**** BEGIN LOGGING AT Thu Aug 04 23:59:56 2005 Aug 05 01:59:01 koen around? Aug 05 01:59:10 yes Aug 05 01:59:14 cool. Aug 05 01:59:29 Was it you that hacked up viewmtn to work with monotone 0.20+ ? Aug 05 02:00:51 Can my color be nicer? :P Aug 05 02:01:12 dyoung: replace depth with last or something very similar Aug 05 02:01:33 ok, I did that. But it displays ALL of it. Aug 05 02:02:01 so was wondering if something else needed hacking on... Aug 05 02:02:09 or if I need to RTFM. Aug 05 02:03:49 you mean the whole 'manifest'? Aug 05 02:03:57 mickeyl has a patch for that Aug 05 02:03:59 Yeah, Aug 05 02:04:27 http://ewi546.ewi.utwente.nl/tmp/viewmtn/ still has the manifest Aug 05 02:05:35 Yeah, mine looks sorta like that, except worse because the graphic shows the whole thing too. Aug 05 02:05:45 looks like mickeyl is zzzz... Aug 05 02:07:16 its 10am! Aug 05 02:07:31 the depth/last thing takes a paramater to shorten the tree Aug 05 02:13:41 I tried hard coding it in too. limit=5 or somehting.... Aug 05 02:13:47 my pythonfoo sucks Aug 05 06:52:52 rwhitby: should I go ahead and add ntp-tickadj to the firmware now? Aug 05 06:53:07 come to think of it, I'll need to create somesort of init-script for that Aug 05 06:53:24 (and then pull up my serial-enabled slug to test) Aug 05 06:55:32 NAiL: yep Aug 05 07:04:39 rwhitby: ok.. the tickadj initscript... I can't think of anything that can go wrong. It run /usr/bin/tickadj 10101. Do I need to test for stuff? Aug 05 07:05:06 well.. ok, test if it's there... someone might decide to remove the ipkg :P Aug 05 07:05:34 If the init script is part of the ipkg, then that's not a problem Aug 05 07:05:47 make it an nslu2-specific init script Aug 05 07:06:30 I was thinking about including it in base-files/openslug, instead of adding a hack to ntp... but it's probably better to include in ntp-tickadj Aug 05 07:07:04 better to be an nslu2-specific init file for ntp-tickadj Aug 05 07:07:13 do_install_append_ntp-tickad? Aug 05 07:07:25 no, an update-rc.d Aug 05 07:07:31 ah Aug 05 07:07:46 like any other package init script Aug 05 07:09:03 INITSCRIPT_NAME_ntp-tickadj_openslug? Aug 05 07:09:18 or nslu2 Aug 05 07:10:02 nslu2, cause you would use it on unslung too if we decided to add the package to it Aug 05 07:14:49 do_install_append_ntp-tickadj_nslu2() <-- this looks correct right? It's getting confusing when it's that long ;-) Aug 05 07:18:57 no ntp-tickadj in do_install Aug 05 07:19:14 just do_install_append_nslu2() Aug 05 07:20:35 uh Aug 05 07:20:51 that'll hit both ntp, ntp-bin, ntp-tickadj and ntpdate, no? Aug 05 07:21:00 "both"->"all of" Aug 05 07:21:01 its the same do_install for all Aug 05 07:21:18 do_install is for the entire bb, not for each package Aug 05 07:21:52 now I'm not following Aug 05 07:21:54 oh Aug 05 07:21:57 yes I am Aug 05 07:21:59 what later goes in what package is set by the FILES_${PN}-tickadj Aug 05 07:22:04 it's the package stuff that splits it up Aug 05 07:22:09 yep :) Aug 05 07:23:13 but the INITSCRIPT_NAME variable should be set to ntp-tickadj, because that is done during package-stuff Aug 05 07:23:17 right? Aug 05 07:23:23 I should eat and come back Aug 05 07:23:33 My brain is running on vapors Aug 05 07:24:07 hehe, yes initscript_name should be ${PN}-tickadj Aug 05 07:24:12 or, set to Aug 05 07:25:08 INITSCRIPT_NAME_${PN}-tickadj_nslu2? Aug 05 07:25:15 yes Aug 05 07:25:28 ok, so