**** BEGIN LOGGING AT Sun Aug 27 02:59:57 2006 Aug 27 06:26:25 03bzhou * r3958 10optware/trunk/make/eaccelerator.mk: eaccelerator: trigger a rebuild to match php 5.1.6 Aug 27 07:24:19 03bzhou * r3959 10optware/trunk/make/py-turbogears.mk: py-turbogears: upgrade to 0.9a7 Aug 27 11:19:37 03gda * r3960 10optware/trunk/ (5 files in 2 dirs): cyrus-imapd: 2.2.12 with patches from the University of Athens Aug 27 11:26:35 03gda * r3961 10optware/trunk/make/cyrus-imapd.mk: cyrus-imapd: inc version Aug 27 11:53:54 re Aug 27 11:54:58 have running now squirrelmail on my ds101g with lighttpd, php-fcgi and cyrus-imapd :) Aug 27 11:55:34 package for squirrelmail will come soon, the other ipks are all from optware repos Aug 27 11:56:43 hi # Aug 27 11:57:06 03gda * r3962 10optware/trunk/ (make/cyrus-imapd.mk sources/cyrus-imapd/imapd.conf): fixed dependencies and bug in config Aug 27 12:04:03 Could someone explain me what the meaning of amber/red/out flashing means for DebianSlug ? Aug 27 12:04:27 I added a 2nd hdd to the 2nd USB port and it doent boot anymore :-/ Aug 27 12:37:46 bbl Aug 27 13:03:41 i have problems german locale in, let's say, the tin usenet reader. I have installed the glibc locale files, I have set LANG=de_DE (tried different variations as well), i even copied /usr/share/locale from another host, but still i'm getting garbage instead of real Umlauts in the terminal. Any idea? Aug 27 13:04:48 mickley: Insead of cp locales tried dpkg-reconfigure locales ? Aug 27 13:05:10 ah, sorry, i should've said, I'm using OpenSlug, not DebSlug Aug 27 13:05:56 mickeyl, is there something like dpkg-reconf ? .oO(ipkg-reconf ;-) ) Aug 27 13:06:04 yeah, ipkg configure locales Aug 27 13:06:07 "Nothing to be done" Aug 27 13:06:21 ok, reconfigure is it. Aug 27 13:06:23 no Aug 27 13:06:48 is there anything i can do manually? Aug 27 13:07:15 mickeyl, see http://wiki.nslu2-info.de/pmwiki.php/HowTo/DebianSlugUmlaute Aug 27 13:08:37 Sinclair73de: thanks, I'm afraid it doesn't work like that w/ OpenSlug. there isn't a "locales" packages in the ipkg feed, perhaps that's the problem Aug 27 13:08:50 could you give me a list of files included in the locales packages on a debian system? Aug 27 13:10:29 have u the export LC_ALL-line in /etc/profile? Aug 27 13:12:09 yes Aug 27 13:13:34 interestingly, i can see öäü on the console Aug 27 13:13:39 it's just not working in applications Aug 27 13:14:04 well... tin at least Aug 27 13:14:57 let me see what samba says Aug 27 15:59:39 03bzhou * r3963 10optware/trunk/make/ruby.mk: ruby: upstream upgrade to 1.8.5 Aug 27 17:08:26 03bzhou * r3964 10optware/trunk/ (3 files in 2 dirs): cyrus-imapd: replace outdated config.{guess,sub} to recognize x86_64 host Aug 27 17:37:59 eno-away: are you really away? I have some questions. Aug 27 17:41:18 gda_, just got back Aug 27 17:42:16 actually i'm wondering cyrus-imapd seems to x-compile just fine Aug 27 17:43:34 and i don't know the reason behind r1794 commit of make/cyrus-imapd.mk (removed the strip step) Aug 27 17:44:10 if you can do some runtime testing, we can move it from native to cross Aug 27 17:46:39 currently i don't know what you mean, looking for the ticket Aug 27 17:48:32 of course it x-compiles and I use it currently, my patches are great, no need for the cyradm to create an acount Aug 27 17:48:46 bbl Aug 27 17:49:02 alright, i'm also heading out Aug 27 17:49:24 but currently native cyrus-imapd is broken Aug 27 17:50:05 see http://logs.nslu2-linux.org/buildlogs/autobuild-gastro-last.txt Aug 27 17:50:58 and with-perl it is also broken for nslu2 target Aug 27 17:51:15 x-compiel Aug 27 17:53:13 03bzhou * r3965 10optware/trunk/make/cyrus-imapd.mk: cyrus-imapd: use OPTWARE_TARGET instead of HOST_MACHINE for x-compilation Aug 27 18:00:04 03bzhou * r3966 10optware/trunk/make/cyrus-imapd.mk: cyrus-imapd: try to fix "armv5b-softfloat-linux-gcc: No such file or directory" error in native build Aug 27 18:02:21 why does it work for me? Aug 27 18:05:01 eno_away: would like to add a dir /opt/etc/lighttpd/lighttpd.conf.d to the lighttpd package, for