IRC Logs for #crux-devel Tuesday, 2014-02-11

*** Amnesia has quit IRC01:01
jaegerregarding the comments about github earlier, I've considered installing something like gitlab just for a more modern look to our repos01:36
*** Amnesia has joined #crux-devel01:43
jaegermade a quick lint script for the defconfig:
prologicjaeger: what's this for?02:29
jaegerfor making sure I don't leave those options out of the defconfig that ships on the ISO02:30
jaegershould cut down on some of the installation questions02:30
prologicTo make updating the ISO a bit easier?02:32
prologicAnd keeping a track of kernel config options that are required? :)02:32
jaegerWell, they aren't all required for everyoe but the idea is to make the default config more usable without modification02:33
prologichmm ic02:33
prologicI don't need half of the file systems your script checks for though02:34
prologicjust saying02:34
prologicI think I use EXT3 and BTRFS currently02:34
jaegerI'm pretty sure I just said something like that02:35
jaegera few seconds ago02:35
jaegerperhaps I'm losing my mind, though02:35
prologicsorry I misread :)02:35
prologicno no02:35
prologicjust my eyesight :)02:35
prologicyeah ok scrap that02:35
prologicgood :)02:35
jaegerLet's say you're a newish crux user and you don't know kernel configs really well yet - the default config should get you booting on some common hardware02:35
prologica sane default config sounds like a good idea in any case for new users02:36
prologicgood use-case02:36
prologicand probably good to just get up and running quickly too02:36
prologici.e: lazy crux user :)02:36
jaegerthe 3.0 ISO has one, even, I've just added more support02:37
*** mavrick61 has quit IRC03:39
*** mavrick61 has joined #crux-devel03:40
Romsterjust adding every filesystem and sata ide driver as builtin n the ocnfig with IA32 set so those that sysup gcc glibc-32 works07:22
Romsterprologic, my build system is nothing more than romster/pkg-clean and a chroot07:23
Romsterthough i have been lazy to setup docker yet07:23
Romsterand libata subsystem07:24
Romsterheck isn't there localdefconfig or some other way to configure all device drivers?07:25
prologicRomster, okay -- I'll build a build sytem then08:17
prologicprobably next long weekend when I have more time08:17
Romsteri got myself a little build farm here. i should have a few more boxes soon08:27
Romsteri should be able to make it to -j3008:27
Romsterif i setup with docker i could buildbot everything08:27
Romsterdefconfig  - New config with default from ARCH supplied defconfig08:44
Romsterthere we go we can provide a defconfig08:45
*** deus_ex has quit IRC12:03
*** deus_ex has joined #crux-devel12:06
*** deus_ex has quit IRC12:13
*** deus_ex1 has joined #crux-devel12:13
jaegerdefconfig and localyesconfig are very different, defconfig is not very useful12:39
jaegerlocalyesconfig depends on the specific hardware being installed12:39
jaegerMight make more sense for me to say that my defconfig is more useful for a new install than the arch defconfig12:39
prologicI think Romster misunderstood what you meant by "defconfig" :)12:42
prologicI took it to mean you're creating a better default kernel config for the ISO12:42
prologicOne in which new users can copy to their new system with some sane defaults12:43
prologicif it boots the iso/cd it should boot the system right? :)12:43
pitillosorry :)12:43
teK_prologic: wrong. You have to take the used initrd into consideration, too12:48
jaegerthere are 2 kernel configs shipped12:48
jaeger1 is the initrd which uses nearly all modules12:48
jaegerthe other is the "defconfig" in my terms which has a few things changed to builtin for the aforementioned "sane defaults"12:49
jaegerWhen you run setup, the latter config is the one that gets copied to the installed system12:49
prologicyeah sorry forgot the initrd stuff :)13:01
prologictbh never really used it13:01
Romsteryeah i'm awre of the 2 kernel configs13:45
Romsteri meant to include the common drivers that most people use to get root to mount13:46
Romsterthe bloat can be cut back later.13:46
*** Roomster has joined #crux-devel13:48
jaegerHrmm, I need to test the ISO in hyper-v again13:56
