IRC Logs for #crux-devel Wednesday, 2012-04-25

juegood morning01:45
jueteK_, jaeger: the two are not really integrated, but after you've created a new accout FS sends you a confirmation email with some code you need to complete your registration01:51
juehmm, I just say that you cannot register two accounts with the same email address, it would have been better to disable the spam user and not to delete the account01:57
jues/say/saw/01:57
*** mike_k has joined #crux-devel02:17
*** joe9 has quit IRC02:21
*** rines has joined #crux-devel02:46
juethere's a new procps port in my private repo using the sources from debian, they forked procps because the original sources seems unmaintained03:57
jueplease give it a try and report back, thanks03:58
juebtw, gentoo is using it as well03:59
frinnsttilman: reverting cairo may be prudent04:10
Romsteri'm not noticing any cairo issues but i'm on binary nvidia driver04:13
Romsteris it just the intel cards that get it frinnst ?04:13
juenouveau is also but differently affected, it's unusable slow with new cairo04:24
juebut right, binary nvidia works without issues for me04:26
*** sepen has joined #crux-devel04:36
sepenwondering about to start 2.8 branches for development, what do you think?05:22
sepensince kmod/udev/module-init-tools is one of the tasks to do in TODO2805:24
jueit's still listed there because we have to think about it wrt setup-helper05:26
juebut right, make sense to mark it done ports/core section of TODO2805:27
sepenimho 2.7 series should have the old stuff05:28
sepenwe can just start the development in our boxes to test the new stuff05:29
juedo you know how far a 2.8 release is?05:29
sepenhehe05:29
sepen2.8, ... 3.005:29
jueI don't know05:29
sepenwell, me should start to decide what kind of stuff should be in 2.805:30
jueIMO the kmod thing is not a big issue05:30
sepenthen discard things to 3.005:30
sepenjue: yep05:30
juewhat's the difference between 2.8 and 3.0?05:31
sepenmaybe x64 official?05:31
sepenor new rc-ng to me05:31
sepenwell things that means big changes too05:32
sepenbut we can start to mark for 2.8 the new udev/kmod + 3.x.x series of kernel, ...05:33
juewell, a glibc 2.15.1 release is planned for beginning of may, that might be _one_ reason to think about a new release05:34
juesepen: TODO28 is a wiki, feel free to add stuff etc.05:35
sepenso we can start 2.8 branch for it05:35
juebefore we start a branch for something we should know what we want05:35
frinnstrc-ng gets a -1 for me, for now atleast05:36
sepenjue: I like TODO28, but I'd like more to have all of that as filed tickes in FS, so I think that is easiest to change status for a ticket, for example from 2.8 to 3.0, etc.05:36
sepenyou know, 2.8 milestone, etc.05:36
sepenfrinnst: -1 for me, for now too, but don't seems a bad idea for the future05:37
juesepen: for me the wiki page it's more comprehensive as long as I am the only person adding stuff to it05:38
sepenyeah :D05:38
sepenjue: you have the power!05:39
juebut at all I'm _very_ happy if someone else jumps in to help with all of that05:39
teK_yeah I voted to help with rc-ng and did not forget about it05:41
jueto be more clear, I'm not the leader of our project, but as long as no one else is doing what needs to done ...05:41
sepenjue: TODO files are fine for me, but in most cases we can start to add 'those new ports' to a new branch in our development05:41
jueteK_: nice to hear that05:41
sepenmaybe I can explain that right, sorry05:43
juesepen: sure, but sorry, improvements in such areas are not our biggest problem at the moment05:44
sepenjue: I think you're the potential-leader now05:44
juesepen: no05:44
teK_i think leadership by getting shit done is a good dev-model :)05:45
teK_* as long shit gets done05:45
jueteK_: agreed ;)05:45
sepenyep05:45
sepenwell, I'll try the reboot now, later05:47
*** sepen has quit IRC05:48
juewrt rc-ng: it's fine for me to keep our simple one, but we should find someone who takes care of it, because I get the impression that it's a bit too simple currently05:48
teK_agreed05:49
teK_at least parallel startup or the like should be available05:49
juefor me the whole lvm/mdadm/raid stuff is a bit hairy, mounting the different kinds of required virtual filesystems as well05:51
teK_hairy? why?05:53
juenot sure if we do the right things currently05:56
*** sepen has joined #crux-devel05:56
teK_hm06:00
teK_I understand that lvm has to be done, raid/md-stuff can be disabled via kernel-boot-parameter06:00
teK_and: works for me[tm]06:00
teK_but sure we could/should review that06:00
juebbl06:15
*** rines has quit IRC06:25
*** sepen has quit IRC06:45
*** sepen has joined #crux-devel06:59
juesepen: back to your initial question, I think it's to early for a 2.8 or whatever branch07:12
juewe should do that only when we start a new release cycle07:15
juebut the first thing for that to do is to start a 'official' arch discussion07:16
sepenhmm, just tried to find a way to share our development via git, or something07:16
sepenjue: +107:17
juewhich development?07:17
sepenwell maybe ports07:17
jueah, you mean something for the procps port I mentioned above?07:18
sepenjue: just I was wondering to adopt a release cycle like the xfce guys does, but sounds dificult07:19
sepenyeah, in that mean we could work with new ports, or stuff07:19
sepenATM the only way is to use your private repo for that07:20
juewell, for that a test repo should be sufficient07:20
sepensorry, its dificult to me explain the concept07:21
sepenhttp://wiki.xfce.org/releng/4.10/roadmap07:21
juesepen: yeah, I know you like such things much ;)07:23
sepenjue: I want to say something like ATM our -product- stuff is the one in HEAD, but we could just release the -TAG- we want07:23