something like "include /opt/etc/lighttpd/lighttpd.conf.d/*.conf" in the lighttpd.conf Aug 27 18:06:36 That would allow me to add symlinks to this dir to conf files for squirrelmail and php-fcgi without the need to edit the conf file of lighttpd, what do you think? Aug 27 18:08:42 eno-away: general question: is it allowed for ipk postinst prerm scripts to change files that dosn't belong to them? Aug 27 18:09:34 eno-away: I ask only because we have a policy in our distri to forbid this Aug 27 19:40:17 a nslu2 user reading? Aug 27 19:40:55 that is using optware? Aug 27 19:41:37 does the nslu2 use shadow for passwords? Aug 27 20:12:58 gda_: adding /opt/etc/lighttpd/lighttpd.conf.d sounds like a good idea Aug 27 20:13:58 gda_: it makes sense to PROHIBIT one ipk's postinst prerm messing with files not belong to them Aug 27 20:14:29 gda_: nslu2 unslung + optware is not using shadow password Aug 27 20:21:58 eno: and I asume they don't use pam, right? Aug 27 20:23:30 no, just plain old passwd Aug 27 20:25:07 so I have to change saslauthd to use shadow on the ds101g and pwent on all the others Aug 27 20:25:42 k Aug 27 20:26:42 gda_: cyrus-imapd x-compilation is not ready, neither nslu2 nor ds101g Aug 27 20:27:25 LD_RUN_PATH="/home/bzhou/slug/optware/ds101g/staging/opt/lib" cc -shared -L/usr/local/lib IMAP.o -o blib/arch/auto/Cyrus/IMAP/IMAP.so ... Aug 27 20:27:34 it should not use "cc" Aug 27 20:29:22 do you have files not checked in yet? with-perl also tries to use (cd $(CYRUS-IMAPD_IPK_DIR)/opt/lib/perl5/site_perl/5.8.6/armv5b-linux/auto/Cyrus ... Aug 27 20:29:47 which should definitely use $(PERL_VERSION) instead Aug 27 20:31:01 03bzhou * r3967 10optware/trunk/make/cyrus-imapd.mk: cyrus-imapd: --with-perl should use $(PERL_VERSION) Aug 27 20:32:54 03bzhou * r3968 10optware/trunk/make/py-mysql.mk: py-mysql: upstream upgrade to 1.2.2b1 Aug 27 20:36:39 eno: sorry, should have checked, I did never use the perl extension Aug 27 20:38:27 the cyrus ipk is not from me, I have thought it was okay and I wanted only to have more functionality Aug 27 20:40:33 please check http://logs.nslu2-linux.org/buildlogs/ whenever you commit a change, to make sure no breakage Aug 27 20:40:44 k Aug 27 20:41:14 how often happens the rebuild? Aug 27 20:41:25 every 15 minutes or so Aug 27 20:41:39 gastro native, and nudi cross Aug 27 20:45:33 eno: with your last changes it even don't build on my machine anymore Aug 27 20:51:33 okay now it uses the right cc, but -march=i586 -mcpu=i686 and the wrong perl version Aug 27 20:51:59 you mean the HOST_MACHINE to TARGET_MACHINE change? Aug 27 20:52:21 that could very well be the problem Aug 27 20:53:06 I don't know which change, that what I checked in was compiling but after I checked your changes iut it doesn't build anymorw Aug 27 20:54:45 It doesn't use hostperl Aug 27 20:58:07 eno: Aug 27 20:58:09 configure: WARNING: *** This system does not have a working mmap() Aug 27 20:58:09 configure: WARNING: *** Expect a considerable performance penalty Aug 27 20:58:37 cross configure still needs more tweak to match native Aug 27 20:58:40 I get this during configure, because of cross of course Aug 27 20:59:07 at least at the ds101g there is mmap() Aug 27 20:59:20 how about the other platforms Aug 27 20:59:22 ? Aug 27 20:59:30 and my $(HOST_MACHINE) => $(OPTWARE_TARGET) change is so that cross matches native Aug 27 21:00:17 i believe all supported platforms supports it, we should force cross configure to use mmap() Aug 27 21:00:57 and if some platform does not support it, we'll get bug report later on and adjust Aug 27 21:01:18 currently the problem is only the perl extension, so before your change it was only build native, give me a chance to fix it Aug 27 21:01:39 right now, cyrus-imapd is native only package (officially) Aug 27 21:02:00 Unrecognized argument in LIBS ignored: '-Wl,-rpath,/opt/lib' Aug 27 21:02:07 does this do any harm? Aug 27 21:04:00 that is ok Aug 27 21:04:09 we can fix it later Aug 27 21:05:14 i think the problem is that perl LD is using ld, and we're passing parameters