jaegerteK_: can you test the latest one with virtio?13:56
*** rexich has joined #crux-devel14:00
rexichhello people14:00
rexichwhat's up?14:00
rexichone question came to my mind: is CRUX a rolling distro?14:00
jaegerNot officially, no14:00
prologicdefine Rolling distro?14:04
jaegerRolling release upgrades can be done manually but the officially supported upgrade path is ISOs14:04
prologicWell typically this involves (usually) toolchain updates right?14:04
prologicAs well as dropped packages, injected packages, etc14:05
*** Roomster has quit IRC14:05
prologicI think we've history tried to make the ISO the upgrade path?14:05
*** Roomster has joined #crux-devel14:05
*** Roomster has quit IRC14:06
*** Roomster has joined #crux-devel14:06
rexichrolling distro: in my opinion, one which has the most current software and lacks a version, instead it has a monthly ISO, such as Arch14:06
prologicEven though (I think) if you follow the port updates closely and keep your system(s) up-to-date you can not worry about upgrading with the ISO -- but as I've never seemed to be able to keep any of my CRUX hosts up-to-date they tend to get rebuitl from scratch many eyars later after being behind several minor versions of CRUX :)14:06
prologicyour definition is ambigious :)14:07
prologicCRUX does have the latest software14:07
prologicall the time pretty much :0 depending on laziness of port mainterins :)14:07
jaegerIt's not ambiguous, it's subjective14:07
rexichdoes CRUX have a monthly ISO?14:07
prologicBut ISO images have typicalyl been updated/built every year or so14:07
jaegerWe don't officially do either rolling release or "rolling distro" by your definition14:07
prologicjaeger, ahh yes that's the word I was looking for :)14:07
jaegeryou can do it manually if you like, though14:08
prologicrexich, no14:08
prologicnto historically14:08
Roomsteri've done it without a burned cd and loop mounted iso or my own chroot toolchain and more ports all built then pkgadd/pkgrm manually. but i would not advise it14:08
prologicyearly really14:08
jaegerThe official release ISOs are just built at release time but I periodically build updated ones that are unofficial14:08
rexichI keep and use the official one for system recovery or for installing it on other computers14:08
rexichwhat about the plans for version 4?14:08
prologicas do i14:08
prologicon a usb stick14:08
prologicbut I hardly ever upgrade systems via cd/usb/iso14:09
prologicprt-get update or prt-get sysup14:09
prologicwith rejmerge14:09
prologicand keep doing your software research :)14:09
prologicread read read :)14:09
prologichaha what plans for version 4? :)14:09
Roomsterwe went to 3.0 to demote 64bit multilib i doubt we will move to 4 for a very long time14:09
prologicdon't we just whim it? :)14:09
prologicdepending on what's changing in the GNU/LInux ecosyttems()?14:09
prologicrexich, ever remember me saying CRUX is just filesystem + rc + pkgutils?14:10
prologicoptionally prt-get14:10
rexichyes, I remember14:10
jaegerWe don't pick the numbers arbitrarily, a major version change comes with something big like multilib14:10
prologiceverything else is just ports (software)14:10
rexichI understand now :)14:10
prologicwe went form 2.8 -> 3.014:11
prologicx86 (i686 really)14:11
prologicto x86_64 multilib14:11
Roomsterevery 6 months to a year new iso but prt-get sysup usually gets you there14:11
prologicbut I think in the upcoming 3.114:11
prologicwe're dropped the multilib tag from the CRUX version14:11
Roomsterwhy do that?14:11
prologicRomster, whate version did I introduce you to CRUX at ?:)14:11
rexichRoomster: I do prt-get sysup often, so everything's up to date, I compile my own kernels, currently 3.12.2, the stable one14:11
prologicRomster, I didn't :)14:12
prologicfrinnst did :)14:12
prologiccheck the commit logs for core :)14:12
prologica year after I started using CRUX :)14:12
prologicLinux daisy 3.12.9 #1 SMP Sun Feb 2 11:02:45 EST 2014 x86_64 Intel(R) Core(TM) i7-3770K CPU @ 3.50GHz GenuineIntel GNU/Linux14:12
prologicoh yeah 3.12.9 here too14:12
prologicthought I was running 3.13 for some reason14:12
