IRC Logs for #crux-devel Wednesday, 2009-05-13

*** sepen has joined #crux-devel01:24
*** mike_k has joined #crux-devel02:05
*** sepen has quit IRC02:10
*** Rotwang has joined #crux-devel02:35
*** mike_k has quit IRC02:43
*** sepen has joined #crux-devel03:07
*** mike_k has joined #crux-devel03:52
*** Rotwang has quit IRC04:26
*** jtnl has joined #crux-devel04:48
*** jue has joined #crux-devel05:10
jtnlhi jue05:11
juejaeger: can confirm your report of the /etc/securetty problem, it's even more ugly because I get a crash of login after the attempt to login as root05:14
juecan someone with shadow 4.1.4 please test the following:05:14
jue1) open a console 2) login as root05:15
juethat should fail, now press ENTER again05:15
pitiIIolet's see if I can test it under the VM05:17
pitiIIoummmm strange, I have the vmware console gray and freeze05:18
pitiIIoI can't check it in another way atm05:19
juepitiIIo: thanks anyway05:22
pitiIIoI think the only way to get it working again is to reboot it, but I think I must review what happened there, it has a good uptime and very good performance, really very strange05:36
juepitiIIo: try to login on another console as a normal user, that sould work05:46
sepenor call to init by using telinit instead of reboot05:57
sepenjue, no problems here with shadow inside a safe-environment05:59
sepenerr, glibc backtrace ;D06:00
juesepen: thx, same here06:05
sepenoh np, but no idea what its the cause06:06
juewell, it's not the first regression added by the debain maintainer ;)06:09
pitiIIoummmm I can't get the console in vmware (grey screen there) and no way to change between ttys, I'm in by ssh06:11
pitiIIoI will reboot it and check it again (I can't see anything in the VM)06:11
sepenpitiIIo, just use a safe-env. 1 execute /bin/login, 2 login as root (it fails), 3 (press enter)06:13
sepenbut the report I think that will be also positive ;D06:13
jueit's a heavy bug at all, I tend to downgrade to the old version for now06:15
pitiIIoummmm good point sepen, I wanted to test it in a tty directly06:17
pitiIIorebooted and checking it now06:18
sepenpitiIIo, you can use a serial console too, just take a look of /etc/securetty06:18
pitiIIoummm well, here crashed too06:18
sepenjue, +106:32
juepitiIIo, sepen: thanks for the help, link to my bug-report is in the commit message07:17
jueaon: libcap refused to build with gcc 4.4 ->
aonthink that's a problem with the current headers07:56
aonbut i found a patch07:57
jueyeah, you're right, not a gcc problem obviously08:02
*** jtnl has quit IRC08:03
aonjue: try now08:08
aoni should probably upgrade glibc to investigate that further :)08:24
sepenjue, are binutils, libgmp, and libmpfr compiled before glibc?08:40
jueif you want a full bootstrap you must build all of them at least 3 times08:42
sepen3 times?, just I eared about 1 for toolchains08:43
juewell, probably not08:44
*** deus_ex is now known as pedja08:44
jueif you have two pieces each dependent on each other you need a min. of two runs, that's obviously and true for gcc + binutils08:51
juebtw, our ISO is build that way, three runs08:53
sepeniirc I read that you need only 2 steps for the complete toolchain08:53
sepenjue, three runs but the last one chrooted, right?08:54
jueIIRC the last two08:55
sepenwell I think the last one its required only for building ports, and not for the toolchain08:56
aonis it ok to drop ant, bftpd, cdrdao and opera from opt?08:58
aonsince they're duplicated in contrib08:58
jueold ports from sip?09:00
sepenwe have an alternative port?09:00
aonbanana[~/dev/crux/git]$ ls -d {opt,contrib}/bftpd09:01
aoncontrib/bftpd  opt/bftpd09:01
sepenohh pureftpd, xD09:01
aonand vsftpd09:01
aoni don't understand why anyone would want to use anything else than vsftpd09:01
sepenaon, yes its duplicated. I added it as **d** to the orphaned list, also I think I'm the pseudo_official orphaned-list-maintainer ;D09:02
juewell, pure-ftpd has a lot more options09:02
sepenI'm sure we have a lot of things for a next casual meeting09:03
sepenhmm and finally I think I'll adopt adobereader if no one does it before, so it's time to remove 'the rest' of unmaintained ports from opt09:05
jueyeah, I'd say go and remove the orphaned opt ports09:05
sepenonly one thing: the pkg-get* stuff. any idea if simone plains a new release?09:06
juedon't think so, but would be nice to keep that in opt09:06
aonthere you go :)09:06
jueyeah, jue has too few ports anyway09:08
jueaon: how about you? see it as a tribute to sip :-)09:09
aonnever even used it :)09:10
sepenwell no idea if pkg-get works fine09:10
jueyes, it does09:10