to it as if LD=gcc Aug 27 21:05:43 i.e. ld takes "-rpath /opt/lib" Aug 27 21:05:56 gcc takes "-Wl,-rpath,/opt/lib' Aug 27 21:05:57 eno: make starts perl Makefile.PL and make, on this way my setting for PERL=$(PERL_HOST_PERL) gets lost Aug 27 21:06:33 shall I configure --without-perl and use perl Makefile.PL directly? Aug 27 21:07:55 1st priority is to fix native build Aug 27 21:09:41 when we moved perl to 5.8.8, cyrus-imapd was broken, your check-in just trigger a new build Aug 27 21:11:53 okay, I understand so I change $(OPTWARE_TARGET) back to $(HOST_MACHINE) Aug 27 21:12:26 that basically means native with-perl and cross without-perl Aug 27 21:12:27 but undertsnad that my first priority is cross because mainly I do it for my machine Aug 27 21:12:36 yes Aug 27 21:12:46 only as first fast fix Aug 27 21:14:31 i understand that without a slug, you cannot really test native build Aug 27 21:14:48 or maybe you should test ds101g native build Aug 27 21:15:06 I have tried with Rod without success Aug 27 21:15:46 no i mean test native build on your ds101g Aug 27 21:16:01 ds101g has one big problem: patched ext3 fs, it is not case sensitive breaks perl builds Aug 27 21:16:08 I meant it too Aug 27 21:16:41 ok now i understand Aug 27 21:17:22 this is why I tried to get perl cross, nobody else thought it could be important Aug 27 21:18:05 i feel the importance all along, see how many more perl packages we get once we made it cross? Aug 27 21:18:16 :) Aug 27 21:18:30 But maybe I am the only one that needs them Aug 27 21:19:00 but what I have ment is, for the most users it was okay to have a native perl Aug 27 21:19:36 03gda * r3969 10optware/trunk/make/cyrus-imapd.mk: currently only native builds with perl Aug 27 21:20:34 for end user, it makes no difference, they don't even know whether the ipk is from cross or native feed Aug 27 21:21:03 that was what I mean Aug 27 21:21:20 but there was nothing for the ds101g Aug 27 21:21:32 and it is a so nice machine Aug 27 21:24:28 yeah it got 64MB ram and 16MB flash Aug 27 21:25:23 SATA too Aug 27 21:25:51 GB Ethernet but fan too :( Aug 27 21:28:32 how much is a ds101g? Aug 27 21:29:07 got it as present for 20 years in my company, but I think around 200 EUR Aug 27 21:30:17 with HD? Aug 27 21:30:24 no Aug 27 21:33:38 it's a better machine than the slug, also a bit more expensive too Aug 27 21:34:57 do you use scripts/optware-check-package.pl ? Aug 27 21:34:59 true Aug 27 21:35:12 don't know it Aug 27 21:35:30 if you read optware/Makefile testing target Aug 27 21:36:01 it is using that script to do some sanity check, always a good idea to run that against your packages Aug 27 21:36:24 for ds101g, you might need to point it to the right objdump Aug 27 21:36:54 right now, cyrus-imapd has a few unstripped binaries Aug 27 21:37:39 only some weeks ago I fixed some problems with the ds101g and asked who to get it upstream and suddenly I got the ds101g maintainer Aug 27 21:38:08 I had not really big chance to have a good start, just jump into cold water Aug 27 21:38:21 but you're doing good Aug 27 21:38:41 don't fool me, make only problems Aug 27 21:39:21 the ds101g costs about 400euro in Germany with a 250gb hdd Aug 27 21:39:21 that's why i said a few hours ago, i don't understand the commit to "remove strip" Aug 27 21:39:37 but there a only few sellers Aug 27 21:39:44 dont know why Aug 27 21:40:06 my company has paid 380 EUR with a 320 GB disk, of course I have told them where to buy Aug 27 21:40:46 It is a wd normally I would use seagate, but the wd is cooler Aug 27 21:41:37 i am currently using an old notebook with a p3 600 and 128mb ram as NAS Aug 27 21:41:53 about 30W thats ok i think Aug 27 21:43:45 before I used a igel j thin client with a notebock drive with 15W Aug 27 21:44:09 hm k Aug 27 21:44:27 the nlsu2 with an 2,5" uses about 10W Aug 27 21:44:41 but its just too slow for php Aug 27 21:44:42 eno: now I got the first of two perl parts to build, have now to get it installed Aug 27 21:44:48 und perl Aug 27 21:44:52 -u+a Aug 27 21:46:24 eno: before I search long, what