rexichLinux prenosnik 3.13.1 #2 SMP Wed Feb 5 11:30:02 CET 2014 x86_64 Intel(R) Core(TM)2 CPU T5600 @ 1.83GHz GenuineIntel GNU/Linux14:13
rexichI forgot the 3.12.2, too lazy to compile it14:13
Roomsteri'm on my laptop if ya wondering why i'm in here as Roomster14:13
Roomstertoo lazy to setup znc14:13
rexichI wondered, why two Romsters :D14:13
prologicyeah your nick(s) are confusing at best14:13
prologicjust stick with one :)14:13
prologicznc ftw :)14:13
prologicalways the same nick regardless of wher eyou are :)14:13
prologicwell at least that's how I play it14:14
prologicoh which reminds me14:14
Roomsteri got workster as well at work14:14
Roomsterfrinnst, seems to be making willy nilly changes are they even on the todo 3.1 list14:14
Roomsteror they are now14:15
rexichguys, thanks for the answers, I wanted to see where CRUX is headed to14:15
prologicwhere did you think it was heading?14:15
Roomsterwe are all moving along i'm more concerned with avoiding systemd14:16
prologicwe'll avoid it :)14:16
prologicno doubt about that14:16
rexichprologic: ahead, sans systemd and the other abominations :D14:16
prologicI played with a systemd based system today14:16
rexichbtw, do you plan to keep CRUX multilib or you will eventually make it pure 64-bit?14:16
prologicCoreOS (at work on OpenStack)14:16
rexichoh god14:16
prologicI nearly tore my heir out14:16
prologicjust trying to figure out where the damn configs are14:16
prologicand how things work14:16
Roomsterbetter or it'll be a turning point where i'll give up on linux. not putting up with that crap14:17
prologicwell obviously some folk see it as a good thing14:17
prologicseveral of my colleagues do14:17
rexichI've visited the hacklab in my city and were discussing systemd, most of the people just don't care, I'm stronly against it, they just want their Linuxen to Get Shit Done :D14:17
prologicand their justiciation: desktop performance14:17
prologicmy retort to that was14:17
Roomsterits a basterisation of everything wrong14:17
prologicmy desktop at home boots up to XFCE in 5s14:17
prologicwith sysvinit14:17
prologicwho gives a !@#$14:17
prologicwhy would I want it to boot in 1-2s?14:17
jaegerhyperbole all you like, systemd isn't going away. :P14:18
prologicbesides that I never reboot my desktop anyway14:18
prologicunless there's a power failure14:18
prologicjaeger, of course not14:18
prologicbut no-ones forcing it down our thorats :)14:18
prologicgah :)14:18
prologicneed aspell for X-Chat :)14:18
rexichI put my laptop to sleep when I go to sleep and turn it off when taking it outside :)14:18
Roomstersame and it's on a ssd so even if i do 20 seconds later desktop14:18
rexichprologic et al: is there a plan to keep CRUX multilib or will it eventually be made purely 64-bit?14:19
prologicthe only init I think I'd consider is runit14:19
prologicwhich isn't too bad14:19
jaegerThe plan is to stay multilib.14:19
prologicand has all the same performance characteristics of things like upstart/systemd14:19
Roomsteri used to run runit its nice i should use it again14:19
prologicI have yet to test it actually :)14:19
jaegerWe opted for that because we didn't want to maintain both14:20
prologicrexich, you can make it pure 64bit now14:20
prologicjust by removing the 32-bit compat layers14:20
rexichin the Pkgbuiklds?14:20
prologicI don't recall exactly what packages those are now14:20
rexichI think I know which ones14:20
jaegerYou can but it isn't as simple as prologic makes it sound14:20
Roomsteredit gcc --disable-multilib14:20
Roomsterpkgrm glibc-3214:20
jaegerYou need to rebuild the toolchain14:20
prologicand recompiling glibc/gcc14:20
rexichyes, gcc :) that will take a while :)14:21
jaegerIt isn't worth the effort, pretty much14:21
prologicI forgot that bit :)14:21
prologicjaeger's right14:21
prologicmost of us haven't bothered14:21
prologicI asked, got the same answer14:21
prologicwent *meh*14:21
prologicleave it alone in case I ned it14:21
prologicturns out I do :)14:21
Roomsterand removing IA32 out of kernel14:21
prologicstill lots of things that are 32bit :/14:21
rexichI intended to install Skype, but when I saw the dependencies...*gasp*14:21