*** pedja has quit IRC09:10
jueit's pretty cool to distribute packages within a local network09:11
sepenATM I'm using prt-get + pkgmk-patched to obtain a similar behaviour for pkg's installation only09:11
*** pedja has joined #crux-devel09:11
sepenall you need is a nfs exported directory with pkg's09:11
juewith pkg-get yopu need nothing as an ftp or http server09:13
sepenbut you need to create a REPO09:13
jueyeah, sure :)09:13
sepen^^ sorry for some bad words (like 'odd'), my bad english sucks sometimes09:14
sepenwell, always, ;D09:14
sepenwith this method I'm sure that I'm using the prt-get's dependency algorithm, no idea if pkg-get uses the same09:16
*** jesse_ has quit IRC09:46
sepenaon, ping10:06
sepenthe ports you removed were in 2.6 branch10:06
aonyeah, i know10:07
sepenand 2.5?10:08
aonyeah, i guess it should be done there too :)10:08
sepenwell I'm removing them from the orphaned list10:09
sepenbut still requires to push the same changeset you made in 2.510:10
sepenah ok10:10
sependone, I'll wait until the weekend, if no one adopts some of 'the rest' I'll remove them10:12
*** sepen has quit IRC10:26
*** prologic has quit IRC10:53
*** prologic has joined #crux-devel10:58
rehabdolltilman: how can i create a 2.6 branch of opt-x86_64 without ruining everything? :)11:06
tilmangit fetch11:21
tilmangit checkout 2.511:21
tilmangit rebase origin/2.511:21
tilmanat this point, your local 2.5 branch is up to date wrt the remote end11:21
tilmangit branch 2.611:21
tilmanthen just: git push origin 2.6:2.611:22
tilmanrehabdoll: just be careful to not rebase 2.5 onto 2.6 or vice versa :D11:23
*** sepen has joined #crux-devel11:33
aonisn't just 2.6 the same as 2.6:2.6?11:35
tilmanaon: not if that branch doesn't exist on the remote end11:36
aonah right11:36
aontilman: i think you should bump nspr in 2.611:37
tilmandoes gcc 4.4 imscompile the current one11:39
tilmanhi jue11:39
aonbut ff is all messed up with it11:39
tilmanare you sure 2.5 isn't affected?11:39
juetilman: core builds smooth so far, the only failed port is httpup11:40
aonno, but it appeared after i built xorg from the 2.6 branch11:40
tilmanjue: the usual c++ build problems? :)11:40
aonrehabdoll too11:40
tilmanaon: i'll bump it in both branches11:40
aonyeah, there's probably no harm11:41
tilmanjue: i'll fix it11:42
juetilman: let's create a ticket for cptn ;-)11:43
tilmanjue: yeah. would be nice to get a tarball out11:44
juedone, ticket #44711:47
tilmanjue: i'll patch core/httpup anyway, ok?11:47
jueyes, sure11:48
tilmanugh, running openrdate just once a day isn't enough for my laptop's rtc :|11:51
juetilman: your opinion on the rdate->openrdate switch?11:56
tilmanjue: there's no downside to switching afaics :)11:56
mike_kchronyd works for me11:56
juetilman: ok, I'll do it tomorrow11:57
juemike_k: yeah, chrony is cool11:57
mike_kyeah, thanks for the port11:58
aonjue: hehe11:58
tilmanjue: it would be nice if we could document that the -a switch will not fix the clock at once, but gradually11:58
juemike_k: it's pretty amazing how fast it syncs the clock11:58
tilmanjue: i'm not sure how to do that though11:58
tilmanhaving a README just for that seems overkill11:58
tilmanmaybe add that to the FAQ in the handbook...11:58
juewe can comment the cron-daily file that comes with the port?12:00
tilmanjue: what do we do about xz? the final stable release 5.0 still isn't there yet :(12:05
jueyeah, that's odd because everything seems stable so far12:07
tilmanof course we could add 4.99..99 to ports12:09
tilmanand try to test the installation really well12:09
tilmanespecially if we use xzipped tarballs on the iso instead of gzipped ones like thrice suggested :)12:09
aonare we on the verge of hitting some limit of space on the iso?12:10
tilmansmaller iso = better iso though12:10
aonare we on the verge of hitting some limit of quality? :)12:11
jueI saw that suggestion but not sure how it should work. We need xz support for pkgadd, pkgmk should have a option to produce .xz instaed of .gz. Or do I miss something12:12
aonie. can't it wait until 2.7 :)12:12
tilmanat least pkgadd needs to be changed, yes12:12
tilmanfor the iso, we might have a hack in the makefile that repacks the tarballs12:12
tilman(or maybe not :D)12:12
sepenbut we don't have a deadline for 2.6, right? so can be tested for a long12:13
juesounds a bit like a hack to ;-)12:13
tilmanit's awesome how everybody seems to play with the 2.6 stuff atm12:13
aonsepen: yeah..12:13
tilmanbut i don't think we'll release soon :D12:13
sepenjust to anticipate the work12:14
juetilman: was my impression as well, everything is eager to do something wrt 2.6 :)12:14
tilmanre. xz again:12:17
