IRC Logs for #crux-devel Tuesday, 2014-05-13

*** jue has quit IRC00:19
*** jue has joined #crux-devel00:26
*** pitillo has quit IRC01:26
*** pitillo has joined #crux-devel01:31
teK__jaeger: jfyi: I cherry-picked grub2-efi from 3.0 to 3.107:23
teK__and I just figured, that we are not hit by CVE-2014-0196 :}07:25
teK__btw jaeger will 3.12.7 really be the kernel on the ISO? It's vulnerable to that local root exploit, too08:20
frinnst3.1/compat-32/openssl-32 is still broken btw08:43
teK__NEW       drwxr-xr-x      root/root       usr/share/08:47
teK__NEW       drwxr-xr-x      root/root       usr/share/man/08:47
teK__NEW       drwxr-xr-x      root/root       usr/share/man/man1/08:47
teK__NEW       -rw-r--r--      root/root       usr/share/man/man1/clang.1.gz08:47
teK__=======> Building '/usr/ports/packages/llvm#3.4.1-1.pkg.tar.gz' succeeded.08:47
teK__fakeroot pkgmk  4989.63s user 123.87s system 560% cpu 15:11.83 total08:47
frinnst=======> ERROR: Building '/usr/ports/pkg/llvm#3.4.1-1.pkg.tar.gz' failed.09:23
frinnstMISSING   lrwxrwxrwx      root/root       usr/lib/ -> /tmp/llvm-work/pkg/usr/lib/libLLVM-3.4.1.so09:23
frinnstnice :)09:24
teK__mental note to self: change $workdir to something frinnst-insulting09:26
teK__10:06 < teK__> really.. upstream is driving me crazy09:44
frinnstconfigure: WARNING: Not building nmap-update because libsvn1 was not found09:45
frinnst--without-subversion probably09:49
frinnstunless we really want it. no clue what it does09:49
teK__update nmap from their repo09:49
teK__so its totally useless  ;)09:50
teK__let's see if nmap fixed parallel builds yet :-)09:52
teK__they did09:54
teK__hey frinnst, what's the smartest way to get something off a tap (with tar) and pipe it through the network (via netcat) without creating temp files on the source/tape system?10:03
Romsternetcat comes to mind10:22
Romsteroh you said netcat... hmm i had a netcat pipe ssh command somewhere10:25
teK__the point is that I have to do tar -xf /dev/nst0 because dd and cat wont work10:26
teK__and tar cannot create archives  from stdin(files)10:26
Romsterwhy do you need todo this for?10:29
teK__productions system + tape ---> backup system10:31
teK__easy ;)10:31
Romsterthere are programs for tape archiving.10:31
teK__oh wow :-)10:32
teK__tar has been the only one to work10:32
teK__depends on thetape/driver10:33
Romsterbut i guess don't want to install any.10:33
jueteK__: is there a particularly reason why you didn't use the merge feature of git?10:53
teK__tig release 1 vs 2 so I used cherry-pick11:01
teK__will this be an issue for future merges?11:03
teK__I have changes to merge, what's the policy?11:15
frinnstjust pull from 3.011:27
teK__I did git merge 3.0 (while 3.1 is checked out)11:28
teK__I'm asking because I screwed opt before 2.8 -> 3.011:28
frinnstdo from 3.1: git pull . 3.011:29
frinnstthats what I do atleast11:29
Romsterthat's what i do and then fix any rebasing issues.11:29
Romsterif any.11:29
Romsterjust make sure you check what branch your on and what your pulling from.11:30
Romsterelse you could make a terrible mess.11:30
teK__I did11:31
Romsteri screwed my git tree up once that bad instead of pushing the crap i checked out git again and redid my commits.11:31
Romsterit's wise to rebase on git checkout 3.1, before git pull . 3.011:32
teK__I did git pull before git pull . 3.0 which should be the same?11:32
Romsteri like using gitk --all11:32
frinnstdont pull from, use fetch and rebase11:33
frinnstiirc just pulling from upstream will fuck shit up11:33
Romstereh i prefer doing git checkout 3.1 ; git fetch ; git rebase origin/3.1 ; git pull . 3.011:33
