd2, opening contrib for all is quite dangerous in my opinion. You have the choice of private repos ("are untrully" and you can add something bad to contrib, that means you need to track contrib too, not only private repos). I think the point is select ports (welldone) and talk to the maintainer if he/she is not a contrib member. Or like cptn answered, tell the user that he has the option to make a suggestion to the maintainer to ad11:43
pitillod it to contrib. (I tried to explain it better at ML)11:43
*** Roomster has joined #crux-devel11:51
bd2pitillo, will not work. I can happily use my ports privately, and provoke work duplicating. Now, think of portsdb as just private repos. portsdb is not handy at all, as opposite to cental repo.12:02
*** Romster has quit IRC12:02
bd2As for asking other maintainer to add port to his collection - this will not work too. Ask yourself, would you add my ports to your repo, and periodically sync with me, check my ports? It's additional work for you12:04
bd2you'll tire very quickly, believe me12:04
pitillobd2, no, I mean to suggest one of your private ports to be in contrib.... (if you are not a contrib member... I think adding that port makes you to be a contrib member)12:06
pitillobd2, I tried to explain my opinion at ML. If you have a few minutes take a look to it and feel free to share your opinion, here or at ml. I like to know differents points of view and reasons.12:08
bd2pitillo, current contrib rules enforces me additional worries. I don't want them, and most ports maintainers don't want additional worries. Why they have to bother? Yup, they just don't go to contrib.12:10
bd2think of new contrib as if I'll create my own git repo, and will invite there crux portsdb members there.12:10
bd2and notice, w/o *any* rules, except they should not touch other's ports, and should try not damage repo12:11
pitillobd2, which are the additional worries?12:11
bd2pitillo, you have to follow email mangling standart, you MUST review candidate's ports randomly.12:12
pitillobd2, first I think is a standard.... I do not find it like a worrie. Second may be. If I had the knowledge to review other ports, why do no review them? I think is a good way to get more ports and make them better. May be, adding the suggestion to who is maintainig it can be usefull.12:15
sipc'mon it's not a great burden12:16
sipyou won't have 1000 submissions/day12:16
bd2pitillo, "why do no review them?" - why I have to do it?! I WANT to give you few ports, not less not more. I don't want bother with other stuff but my own ports.12:17
sipin case you didn't notice, the contrib repo is meant to be a community-based project12:18
bd2this is just simple as it is. Just a common/central place where crux users can put their ports, and not duplicate their work (because portsdb is not handy, everybody just copy Pkgfiles).12:18
sipnot some geocities-like webspace12:18
bd2sip, pardon, and opt is not community-based project?12:18
sipnot completely12:19
bd2what the difference?12:19
sipthe CRUX team took Per tasks when he left12:20
sipplus additional ports12:20
sipif that's what you're asking, I'm in favour of reducing the opt port to some extent and move some stuff to contrib12:21
sipdinner time. later :)12:22
bd2sip, I thought "CRUX team" was CLC, CRUX Linux Community. Maybe you're right, it's not community anymore.12:26
tilmananyone can contribute12:26
tilman"community project" doesn't involve anonymous write access either12:27
bd2tilman, of course they can. And you made such conditions that nobody want to contribute12:27
tilmanthat's debatable ;)12:28
jaegerarg, switch_root still seems to be messing up12:29
bd2If you think that there is just less and less "advanced" people, and that's why CRUX slowly loosing contributors, then you're wrong. You just don't made conditions to attract new ports maintainers (and later they could reborn to new "core" devels).12:33
pitillobd2, I think that is a good reason. You are contributing with your well done ports but like sip says, sounds a bit egoist. The best way to review one port I think is using it. (your contributed ports are tested by you and it is a good review if you done it well). If you have the sufficient knowledge to do good ports, review/overview another one (that can be reviewed by other member) and share your opinion and make it in the best way12:36
pitillo, can make the project bigger and the most important, better.12:36
sipbd2, to better explain, we've always be opened to contributions and it's great when somebody wants to join the crew. But the efforts of many can be useless if there is no or little coordination12:41
sipso we have (few) rules to opt and contrib12:41
*** Viper_ has joined #crux-devel12:42
bd2pitillo, I just don't have any time and, more importantly, I don't have any wish to review ports that I don't use. I don't care about them. If you will force me to, I'll quick get tired, and will retire. You saw it. I don't want to bother with that crap. Though, as I'm egoist, I will do my ports good, AND I'll try to improve ports that I *do* use. This is natural, implicit, and don't desire any explicit rules.  ( here "I" -- imaginary new contributor)12:44
sipAs a final note, I'm not the holder of the truth, maybe a more open contrib would end up as a better collection than the current one; it's only that we don't want to cope with the risks of such a change.12:47
pitilloI understand it, but if someone tell you to take a look to a new candidate to know your opinion (diferent of must to review that port because there is rule that tells that), you would not do it too?12:47
sip(not on at least)12:47
bd2pitillo, probably if I'll be in good mood, yes. But that does not mean, that I'll always be in that mood. With current rules, I'll be punished for that.12:49
