IRC Logs for #crux-devel Thursday, 2012-04-05

*** mike_k has joined #crux-devel02:23
juegood morning02:25
juefrinnst: already fixed with 3.9.5, which is our current version02:26
frinnstare you sure? 3.9.5 dates back a while. 3.9.6 is from feb i think02:33
jueyeah, looked at the changelog02:34
frinnstsweet.. got the mail from the debian ml late last night :)02:36
juefrinnst: the relevant part of it ->
frinnstwow, from 2011?02:37
frinnstIts not the same CVE number though02:38
juehmm, so 2.9.6 didn't help here02:43
frinnstNo, as I understand there is no upstream patched & packaged version that resolves it02:43
frinnstthe debian folks also released an advisary for libpng, but reading the libpng website i *think* we are up to date02:44
*** teK_ has joined #crux-devel02:45
frinnstgot the advisory late last night so didnt look into it that close02:45
jueyeah, libpng looks ok02:48
Romstermorning jaeger08:06
jaegerswapping the SSD in my desktop last night was a bit more of a chore than I expected08:21
jaegerI know what to expect for the laptop one today, at least08:22
*** Romster has quit IRC08:30
*** Romster has joined #crux-devel08:33
Romsterhow hard is it to clone the system to the SSD or did you have to reformat install from scratch?08:35
jaegerIt's pretty easy as long as you know what to expect. If you don't do it right, alignment gets buggered08:39
jaegerwhich is VERY bad for performance on an SSD08:39
jaegerdoesn't matter much for platter drives08:39
jaegerI was cloning from a 120GB drive to a 256GB drive last night08:40
jaeger1) clone the drive with clonezilla08:40
jaeger2) resize partitions if needed, while preserving alignment08:40
jaegermy gparted CD was too old for step 2, it didn't understand any alignment but by cylinder, which doesn't work for SSDs08:40
Romsteryeah i'm aware of block alignment.08:43
Romsterit's a non-issue mostly on spindle drives.08:43
jaegerThe laptop will be more complicated08:43
jaegerThe desktop only had windows on it and the drive was smaller08:43
jaegerThe laptop runs both windows and crux and the drive is 320GB, moving to 256GB08:44
jaegerso I'm gonna shrink it before I do the clone08:44
jaegerI'm not using anywhere near 256GB currently so it shouldn't be a problem08:44
Romsterpartclone might be easier... backup restore it knows about the FS so should align itself?08:44
jaegerit uses partclone08:44
jaegerclonezilla is just a collection of things like partclone, ntfsresize, etc.08:45
jaegerit's pretty nice08:45
Romsterah k, i'm aware of clonezilla but i haven't used that one personally.08:46
jaegerThere are a million cloning solutions out there, it just happens to be one of the best free ones08:46
jaegerThe SSD even came with a copy of norton ghost but I was already familiar with clonezilla so used it instead08:47
Romstermind taking a look at my pppoe connect issue in #crux ? i odn't understand why it wont work on kernel 3.3.1 searching the net is no help. and an extra set of eyes wouldn't hurt.08:48
Romstermost issues are my fault :D08:49
jaegerperhaps kernel 3.3.1 is just broken with pppoe?08:49
jaegerI don't use pppoe so I can't test it myself08:49
Romsteri tried 3.2.2 in the past with the same deal.08:49
Romster3.2.1 works fine.08:49
Romstererr 3.1.208:49
Romsterseriosuly wonder if it's a regression or do i need to recompile pppd/iproute2 perhaps i didn't try that.08:50
jaegerWhat's the output from grep PPP linux-3.3.1/.config ?08:50
jaegerI find the diff output from kernel configs a pain in the ass to read08:51
Romsteryeah diff on kernel is a pain.08:51
jaegerDo you get errors when trying to connect or does it fail silently or what?08:51
horrorStruckRomster: no problem with 3.3 and pppoe here08:52
Romsterand 3.1.2 is
RomsterApr  5 23:29:16 fire pppd[1189]: Timeout waiting for PADO packets08:52
RomsterApr  5 23:29:16 fire pppd[1189]: Unable to complete PPPoE Discovery08:52
RomsterhorrorStruck, i think it was you that told me that i didn't need rp-pppoe anymore? i've been using pppd with it's pppoe module for at lest 3 months fine now.08:53
jaegerIt's been like 15 years since I touched PPPoE so I have no idea, sorry... perhaps horrorStruck will have better ideas08:53
horrorStruckRomster: yes that was me. i'm still not using rp-pppoe FWIW08:54