juesorry, was AFK11:33
Romsteri don't like using git pull personally.11:33
teK__I assume pulling would not fuck things up, as git status says:11:34
teK__On branch 3.111:34
teK__Your branch is up-to-date with 'origin/3.1'.11:34
Romstersure you could do git pull --rebase but eh....11:34
teK__but that may be a misconcetion11:34
Romsterdepends on git pulls strategy or something.11:35
RomsterSee pull.rebase, branch.<name>.rebase and branch.autosetuprebase in git-config(1) if you want to make git pull always use --rebase instead of merging.11:35
Romsterwell you could set that then it be safe to use git pull.11:36
Romsteri never bothered too.11:36
jueif you have nothing committed locally, it's save to do a git pull, otherwise use fetch/rebase11:36
Romstergit pull can merge otherwise.11:36
Romstergood morning jue11:37
teK__that's in opt/llvm (there are conflicts for llvm, which is ok)11:37
Romsteri did mess up contrib ages ago, forgot how i did that...11:37
Romsterjsut edit the files it's pretty easy to see the merge lines that need to be removed then git add them files.11:38
Romsteronce merges are fixed git add, git rebase --continue or what ever it says needs to be done in. it'll give you suggestions11:39
Romsteronce t's all checked and fine git push and hold your breath lol11:39
frinnstgit status; git log; git push :)11:40
Romsteri honestly should stop adding ' to its11:40
teK__I locally deleted && recreated 3.111:41
Romsterapparently English does not require it. even though it is.11:41
jueteK__: hmm, just fix the llvm conflicts and run git commit, that's all11:42
teK__I ran git merge 3.0 in 2.8 back then11:43
teK__i.e. the other way round11:43
Romsteri guess git is a hard tool to get to use.11:43
teK__na I'm pretty satisfied :)11:44
juewhat we need isn't complicated at all11:44
teK__yes and commits are even not cross-directory :)11:45
Romsteri've grasped how to recover from a failed merge.11:45
jueimportant is only to follow our merge strategie: 3.0 -> 3.111:45
juewe did it the other way arounf in the past, but it's better to merge form old -> new11:46
Romsterdoens't make sense to do it the other way though i have cherry picked commits to put in new -> old in some cases.11:46
teK__learned that the hard way11:46
Romsterit often makes a mess.11:47
Romsterif done wrong.11:47
frinnst [ANNOUNCE] xf86-input-synaptics 1.7.611:53
frinnstand 1.8.0 too11:53
frinnstbtw Romster, are you able to fix compat-32/openssl-32 for 3.1 ?11:56
frinnstpatch seems broken11:56
frinnstfor perl11:56
frinnstcore/openssl patch should work just fine11:57
juefrinnst: looks like 1.8 needs a new lib -> libevdev11:59
Romsteri thought i did fix that.11:59
juefrinnst: probably instead of mtdev?12:00
juehint: use the -x option with cherry-pick ;)12:13
teK__wrt cherry-picking vs merging12:15
teK__I did the merge in contrib12:15
teK__and all cherry picked commits appear twice now12:16
teK__sorry for cluttering the history.12:16
Romsterfrinnst, fixed.12:16
teK__this is going to happen in opt, too. Is there a way to avoid that?12:17
jueteK__: stop !12:18
jueteK__: if you are unsure I can do the merge in opt for you?12:18
teK__still.. is this supposed to happen?12:20
*** frinnst_ has joined #crux-devel12:20
teK__well that's funny because I did:12:20
teK__git checkout 3.112:20
teK__git pull . 3.012:21
teK__I need to push another change to 3.0. After that we/I/you could merge in 3.1 (with conflicts because llvm changed because of multilib and clang things)12:24
teK__just for the record: committing this:
jueok, I'll push the merge12:33