sepenfor example, I've two production boxes running postfix at office, so I can just sysup these boxes07:24
sepenmaybe we could just say the released packages are fine for stable, right?07:24
sepens/say/say that/07:25
sepensorry if I can explain the concept with my poor english07:25
juesepen: maybe it's easier to explain the problems you see currently?07:27
sepenjue: think in production boxes running crux and a maintainer who wants to maintain their systems up to date07:28
jueyeah, and what's the problem with CRUX here?07:29
sepenstable method should use only packages relased in 2.7, 2.7.1, etc.07:29
jueisn't that what we try to do?07:30
sepenno, its fine, my question is that where is the point when development would be stalled before switching to new release?07:30
sepenso how many -new changes- should contain 2.7 branch before we finally switch to 2.8 or whatever07:31
juewhy stalled? we always maintained two branches until we release the next version07:31
sepenyep I forgot that07:32
sepenbut how we determined in the past, when is the time to switch07:32
sepen?07:32
sepencan't remember it07:32
juebut the ongoing merges are a lot of additional work, so I always prefer short release cycles07:32
jueafter we have a good feeling that everything is fine ;)07:33
sepenah right, we used merges .. can't remember it, sorry07:33
sepennah nah, sorry jue for loosing your time07:34
sepen*waste07:34
juewe mostly had one or two test-releases and at least one rc after that07:34
sepenyep, I need to refresh some concepts07:35
sepenthe problem was that the first we created 2.8 the first we should start to merge lot of commits, and its really hard07:36
sepenin other words 2.8 should be created at the right time07:36
jueyeah, after the branch is created we have to maintain 2 branches in parallel07:38
sepenjue: you said you were not the leader, but guy, you'are the basis behind many things :D07:38
*** rines has joined #crux-devel07:41
juewell, I try to do my best and are fortunately able to spend quite some time for CRUX, but that's all, real development doesn't happen currently07:45
juethat's the main reason why I like rc-ng, it's new and there is a smart guy behind it07:47
jueTBH the second reason is more important ;)07:48
jueof course, that's only my opinion07:49
juesorry, but have to run now, later07:53
*** mike_k_ has joined #crux-devel08:10
*** mike_k has quit IRC08:14
*** joe9 has joined #crux-devel09:24
*** rines has quit IRC09:34
*** j^2 has quit IRC09:55
Romsteri think sepen is looking at a stable and unstable dev version release scheme10:01
juere10:25
jueRomster: I think our current scheme is the more like the stable one10:26
juewe do not add stuff that needs a lot of user action, like libpng for example10:28
juewithin a release, I mean10:29
jueat all I think this has worked quite well so far, with only very few exceptions10:33
jueand recompiling the kernel with two new options isn't a exception for me ;)10:34
Romsterthough that has caught out a few users already10:51
rmullthe udev thing? that one bit me11:13
rmullbut it was quickly sorted. wasted time assuming my hard disk was getting furry.11:13
jueRomster: right, if you have any idea how we can improve the notify thingy11:21
*** horrorStruck has quit IRC11:21
juebut at all we should keep in mind that we are CRUX and not a mainstream handholding distro11:22
jaegerI think something like the udev change would be a good candidate for news on the main crux.nu website11:32
jaegerAs well as any other package updates that require intervention11:32
jaegerThat would require users to read it, of course11:32
jaegerPerhaps another way would be to create a package that would automatically check the site for news in an RSS feed or something11:32
jaegerand print it in the motd11:32
jaegerOr perhaps not the motd since some people will have customized it but on login, at least11:33
jaegerOr alter pkgmk to print out a message if there's one included with that particular version of the port11:34
jueyeah, totally forgot our crux.nu website, would be good to have some more news there anyway ;)11:35
jaegergentoo's portage prints those sorts of messages during the build and at the end, for example11:35
jaegerso there are a lot of options11:35
jaegerFor kernel options we could even have a script check /proc/config.gz (if it exists) for things like DEVTMPFS11:37
rmullWhy not just notify via the mailing list?11:40
jaegerIt was posted to the ML, many didn't read it or weren't subscribed11:41
rmullYeah, I guess I did read it. I misinterpreted it as just warning of an update, and I glossed over the critical kernel dependencies11:42
rmullWhich was entirely my fault, not crux's11:42
jaegerI'm just trying to think of a way to make it hard to miss11:47
rmullThe gentoo approach is a good one, IMO (disclosure: I am ex-gentoo user)11:51
rmullThe other thing that might be added to the post-install process is alerting of a need to run a rejmerge11:52
rmullSince software versions make rely on different config file formats11:52
rmull(IMHO)11:52
rmullGotta run.11:52
*** rines has joined #crux-devel12:05
*** mike_k_ is now known as mike_k13:38
teK_my biggest wish would be to have a script to do 2.x -> 2.y upgrades14:34
teK_I do not consider having to reboot my server to an ISO 'best practice'14:34
*** rines has quit IRC15:25
*** mike_k has quit IRC16:16
*** j^2 has joined #crux-devel18:21
*** sepen has quit IRC19:37
*** tilman has quit IRC20:10
*** tilman has joined #crux-devel20:11
*** mavrick61 has quit IRC21:48
*** mavrick61 has joined #crux-devel21:49

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