rexichI left IA32 in kernel, or gcc multilib won't compile14:22
rexichI'll try this, nevertheless14:22
Roomsterhow many 32bit ports do you use out of curiosity14:22
Roomsterheh i did all the dependences nearly. jaeger did the initial 32bit multilib i pretty much extended it14:23
prologicI'm mostly just using all the ones needed by the stream port14:23
prologicotherwise I would have been using none14:23
prologicI may end up removing stream and all it's deps14:23
Roomsterfeels as if it's just jaeger and myself in compat-3214:23
prologicI cannot get into PC gaming at all14:24
prologicthat reminds me14:24
prologicI should put docker into contrib14:24
prologicmight do that now and test the builds14:24
rexichI do not use anything 32-bit at all14:25
rexichI wanted Skype, but alas14:25
Roomstertoo eeping long dep tree14:25
Roomsterbut it works14:26
Roomsterthough avoid it if you can14:26
Roomstersome must run skype for work14:26
Roomsterbit hard for me to avoid it when it's on windows linux and my android phone14:27
jaegerprologic: what's stream?14:29
prologicdid I say stream?14:29
jaegermore than once, yeah, so I assumed it was intentional, heh14:29
prologicsorry late :)14:30
prologicneed to go to bed!14:30
teK_jaeger: can do tomorrow14:30
jaegerteK_: no rush, thanks :) Do you just use qemu for that or is there something else required for virtio?14:31
frinnst <Roomster> frinnst, seems to be making willy nilly changes are they even on the todo 3.1 list  ?14:32
prologico oh14:32
prologicRoomster's in trouble :)14:32
frinnstno, i just dont understand what you were talking about14:33
prologicno idea :)14:33
prologicI just mentioned the fact you removed multilib from /usr/bin/crux in core14:33
prologicin preparation for 3.1 I presume14:33
Roomsterdropping the multilib out of the x86_64 string14:33
frinnsti dont think that was ever really supposed to be in there, probably just followed over from jaegers toolchain14:34
jaegerNo real reason to keep it there. It was only in originally because multilib wasn't official14:34
frinnstthe 3.1-ports i've committed are just my testversions anyways14:34
Roomsteri hadn't looked just going by what i read14:34
jaegerso that it was obvious if someone was running the multilib version instead of the release14:34
frinnstdid it in fear of losing the work14:35
prologicall good :)14:35
frinnstgranted, not much work done but still :)14:35
Roomsteroh i guess so as long as it's in gcc -v as what i stated14:36
Roomsteri have to step up and deal with lvm and ports and i've started testing without pthred-stubs14:36
Roomsterbut i've hit a mental wall lately and just been relaxing more14:37
jaegerDefinitely good to take time to relax. Fewer mistakes that way and less stress.14:37
prologicthat's why I'm going on holdiays as of Mon14:40
*** rexich has quit IRC14:41
Roomsteri get too burnt out from work and then everything else i do so yeah i have to relax sometimes14:45
Roomsteralso means i don't go bitching around here as much either :D14:45
*** rexich has joined #crux-devel14:53
jaegerok, hyper-v install worked fine14:53
frinnstcrux on hyperv?15:06
jaegeryes. someone requested it a while back15:07
frinnstnice, and poor whoever-requested-it15:07
prologicI probably want CRUX on OpenStack for work soon15:08
*** rexich has quit IRC15:23
*** Roomster has quit IRC15:54
*** jue has joined #crux-devel18:35
jaegerheyo, jue19:24
jaegerjue: how goes?19:27
juejaeger: great, just back from a short vacation :)19:52
jaegerwelcome back, hope it was a nice vacation19:53
juejaeger: thanks and yes, it was nice and relaxing19:54
jaegergood :)19:54
juehope that I'll have a bit more time to work for crux the next weeks19:55
jaegerShould we start really focusing on 3.1?19:55
jueyes, I'd say19:56
jaegerI've worked on the ISO a bit in the last few days in preparation19:59
jueour TODO31 is not very long and not too complicated so it should be possible to do the final work in a shorter time19:59
jueahh, that's great19:59
juelooks like only a few points in TODO31 needs decicions20:03
jueI see PAM, eudev and 'consolidate network-tools'20:03
jueeven thoujgh the latest is a simple thing20:04