Romsteryea after setting the clamp MTU to MSS it's been fine.08:54
Romsteri can't see anything wrong with that config the PPP stuff seems ok.08:55
Romsteronly other thing i didn't check is if my eth device is present on new kernel.08:55
horrorStruck[community-patches-queue] krb5-1.9.2-1.9.3.patch has been added08:55
horrorStruck[community-patches-queue] freetype-2.4.8-2.4.9.patch has been added08:55
* horrorStruck , spamming your channel since 201208:56
Romsterhehe, i thought of a bot for all my commits but then no one would be able to chat :P08:56
horrorStrucknot a bot, just me :P08:57
Romsterthis is frustrating me why wont it find PADO packets.08:57
horrorStruckno idea about pppoe, i can paste my config if you want. a git bisect may help otherwise08:57
horrorStruckdid you make oldconfig first?08:58
jaegerRomster: make an RSS feed instead, or similar08:58
Romsteryep i did silentoldconfig though09:00
horrorStruckdoes this prompts you on new options as well?09:00
RomsterRSS wouldn't be a bad idea.09:00
RomsterhorrorStruck, yes09:00
Romsterpppd says it's loaded fine09:01
Romsterbut then it spits out "Unable to complete PPPoE Discovery"09:01
horrorStrucki had the same error lately but that was my dsl signal that was down09:01
RomsterhorrorStruck, can you grep PPP in your config?09:02
horrorStrucksure but moving to #crux, see you there09:02
Romstermy connect command is just /usr/sbin/pppd call ispname09:03
horrorStruck[community-patches-queue] chromium-18.0.1025.39-18.0.1025.142.patch has been added <--- fixes 9 CVE09:54
*** Romster has quit IRC09:58
*** Romster has joined #crux-devel10:00
juefrinnst, jaeger: have you seen ticket #765? I've prepared a tzdata port, see TODO28 for more infos. I think we should do that for 2.8, maybe you can try it with your glibc 2.15 ports10:58
jaegerI hadn't looked at it before but seems like a good idea11:02
juebtw, the patch is shameless stolen from our Arch friends :)11:06
*** deus_ex has joined #crux-devel11:36
*** mike_k has quit IRC12:58
jaegerhrmm... cloning the laptop drive properly is going to be way harder13:15
jaegerdue to limitations in parted, it seems13:15
jaegerI uploaded a couple patches to for gtk+ and friends13:48
jaegerno problems on my test box so far but I do keep a fairly minimal system these days13:49
*** deus_ex has quit IRC14:30
horrorStruckjaeger: do you mind if I put your patch on my repo so we keep track of it and centralize patches? so far I've updated older branches of as our glib is too old whereas jue has a newer glib in his repo since a long time.14:34
*** deus_ex has joined #crux-devel14:49
jaegerYou're welcome to do whatever you like with it15:05
*** Tashbear has joined #crux-devel15:20
*** deus_ex has quit IRC15:45
frinnstjue, thanks i'll have a look15:47
*** deus_ex has joined #crux-devel16:04
*** Tashbear has left #crux-devel16:14
*** deus_ex has quit IRC16:32
jaegerAh, I think I've gotten the drive situation sorted out, finally... I may have to restore my crux partition by hand to shrink it properly but that's not the end of the world16:52
jaegerIf this doesn't work I just thought of a way to do it that should :) guess I'll find out16:59
jaegerok, didn't work... on to the next option! except not yet, time to head home17:12
RomsterhorrorStruck, if you ever bump glib please only do so if atk builds fine, infact test it agenst prt-get depinst webkit, (after bumping libsoup as well)19:37
*** tilman has quit IRC19:58
*** tilman has joined #crux-devel20:00
*** tashbear has joined #crux-devel21:34
*** mavrick61 has quit IRC21:35
*** mavrick61 has joined #crux-devel21:36
Romsteroh right that's in core-multilib jaeger needs a bump22:03
*** tashbear has quit IRC22:26
RomsterteK_, gnupg 2.0.1922:36
horrorStruckjaeger: FWIW when i switched to SSD, i just partitioned it before and used rsync -av, worked great and fast.22:43
jaegerI've got a fairly complicated disk layout or I might have done just that22:43
jaegerGPT, EFI, windows+linux+recovery, etc.22:43
horrorStruckRomster: ok I'll have a look at that but I wonder why jue_ never pushed his newer glib, there must be a reason22:43
Romsterbecause it broke atk22:44
RomsterhorrorStruck, i pushed libquicktime update.22:44
Romsterit's nearly pointless updating my ports in your git tree i usually get it done in 24-48 hours after it's reported.22:45