tilmanit would be good to have a port for xz at the time 2.6 is released, so we can update it 5.0 _before_ 2.712:18
tilmanso sometime between 2.6 and 2.7, people will be able to use xz dist archives in their ports12:18
tilmanadding new core dependencies in between releases will fail anyway (remember the libarchive mess?)12:18
jueok, I'd suggest to add the xz port and add support to pkgmk12:20
*** Rotwang has joined #crux-devel12:21
tilmanjue: awesome. some ntp server disagrees with re. the current time. the difference is 20+ seconds o_O13:05
tilmanjue: i'm testing the local openrdate vs the ntpd on crux.nu13:11
tilmani just pointed at se.pool.ntp.org13:11
tilmanand i'm now running rdate -ncv as well13:11
tilmantime diff still ~20s13:11
tilmansame when using some specific ip (
*** treach has joined #crux-devel13:13
juehmm, got the same, but if I omit the -c option the result is the same as with msntp13:14
tilmanoh boy13:16
tilmanjue: the information on -c in the man page rules13:16
tilman"check against a local radio controlled watch"13:16
tilmanawesome o_O13:16
jueyeah, but works for me ;-)13:17
juetilman: please have a look ->
tilmanworks for me as well13:18
tilmanjue: yep, that's great. s/systemcall/system call/ though :)13:19
tilmansoooo. should -c be the default? :)13:19
juedon't think so13:20
aontilman: for example :)13:21
tilmani could use my freerunner's AGPS ;)13:22
aoni wonder where my usb gps receiver is13:23
aoni got one and put the project i need it for "on hold"13:23
aonhave used it probably once or so :)13:24
tilmanaon: so my run-of-the-mill radio controlled clock uses DCF77?13:25
aoni'd imagine so13:26
tilman :17 Z1 Set to 1 when CEST is in effect.13:27
tilman :18 Z2 Set to 1 when CET is in effect.13:27
tilmanuhuh. i hope there's a reason they are wasting one bit there ;)13:27
tilmanaon: what purpose did DCF77 serve _before_ time and date data was added in 73?13:29
aon"DCF77 has been in service as a standard-frequency station since 1959;"13:30
tilmansounds like they just emitted 77.5 khz signal. woo.13:30
tilmanaon: yes. what does that mean/what's the use?13:30
aonyou can calibrate other devices when you know that dcf77 is exactly 77.5kHz13:31
tilmanmmmmh ok13:32
aoni've used rwm for that though13:33
juetilman: objections against the libusb update? Works for me, but have only a camera (gphoto2, gphotofs) and a scanner (sane) to test13:44
tilmannot sure i have any device that uses libusb13:51
jueI think it's save to commit the changes so people can test it, if something goes wrong we can replace libusb-compat with the old port, called libusb-legacy maybe13:55
aonfyi: readline installs info to /usr/share/info instead of /usr/info now14:15
aonso the rm -rf line needs to be changed14:15
tilman-f considered harmfu!14:15
aonoh, it's even in the footprint14:16
jueoops, shame about me14:17
jaegerwhen git says I need to set push.default, should I set it to 'tracking'?14:19
tilmanjaeger: i think 'matching' is okay14:19
jaegersince my local 2.5 is tracking origin/2.514:19
tilmanmatching was the old default value14:19
jaegerthat's what it says is default, figured I'd ask before breaking something :)14:20
jueaon: thanks14:21
juetilman: agree, forgot to remove the -f, in most core ports I've done it14:23
tilmani was bitten by the same problem more than once, too14:23
*** jesse_ has joined #crux-devel14:30
juejaeger: you've read what I said to you some hours ago wrt securetty?14:33
jaegerjue: yeah, I saw it, sorry, thought I had responded14:35
jaegersame behavior here14:36
jueit's strange that such obvious bugs pass their tests14:40
jue.oO and my test as well ...14:41
*** mike_k has quit IRC15:01
tilmanjue: i guess you'd rather not update util-linux-ng in 2.5? seems like the current udev might not work with current (2.5's) util-linux-ng15:04
tilman(see #crux; udev removed the vol_id program)15:04
*** jue has quit IRC15:47
*** Rotwang has quit IRC16:41
*** sepen has quit IRC17:39
*** treach has quit IRC18:19
*** prologic has quit IRC20:03
*** teK has quit IRC20:03
*** pedja has quit IRC20:03
*** pitiIIo has quit IRC20:03
*** nipuL has quit IRC20:03
*** rehabdoll has quit IRC20:03
*** j^2 has quit IRC20:03
*** jesse_ has quit IRC20:03
*** prologic has joined #crux-devel20:03
*** pedja has joined #crux-devel20:03
*** pitiIIo has joined #crux-devel20:03
*** j^2 has joined #crux-devel20:03
*** nipuL has joined #crux-devel20:03
*** rehabdoll has joined #crux-devel20:03
*** teK has joined #crux-devel20:03
*** jesse_ has joined #crux-devel22:33
*** jesse_ is now known as Guest8891722:33

Generated by 2.11.0 by Marius Gedminas - find it at!