jaegerI'm fine with the network tool consolidations, they all make sense to me20:04
jaegerI've been using eudev for a while from your repo with no issues, too20:04
jueI see more open questions with our 'clean up rc' point20:05
prologic+1 for cleanup of network tools20:07
prologicIHMO /sbin/ip is all the network tools you need20:07
prologicisn't that iptools2?20:07
prologicAlso +1 on the eudev - we need to keep the CRUX/Gentoo (and others) rhederic20:07
jaegerthe things that are listed as postponed, were those postponed until 3.1 or later?20:08
jaegerit's iproute220:10
juemy understanding is until 3.120:10
prologicyeah that one :)20:10
jaegerok. I wasn't sure about the wording20:10
prologicwhat's with the "rc" cleanup?20:11
prologicwhat's needed there? I could help do this20:11
prologicI originally a long time back rewrite all the CRUX rc scripts for dash so they were sh compatible20:11
jueprologic: it's more than ip, stuff like hostname, whois and traceroute e.g.20:12
prologicaren't they in binutils or somethign?20:12
prologicwhat are we doing?20:12
jaegertake a look at the TODO31 doc, it has the proposed changes outlined20:12
prologicdropping the old iproute20:13
prologicClean up rc. Move btrfs, lvm to udev? <-- wtf?20:14
prologiccore/net-tools is unmaintained and works only with a couple of patches. It's main functionality has been replaced by the iproute collect <-- so things like /bin/hostname have been replaced by?20:14
prologicgotta get to work, I'll do more reading there and comment :)20:15
prologiclooks okay most of the points20:15
*** Romster has quit IRC20:22
jaegerSomething that I haven't added to TODO31 is UEFI support. The work for that is mostly already in place but we'd need to have efibootmgr and gnu-efi on the ISO20:25
jaegerthey currently live in contrib20:25
jueprologic: the idea is to replace the conditional calls of vgscan and btrfs in /etc/rc by some clever udev rules20:26
jaegerjue: how would we go about that, anyway? I'm familiar with creating udev rules to adjust permissions and the like but not sure about replacing something like vgscan/vgchange20:26
jaegerudev already has some dm-lvm rules for device naming but not for the scanning and activation20:28
juejaeger: not replacing but letting udev it call, IIRC20:28
jaeger <-- something like that?20:30
juedunno who we come to the idea that something like that might be work, IIRC frinnst, Romster and teK_ were involved20:31
jaegerPerhaps we should step back and ask: is there a benefit to changing the way it is now?20:32
juenot really, it's more a question of style ;)20:35
jueIIRC the whole debate starts at time we've added official btrfs support20:37
jueand we had to change re for that20:37
jaegerIs there a best practice from the lvm upstream?20:39
juesorry, no idea20:39
jaegerI wonder how fedora/redhat do it20:39
jaegermaybe worth checking since they're the upstream20:40
jaegerI think I have a fedora 20 VM around somewhere20:40
jaegerhrmm... not obvious how it's called20:42
jaegerAh, I guess it's systemd there. I'll look at something older like centos or RHEL5/620:45
juejaeger: have to go now for today but will be back tomorrow20:51
jaegerok, take care20:53
frinnsteudev has automagic btrfs-scan20:55
frinnsti dont need my custom btrfs-rule anymore20:55
frinnstyou need to run "btrfs dev scan" to properly mount arrays and do other cool stuff20:56
frinnstwe do mdadm stuff in udev already20:58
*** Romster has joined #crux-devel21:41
prologicjue: prologic: the idea is to replace the conditional calls of vgscan and btrfs in /etc/rc by some clever udev rules <-- why not ship something like /etc/rc.conf.d for lvm/btrfs or rc.d startup scripts - if the user doesn't use either they can simply disable/remove? I could be talking crap here :)21:42
frinnstyou are :D21:43
prologicawesome :)21:43
prologicI just thought instead of conditional expressions and running said commands in /etc/rc (or /etc/rc.multi)21:44
prologicmake them normal rc.d services21:44
frinnst <- btrfs magic21:44
frinnstuses the kernel to do the stuff21:44
prologicgood lord udev just feels like a DSL to me21:46
frinnsta rc.conf.d could introduce all sorts of race conditions, probably21:49