horrorStruckjaeger: i use gpt too (see FS#764 :P ) but you no windows or recovery indeed22:45
horrorStruckRomster: yeah saw that thanks.22:46
jaegerRomster: are you saying iproute2 in core-multilib needs updating?22:46
Romstercorrect jaeger22:46
Romsteri'm randomly skimming over version sort output i just spent some time fixing the branches that i broke earlier.22:47
jaegerI just didn't quite understand by your wording22:47
Romstersorry my wording is crazy.22:48
horrorStruckRomster: it broke older atk, no? i have no issue with current one22:48
RomsterhorrorStruck, this should help you weed out version bumps anything with the 06-Apr I am still parsing the Pkgfiles right now.22:49
RomsterhorrorStruck, hmm ok see if you can bump libsoup as that needs the newer glib then build atk with recent atk gtk cairo pango fontconf22:50
Romsterit failed for me last time. that was like some time last year IIRC22:50
Romsteri could probably add your repo to the version sort too to save you some effort.22:51
jaegerI was able to build all the gtk stuff today without trouble, posted some patches for core and opt for them22:51
horrorStrucki'll try this weenkend Romster22:51
Romsteri see the core and opt diff in your tmp jaeger22:52
horrorStruckin general, are post-install scripts to avoid?22:52
jaegerRomster: yeah, those are the ones22:52
jaegeruploaded them today22:52
Romsterdamn your server wanting to download diff files than treat them as mime text22:52
jaegerhorrorStruck: we generally avoid install scripts in core and opt, it seems22:52
horrorStruckok thanks22:52
jaegerI'm not sure why, though22:52
Romsterdo we i wasn't aware of that.22:52
Romsterbut i don't think i've added any in opt.22:53
horrorStruckthat would solve the need to supply *.modules files in our gtk ports22:53
jaegerThat's my observation, don't know if it's official22:53
Romsteri think it mgiht be do do with pkgadd off the iso issue22:53
jaegeryes, it would22:53
Romsteras post-install would not be there then.22:53
jaegertrue. there was talk in the past of adding some metadata processing but never went anywhere, obviously22:53
RomsterhorrorStruck, don't forget to bump gdk-pixbuf as well when you test new build of everything.22:55
Romsterpretty much just go though all the prt-get deptree webkit for the stable versons.22:56
Romsterit's a bit of a task i might try it later just preocopied with other stuff atm.22:56
Romsterafaik even numbers are stable on download.gnome.org22:57
horrorStruckmy machine is pretty much up to date on the gtk side, i'll try that Romster22:57
Romsteron another note i wonder when we'll start testing python3 and gtk3? i see newer sakura terminal uses gtk3 now.22:58
horrorStrucki think i have a gtk3 port somewhere22:59
horrorStruckwebkit can take advantage of it too IIRC23:00
Romsteryeah it can.23:00
Romsterbut i force it to use gtk2 for the time being.23:00
jaegerthose patches included gdk-pixbuf, for what that's worth23:01
jaegerglib, atk, pango, gdk-pixbuf, gtk23:01
Romsteri'd like to have gtk3 alongside gtk223:01
horrorStruckthe whole family :)23:01
jaegerYou can install both23:01
jaegerI had them both installed when I was testing gnome323:01
horrorStruckRomster: bring it in contrib, no? they can happily live together23:01
Romsteryeah add it in contrib would be nice23:02
Romsteri just haven't had time to test a build of gtk323:02
Romsterthen i can bump webkit to use gtk3 and test out midori23:02
*** c0x has joined #crux-devel23:03
Romsterand bump sakura to the new version23:03
Romsteri feel guilty giving horrorStruck more work when i'm usually the one doing the testing xD23:07
Romsterwell jaeger saved horrorStruck some work.23:08
horrorStruckat first, the idea was to update ports locally updated by a user, to avoid work duplication mainly and so it can benefit to everyone23:11
horrorStruckbut i dont know about the gtk stuff, would be nice to have jue's feedback about holding glib23:12
Romsteri did xorg how i did so i can later merge in updates of tilman's commits and redo the patches rebased on the diff of tilman's xorg to the changes i've done. i fear your way is gonna end in pain, unless you got some strategy planned.23:13