was the best condition to ask whether it is a cross build? Aug 27 21:48:13 i often use: ifeq ($(HOSTCC), $(TARGET_CC)) Aug 27 21:48:21 tnx Aug 27 21:48:34 ifneq Aug 27 21:48:40 right Aug 27 21:49:30 cyrus-imapd native build succeeds! Aug 27 21:49:42 k Aug 27 21:50:05 I have now the first part of perl ext in my ipk Aug 27 21:50:15 cool Aug 27 21:56:15 gda_: can you strip the binaries too? i don't know how to set it up to do runtime test Aug 27 21:56:35 will try when I am ready Aug 27 22:09:59 eno: will not get it ready today anymore: ../lib/isieve.c:224: error: `SIEVE_SERVICE_NAME' undeclared (first use in this function) Aug 27 22:10:23 it is too late, will get it ready tomorrow Aug 27 22:10:50 I am sure it is simple to find Aug 27 22:18:08 no problem Aug 27 22:19:34 n8 Aug 27 22:19:41 g'nite Aug 27 22:38:42 03bzhou * r3970 10optware/trunk/ (2 files in 2 dirs): strace: fixed arm-bad-syscall by applying http://www.fluff.org/ben/patches/strace/strace-fix-arm-bad-syscall.patch Aug 27 22:46:43 k inhere Aug 27 22:47:07 whats the strace output suggesting? Aug 27 22:47:30 Nymus: ipk is not made to run old ipks Aug 27 22:47:55 lol i know was only a suggestion Aug 27 22:48:33 best thing would be that old files were kept as long as new ones arent found working Aug 27 22:48:40 know that its not easy Aug 27 22:50:09 i think that's what stable/unstable is for, but we never get to set it up that way Aug 27 22:50:55 k Aug 27 22:51:33 just saying that in those cases where 'old' working files were replaced by buggy new ones its a shame Aug 27 22:52:43 i think the cross feed actually keeps old versions Aug 27 22:53:16 it's just that the semi-annual clean build removes anything older than 6months Aug 27 22:53:45 am looking at my lftpget.strace as well Aug 27 22:54:02 thx Aug 27 22:54:12 ok then rep is organised ok :) Aug 27 22:54:18 just not lucky (again) Aug 27 22:54:29 but appreciated the effort u and others put into it Aug 27 22:54:50 others as in drif :) Aug 27 23:01:05 found out something eno? Aug 27 23:01:07 I'm glad my slug is off the internet :P revealed my temporary passoword there... Aug 27 23:01:22 -o Aug 27 23:02:53 Nymus: still looking, comparing x86 lftp.strace with slug lftp.strace Aug 27 23:04:43 ok will try to limit my enthousiasm Aug 27 23:37:02 Nymus: could not find anything obvious in lftp.strace Aug 27 23:38:57 damn Aug 27 23:43:33 the next step would be to build with debug, and run it under gdb Aug 27 23:45:52 hmz trying something Aug 27 23:45:58 pretty weird but seems to work Aug 27 23:46:01 testing again Aug 27 23:54:48 got kicked off Aug 27 23:54:55 looked like server was down Aug 28 00:03:35 eno got a very stupid but working solution/bugg bypass for lftp Aug 28 00:04:18 ? Aug 28 00:04:38 or arent u interested :p Aug 28 00:04:56 i just ran two tests Aug 28 00:05:08 they were both executed like they should Aug 28 00:05:17 when i "mirror directory" Aug 28 00:05:24 i get segmentation fault Aug 28 00:05:32 when i "queue mirror directory" Aug 28 00:05:44 it runs it as a job in the background and executes like it should Aug 28 00:05:50 totaly f*cked up Aug 28 00:05:52 but it works :) Aug 28 00:06:30 but it's definitely very buggy, i can reproduce the segfault with a different scenario very easily Aug 28 00:07:14 no doubt about i being buggy Aug 28 00:07:23 but at least i got the mirror function workin properly Aug 28 00:07:34 so that it can download full dirtrees Aug 28 00:07:44 why not use rsync? Aug 28 00:08:05 cuz ive been trying to get lftp workin for ages Aug 28 00:08:12 and now that it does ... Aug 28 00:08:38 thought u wanted 'workin' lftp as well Aug 28 00:11:08 anyway seems to work great when u put queue in front of the mirror and mget command Aug 28 00:12:03 going of now cya guys tomorrow Aug 28 00:16:41 got a gdb backtrace of lftp segfault Aug 28 00:17:34 it's very similar to http://lists.debian.org/debian-68k/2004/09/msg00063.html Aug 28 01:59:45 03bzhou * r3971 10optware/trunk/make/py-4suite.mk: py-4suite: upstream upgrade to 1.0rc3 **** ENDING LOGGING AT Mon Aug 28 02:59:56 2006