jaegerfrinnst: can you send me the output of 'udevadm info --query all --name /dev/sdwhatever' from a btrfs partition?21:49
jaegeror paste it, whichever21:49
frinnstyou want md done before doing btrfs-scan etc21:49
prologicfrinnst: how so?21:50
frinnstbecause you might have btrfs on an md?21:50
prologicjaeger: this might seem a little silly at first -- but would anyone oppose the idea of including a suitable Docker kernel config or tool/script to enable all the required kernel options with the ISO?21:51
prologicfrinnst: what? the md won't come up before the services are started in rc.d?21:51
jaegerWhat would be the point of having docker support in the ISO? It's not really a suitable live system for running anything inside21:51
jaegerOr do you mean a tool that specifically checks if the kernel you're about to build has the proper options?21:52
prologicno, not in the ISO21:52
frinnstprologic: the md has to be up before mount -a21:52
prologica suitable config or tool to enable a suitable config21:52
prologicuse case being21:52
prologicboot cd, install, setup docker21:52
prologicready to go21:53
frinnst22:53 - time for bed21:53
jaegerI'd think the appropriate place for that is in your docker ports21:53
prologicfrinnst: hmm ok sounds like udev is a better place for it :)21:53
frinnstwhy do we have to have interesting discussions when i have to go to bed? :)21:53
jaegernight, frinnst21:53
prologicjaeger: maybe you're right21:53
prologicand it already is :)21:53
prologicit was *just a thought821:53
jaegerYou'll have to install those ports to use it anyway, right?21:53
prologicalmost a ... what if I could build CRUX+Docker hosts on the metal straight from the ISO :)21:54
prologiceither docker-bin or docker+go21:54
prologicand lxc from opt21:54
*** Romster has quit IRC22:02
nrxtxwhat debian chose systemd?22:03
prologicthat was yesterday's news22:11
prologicat least for my timezone :)22:11
nrxtxi'm a bit late because i have a lack of free time the last days :/22:11
prologicthis is freak'n awesome :)22:13
prologicI need a pypi feed for "docker" anywhere in the package :)22:13
nrxtxi wonder when people will notify that a single architecture supporting everything will not be a good solution ...22:19
nrxtxprologic: systemd based systems do not run in docker afaik?22:20
prologicno they do22:21
prologicany init system will22:22
prologicyou can fully boot up a full distro inside a docker container22:22
nrxtxi know you can boot a full one, but my last information was it does not work out of the box when using systemd22:25
nrxtxthat was also a reason why the docker archlinux container uses initscript -> Line 4022:27
nrxtxah the new ones switched22:29
nrxtxseems like they solved the problems by patching the hell out of systemd22:29
prologicyeah no22:30
prologiclike I said this works fine "now"22:30
prologica few versions of Docker back they were bind mounting their dockerinit to /init in the container22:31
prologicand not unmounting it after bootup22:31
prologicbut now they bind it to /.dockerinit22:31
prologicso any init system should just work22:31
nrxtxi hope all that systemd stuff results in a group of developers writing another "modern" init system with unix flavor :)22:35
prologicYeah I believe it's called sysvinit22:42
prologicor runit22:42
nrxtxif you need a init system influencing anything from boot, log, service managment, hardware management up to desktop environment and does not allow you to easily replace single parts and they all depend on each other you can also use windows, not a good unix spirit in my eyes22:45
nrxtxenough for today on that topic22:45
nrxtxmakes me just angry22:45
nrxtxprologic: already plans when adding go to contrib?22:46
jaegercentos6/rhel6 call vgscan via initscripts23:21
prologicI completely agree23:25
prologicit's completely against the grain of a componentized system23:25
prologicala UNIX23:25
prologicjaeger: cool :) so my suggestion wasn't so bad :)23:26
prologicnrxtx: re go? eh?23:26
prologicit and docker ports are already in contrib23:26
prologicas of last night23:26

Generated by 2.11.0 by Marius Gedminas - find it at!