Romsteri think jue is for glib bump provided it don't break stuff. else he wouldnt' of been testing glib, it wasn't committed to core because of atk breakage.23:13
Romsterthings may be fine now though.23:14
jaegerit also adds a dep, libffi23:14
jaegerwhich I probably forgot to include in my diff23:14
horrorStruckbut atk just needs a bump23:14
Romsteryeah nothing a notify commit wont do.23:14
Romsterto let users know.23:15
horrorStruckor that depsysup that was posted earlier, would be nice to have IMO23:15
Romsteri was nearly about to add libffi for some other port to contrib but i decided to hold off.23:15
Romsterdunno about depsysup... how hard is it to see it broke and use deptree to see what new deps arn't installed?23:15
Romsterwe arn't noobs after all this is targeted to experienced users.23:16
horrorStruckit can build but still you're missing some dep, see gstreamer and orc for example23:16
Romsterjsut seems bloat23:16
Romstertrue but then it's the users fault?23:17
Romsterfor not checking23:17
Romsterand most things will work jsut with missing features or slower.23:17
horrorStruckdo you review all ports before sysup'ing?23:17
Romsternope :D23:17
Romsterwell i do look at the ML23:18
Romsterso i genearlly know when something has a new dep.23:18
horrorStrucki do too, that's indeed only useful for those who don't or if it's not posted23:18
Romsterif something breaks i look at deptree to see what i'm missing. if it builds without user may not need that dep or may have --ignored= that dep on purpose.23:19
Romstererr --ignore=23:20
horrorStruckbut --ignore should still be honored, no?23:20
Romsteri wouldn't mind and i would suggest that prt-get emit a warning that there is ports missing for a port but not stop the sysup from working. user can ctrl-c and make corrections as needed.23:21
Romsterwe don't have as a robust add/remove deps as arch does.23:21
Romsterit's expected that the user knows what there doing23:22
horrorStrucka warning is a nice idea too23:22
Romsterbut that is only my opinion23:22
Romsterwhile i'm thinking about it i've been thinking of a different improvement to pkgmk not related to this topic23:23
horrorStruckwhat kind of improvements ooc ?23:24
Romsterstuff like source=( saves as downloads.php?file=foo.tar.bz2 and subsequently wont get extracted, i was thinking of some syntax like this source=(|foo.tar.bz2) so that wget can rename the file then libarchive does it's thing.23:26
Romsteralso something about these files with no versions in them to save into /md5sum/filename so a sysup later can download the new version less filename if it can't find the md5sum/filename file.23:27
Romsterit's irritating having to rm the version less files to get the new one.23:27
horrorStruckwords of truth ^23:27
Romsteri've avoided that in xtables-geoip by hosting them files myself.23:28
horrorStruckyoutube-dl does that too, hate it23:28
Romsterit's even more irritating when i have my own distfiles mirror of ports.23:28
Romsteryes indeed.23:28
Romstersave to md5sum/name would be a easy way to solve that.23:29
Romsterignoring the whole md5 is not secure anymore debate.23:29
horrorStruckwell it's more archive integrity than anything else that is checked if i understand correctly23:30
Romsteron a similar note i plan to add modified time checking in version sort to do something like the md5sum/foo moving of files (that are older than current one) with another front end tool to keep downloading the most recent versions.23:34
Romsteron top of that if i can i'l also check gnupg and hash files if possible.23:35
Romsterbut that'll take more effort.23:35
Romsterto save me messing up i've started to use git for version sort too.23:38
horrorStrucki've followed more or less the move to signed packages in pacman, it seems to be *a*lot* of work23:40
Romsteryeah i can imagine.23:41
Romsterthe gnupg part may be left to a later time. the hashes will do for now.23:41
horrorStruckbbl lunch23:42
Romstersee a bit down the track i'm going to write a spider to troll all the distfiles on the most useful sites, look for hash files time stamps and test if they are original or modified, making a good/bad list of mirrors to parse over. and generate a bunch of metalink files.23:43
Romsterfood is not a bad idea i'm getting hungry too.23:43

Generated by 2.11.0 by Marius Gedminas - find it at!