pitillounderstand it, like I said, I think it is not a good way to test a port. Using it is the best. But taking a look to it can be a good way to take a first shot of it. I think that rule is not very good, and trying to ask for another opinion is better. Ask in general to make a review and share opinions (not all contrib members may do it, but if there are some with time and "good mood" can be a good way to add it)12:52
bd2pitillo, yes, but current rules says that randomly chosen two contrib members must take time and review new candidate's ports.12:54
pitilloif It is a bit hard to explain my opinion in spanish... thinking in english... xD12:55
sipbd2, that role could also be played by a single contributor that does not mind spending a couple of minutes on that12:55
pitillobd2, all can be talked... share opinion and study the best for the comunity. (if there are differences... can be voted. And be polite and respect the results)12:56
sipmeanwhile, rc1 is out: :-)12:58
jaegeryay :)12:58
pitillomay be I am in wrong and people who work hard on this project can not be aggred with this. I am new to a comunity project and I need more experience. Sorry if I am saying something that sounds crazy.12:59
jaegersip: got a static uClibc bb built, now the problem is that the /sbin/init on the new root doesn't want to run12:59
pitilloniiice. Tomorrow I will test it too :D12:59
siphhmm, double-checked the inittab format?13:00
bd2sip, page does not say that. And anyway, it's just one sentence in current rules I brought up. The whole rules looks unattractive, long procedures, additional worries to maintainers. It just looks like you think new contributor needs you, but actually *you* need new contributors.13:01
bd2sip, thanks for the RC!13:01
bd2as for new contributors, I need them, and I'm sure pitillo need them too. Because the more contributors project have, the less work have to be done by individual person.13:03
sipbd2, as said before I suggest the contrib members should be able to decide on policies13:03
bd2sip, ack.13:03
sipas long as there's some *minimal* control and no random people access, we're fine hosting contrib on crux.nu13:04
jaegersip: not yet, though it's straight from the packages13:04
sipwhich ones?13:05
jaegerat this point, the ones from my last updated iso13:06
jaegerwhose /sbin/init does work on a clean install13:06
sipwas the previous init binary from bb too?13:07
jaegerthe previous (initramfs) init is a sh script13:07
jaegerwhich executes properly13:07
jaegerit's after switch_root that the kernel panics13:07
pitillobd2, true. I thinked in the rule about reviewing ports... xD13:08
jaegeroddly, if I chroot to the new root and run init, I  get the expected usage message13:08
pitillobd2, why not change it to recruit new members?13:08
pitillobd2, if you review one port and seems well done, why not ask who made it and suggest him to maint at contrib?13:09
jaegerif I run it from the bb shell, I get "/bin/sh: init: not found"13:09
pitillos/maintat/maintaint it at13:09
bd2pitillo, because you have to take your time and recruit them. You'll get tired, I'll get tired. What I purpose is: made rules much simple and attractive to new comers, and you don't have to do anything, but wait. If person will prove himself as "good porter", then just move his ports and access to opt repo (or, just create new "official-contrib" repo)13:10
bd2"contrib" will be inbetween official and "crap", to teach and grow new devels13:12
sipjaeger, does fullpath work?13:12
pitillobd2, I think the crap are private repos... and official contrib is contrib.13:13
bd2pitillo, contrib is not official. And if you think private repos are crap, why you want to invite them to contrib?!13:14
pitillobd2, now we are working like you said, only the "good porter" needs to adopt the rule talked before.13:14
jaegersip: actually, I'm using the full path =/13:14
sipjaeger, I used init=/linuxrc in grub params in my tests, if that helps13:15
bd2oh.. you want force them to made their ports "standart" and only then permit them join to contrib. You see? They don't grow theirself, you force them.13:15
jaegeryeah, can't do that with initramfs13:15
jaegerthe boot-time init isn't a problem, though, it's when I switch to the real root that init fails =/13:15
pitillobd2, yes, I am agree with that. But it is a comunity. If that rule does not exists... you would be in contrib man.13:16
pitillobd2, or I see in that way. The only problem is that rule, because adopt some patern I think it is not difficult13:16
sipjaeger, i suppose you're already seen that13:24
sipthe green note on kernel panic seems interesting13:25
jaegerI have indeed :/13:25
jaegerI can post the init script if you'd like to see it13:26
sipyes, I love reading init files :)13:27
jaegerwell, you certainly don't have to look at it :P13:28
bd2pitillo, why you want to force them adopt patterns? Then they won't go to contrib members, because it's easier to not. But they're already made working ports (we all copy them from portsdb), and you force them to made something else, because you like it. I agree, if they will want to be in "official" repo, then and only then they have to follow standarts. Otherwise, it's mandatory just to have safe and working port (which is still unofficial).13:28
bd2all I suggest is replace "portsdb" by something which is more handy and prevents work duplication, i.e. git repo.13:29
bd2central repo13:30
pitillobd2, the last part is the most important I am agree. But some patterns (voted) ,to keep the unity in comunity, are good, Pkgfiles can be done in one way and in this way can be better read them.13:32
sipjaeger, is "attempting to kill init" the only relevant message you get?13:32
jaegersip: yeah.