teK__please paste a log  of what you did12:33
juenothing fancy: git checkout 3.1 ; git pull . 3.0 ; (resolve the conflicts with llvm) ; git add llvm ; git commit ; git push12:35
teK__but dovecot, fcron etc. don't show up as merged from 3.012:36
teK__is that because they were not or gitweb sucks12:36
juenow go and add your last symlink fix to 3.112:36
juethey were, but gitweb doesn't show the merged files in the commit message12:38
jueuse tig to see it :)12:38
juethe commits a there in gitweb, see dovecot for example12:39
teK__but only because of your recent pull/push12:40
teK__I'm asking because I think I did the exact things in parallel (for testing)12:40
teK__still  I'm not sure why the commits show up twice in contrib12:40
jueyou mean in gitweb?12:43
teK__and git log12:43
juehmm, not in git log resp. tig for me ?12:44
jueI see the same commits in 3.0 and 3.112:45
teK__3.1 contrib?12:45
teK__ok wtf :D12:45
teK__you did git pull recently?12:45
jueI see the same commits in 3.0 and 3.1, guess that are your cherry-picks12:45
jueyes, my repo is up-to-date12:46
jue14:13 < jue> hint: use the -x option with cherry-pick ;)12:46
teK__rebuilding llvm the 10th time12:47
teK__after that I'm going to push12:47
teK__so beware and watch after your children12:47
juejust rebuild it on 3.0, still missing all *.a files12:47
teK__so these are missing? -rwxr-xr-x  root/root usr/lib/libllvm.a   etc..12:49
teK__so they seem to belong to ocaml, too…12:50
teK__i.e. another two commits :}12:50
juefrinnst: yeah, I compared 1.7 -> mtdev ; 1.8 -> libevdev13:12
jueso we should probably use 1.7.6 for 3.0 and 1.8.0 for 3.113:14
Romsterjesus tek your commits on llvm13:22
Romsteryou ddin't read that --enable-experimental-targets=R600 is now a default and not experimental anymore.13:24
Romsterin the changelog13:24
*** pitillo has quit IRC13:27
*** pitillo has joined #crux-devel13:29
*** sammi` has quit IRC13:40
*** pitillo has quit IRC13:41
jaegerteK__: 3.12.19 is available now13:52
teK__ah ok :)13:53
Romsterdunno if --enable-libffi is of any value to llvm?14:01
Romsterits disabled by default14:01
teK__file a bug, I dont care :P14:04
Romsteri dunno if libffi is even useful on llvm.14:04
Romsteryour the llvm expert14:05
teK__" "14:06
teK__Call me Lord of the Branches14:06
teK__oh btw folks.. during the last Updaterun on CRUX I set the default HEAD for our ports  to 3.0 to make them appear in the updated gitweb. This has the (positive) side effect of not having to manually checkout / create a branch after the initial cloning14:14
teK__git clone git:// && cd opt && git branch14:14
teK__* 3.014:14
Romsteri'll do some testing but i can't test the r600 stuff due to no hardware14:14
teK__the wiki says to do git branch x.y origin/x.y ; git checkout x.y which is not required any more14:15
Romsterhe never used HEAD at all...14:15
Romsterand i've always complained that but it's in 3.0 branch and that's what gitweb is tracking.14:16
teK__ <14:16
Romsteryes read that a million times14:16
frinnsthey you found my cheatsheet! (I always forget how to work on two branches between releases)14:16
teK__I'm quite sure that you were on master without any branch created/checkedout before.14:17
teK__And I was right.14:20
teK__git clone git://  etc.14:20
teK__git clone git://  etc.14:20
teK__xfce will put you in (the empty) master branch14:20
teK__so the branch/checkout step from the wiki is obsolete for {core,opt,contrib}14:21
teK__which reminds me that I have to change default HEAD for xfce :-)14:21
frinnstwow, gtk3 3.12 really fucked up themes14:23
frinnstwell to be honest, its not worse14:24