jaeger"Switching root." is an echo in my script, "switch_root: moving root" is output from switch_root, and the kernel panic is after the switch13:33
bd2pitillo, yes, they can. But would you reject member and loose few ports, if he using spaces instead of tabs? or he not mangles his email? or does not write Description: field?13:33
pitillobd2, I do not understand why not adopt a few rules to make more readable Pkgfiles....13:34
sipmy take: the real root init fails, thus not replacing the proces with pid=1 and causing a panic13:34
jaegerthe question is why? :)13:34
jaegerwhich I'm having trouble figuring out13:34
sipjust trying to summarize ;)13:34
pitillobd2, are that rules voted by the comunity? if yes, why not be a bit mature and adopt them if more people thinks is better?13:34
bd2pitillo, because *you* like this patterns, other person don't. But his port works and safe. You will not use such?13:35
sipso the rr init either looks for a non-existing file or a bad formatted one13:35
pitillobd2, I understand it. And see it in one member. But I really do not understand why not adopt that voted rules13:36
jaegeror perhaps a file descriptor that doesn't exist or isn't correct?13:36
bd2pitillo, no, rules voted by the core members mostly. But this is okay for their official repo. And if contrib member will want to join official repo, it's okay to force him follow pkgfile "good looking" format.13:36
pitillobd2, you are working in a comunity, if someone explain and idea and is voted, and you do not like it... you leave it...13:36
jaegerI could try to build a static strace into the initramfs, I guess13:36
pitillobd2, sip told that contrib rules are different. (if I understand it well)13:37
pitillobd2, contrib members can vote the rules... and if majority says to follow core for example? you are not agree with it and leave the comunity13:38
bd2pitillo, no. there is one contrib rules, but he just verbosely permitted to not follow one of them. ;-)13:38
sipthat's not true13:39
pitillobd2, there are more than one rule I think (and all the procedure).13:39
bd2sip, no?13:39
sipI said that contrib members should make up their own rules (to a certain extent) and agree on those13:40
sipthat's a bit different from ignoring the rules :)13:40
bd2sip, pitillo, you're talking about two communities, right? About "core" and "contrib". This is not correct. We're all crux users, but some of us is "official" and somebody not. Those who are official or want to be in meantime, they do and must follow standarts.13:41
pitillonow understanded...13:41
sip...the others can publish their repos privately :P13:42
bd2sip, which is work duplication in case of portsdb.13:42
sipthe portdb is there for people NOT wanting to join contrib13:42
pitilloand less "well done" ports in contrib13:43
sipwhatever contrib becomes, there will always be people not wanting to join13:43
bd2sip, yes! because they don't bother to follow standarts and other stuff. They just made their ports, and they work. All I suggest: remove portsdb, made one "unofficial repo, for persons who don't want to join contrib for one or more reasons"13:44
pitilloif you want, follow them, if not, walk near...13:44
sipI think we agree on the core question, though. I'd very much like a bigger contrib instead of a thoushand scattered repos13:44
sipbut a big contrib without some teamplay is something I'd like to avoid13:45
pitilloI think people do not like teamplay... The hard point is review the other ports... adopt standars are only put spaces xD13:47
bd2no need teamplay (portsdb don't have any teamplay). Just moderator to deny persons, who, for example, [un]intentionally touched other's ports.13:48
*** sepen has joined #crux-devel13:48
bd2several times, of course.13:48
*** treach has joined #crux-devel13:48
pitillodo not touch other ports... it is easy. Contact him and talk....13:48
bd2yup. As I said, for my ports I'd wrote "# Please feel free to update this port if it's outdated", thus other contrib members can touch it for good purpose13:50
bd2but implicitly, it's forbidden to touch other's ports13:50
*** sip has quit IRC13:51
pitillowell, the remove it from contrib and put it in your private repo... if some contrib member see it.. can change Maintainer field and put it in contrib13:51
pitillo(to keep it update)13:51
pitilloI think is better that touch someone work.13:52
* treach gets the rc. 14:04
treachI'll try out the upgrade process tomorrow.14:04
jaegerargh! strace doesn't help14:04
treachIf I didn't know that was unrelated, I'd be worried. :D14:05
*** Viper_ has quit IRC14:59
jaegeroooh... think I might have solved the initramfs problem15:08
jaegergonna build another ISO and test15:08
jaegeryay, it's working15:14
*** jaeger has quit IRC16:05
*** jaeger has joined #crux-devel16:44
*** j^2 has quit IRC17:08
prologicdid sip get my email ?17:15
*** treach has quit IRC17:19
*** pitillo has quit IRC17:56
*** _mavrick61 has quit IRC19:02
*** _mavrick61 has joined #crux-devel19:03
*** Roomster has quit IRC21:38
*** Romster has joined #crux-devel21:41
*** Roomster has joined #crux-devel22:29
*** Romster has quit IRC22:50

Generated by 2.11.0 by Marius Gedminas - find it at!