teK__bad in a different way? :D14:24
frinnstnah, just a different kind of shit :)14:25
frinnstsame degree on the bristol scale14:25
Romstergreat so i have todo xorg and compat-32 now too?14:27
Romsterthemes never worked in gtk314:27
Romsteri see lxde have moved to qt14:27
Romsterciting gtk3 is crap...14:28
frinnsttek i still get a footprint missmatch with llvm14:29
frinnstsorry, my tree was not up to date14:30
teK__don't you make me cry :D14:31
Romsterllvm footprint is fine.14:46
Romsterjust built it.14:47
frinnstyeah built fine this time14:50
Romstererror: 'DisablePrettyStackTrace' is not a member of 'llvm'14:55
Romstergreat now mesa3d broke....14:55
juefrinnst: you read my synaptics comments?15:04
frinnstnot really :)15:04
frinnstah, yeah sounds like a plan15:04
frinnstwould be nice to switch branch before 3.115:04
teK__hey xorg team, should we add another commit for llvm? D:15:05
frinnstgo for it15:05
teK__in the sense of downgrading as mesa 9.x does not  compile anymore15:05
frinnstdowngrade for 3.0 only15:06
frinnstno real reason to switch stuff around there anyways15:06
teK__it's my lucky day :D15:06
Romsteri havne't tested 3.1 for mesa3d yet15:07
teK__I don't think it's going to work there, either15:07
frinnsti run 3.1 at home with gallium drivers. will give it a spin in a few minutes when i get home15:07
teK__ -> With the updates from upstream mesa a few days or weeks ago it does now. Doesn't really work, but it compiles.15:07
teK__that should mean 10.x I guess.15:08
frinnst[ANNOUNCE] X.Org Security Advisory: Multiple issues in libXfont15:10
Romsteroh for god sake...15:10
frinnstFixes will also be included in these module releases from X.Org: libXfont 1.4.8 libXfont (1.5.0 RC 1)15:10
Romstertime to bump that..15:10
frinnstbunch of CVE's for qemu too15:11
Romsterjust bump the version to 1.4.8 and 404 -_-15:12
frinnstFixes *will* also be included15:12
frinnstas in "not released yet"15:12
frinnst*heading home*15:12
Romstersomeone else will have todo it... when it's released.15:13
Romsterunless i'm around when its out15:14
teK__3.1 is on mesa3d-10.x. Compiled for me with llvm-3.4.115:30
Romsteri didn't get that far got distracted...15:34
Romsterso i need to backport a patch for mesa3d i guess?15:35
teK__I'll wait for someone to confirm mesa3d+llvm for 3.1 and then roll-back in the 3.0 branch.. woah.15:35
jaegerI just rebuild mesa3d on my 3.1 ISO VM, wasn't a problem.15:35
teK__k thx15:35
jaegercan't test 3d with it in that VM but it built without issues15:35
jueFYI, I've committed the update to 1.8 for synaptics to 3.1, we have to keep mtdev for now because input-evdev still depends on it15:41
juejaeger: ok if I commit the required stuff to iso.git?15:42
jaegerfor that synaptics update?15:43
jaegernot at all :) Just wanted to make sure I was understanding15:44
jueok, will do :)15:44
jaegerI should push my local updates, too, will worry about that after yours15:45
jueand added to TODO31 for completeness15:56
frinnstim updating llvm now. I currently run mesa from git master. will give that a spin as well as the "old 10.x" branch :)16:14
frinnstdoh, mesa needs a rebuild16:24
frinnstno vdpau for now :(16:24
frinnstrebuilt & works16:28
frinnstthis was git master, ill try with whatever is in xorg next17:30
*** pitillo has joined #crux-devel17:48
*** pitillo has quit IRC18:22
*** pitillo has joined #crux-devel18:24
*** pitillo has quit IRC19:04
*** Amnesia has quit IRC19:44
*** mike_k has quit IRC19:44
*** Amnesia has joined #crux-devel19:55
*** pitillo has joined #crux-devel20:04

Generated by 2.11.0 by Marius Gedminas - find it at!