that's already correct Aug 05 07:29:37 uhmm.. could someone cat /etc/issue ? I managed to write over mine Aug 05 07:30:06 OpenSlug 2.3-beta \n \l Aug 05 07:30:15 kolla: It's not in your internal flash? Aug 05 07:30:16 thanks Aug 05 07:33:40 oh, right - I forgot :) Aug 05 07:34:19 2.4-beta here :) Aug 05 07:37:58 linux_logo was tricky, it creates some binaries that it runs to determin what platform it is compiled on, ofcourse if those binaries are cross-compiled for arm they wont run, and when I fixed problem, they ofcourse didnt find an arm platform and hence gave the linux_logo build wrong information about the layout of /proc/cpuinfo :) Aug 05 07:38:42 linux_logo and linux_logo-native? ;) Aug 05 07:38:46 why all the linux archs have completely different /proc/cpuinfo.. I never got it :) Aug 05 07:39:26 kolla: because linux is a hackish hobby project Aug 05 07:39:55 NAiL: my trick was to tell the .bb to first build native for pc, then I tossed in some fixed makefiles and deleted the neeeded objects and programs, and let it rebuild using oe_runmake :) Aug 05 07:40:29 koen: yes, that's ofcourse it :) Aug 05 07:42:23 NAiL: I think a cleaner solutions is to simply fix the makefile and emulate those helper programs with scripts instead, scripts that dont look into the hardware compiling it Aug 05 07:42:59 or using -native ;) Aug 05 07:43:04 more like "oh, we're crossbuiolding for arm, let's assume it's meant to use the /proc/cpuinfo of an arm" Aug 05 07:43:22 sure, everything is simpler native Aug 05 07:43:53 but my slugs waste CPU on gentoo/armeb, I'll stick with cross-building openslug :) Aug 05 07:44:45 kolla: so when will we be ready to add http://www.nslu2-linux.org/wiki/GentooSlug/Homepage ? Aug 05 07:45:03 The default runlevel on the slug is what? 3? Aug 05 07:45:11 ugh Aug 05 07:45:12 5? Aug 05 07:45:19 is 5, should be 3 Aug 05 07:45:40 the 5 comes from OE, which starts a graphical environment on it's initial distros Aug 05 07:46:11 aha Aug 05 07:46:25 rwhitby: good question :) Aug 05 07:47:26 kolla: well a Gentoo developer can't let DebianSlug get a headstart, can they ? Aug 05 07:47:27 rwhitby: solar wants to create a gentoo flash-image I know Aug 05 07:48:03 and.. I'm doing gentoo/glibc while the others are doing gentoo/uclibc Aug 05 07:50:49 rwhitby: the debianslug doesnt count, it's wrong-endian :) Aug 05 07:51:16 btw, is there a fedoraslug/homepage? Aug 05 07:51:38 cause fedora core3 is very much done, and fedora core 4 seems well on its way Aug 05 07:52:07 there could be, but Lennert tried to push some Fedora/armeb stuff upstream and hit a big brick wall Aug 05 07:52:45 aha? Aug 05 07:53:04 "upstream" being what.. Fedora? Aug 05 07:53:15 yep Aug 05 07:53:32 I see Aug 05 07:53:53 anyways, it should be linked to from somewhere, it offers a nice working build environment Aug 05 07:54:12 I never got portage working properly on the native openslug Aug 05 07:54:35 From Lennert: "All discussions on the mailing list at the time seemed to end up in Aug 05 07:54:35 "Fedora does not support ARM.", with Red Hat employees providing not Aug 05 07:54:35 even the slightest of clues as to what the criteria for inclusion Aug 05 07:54:35 of a new architecture are." Aug 05 07:54:36 too much busyboxism I think Aug 05 07:54:56 figures.. Aug 05 07:55:27 he can release Fedupah :) Aug 05 07:56:30 Why is there a openembedded/packages/linux/nslu2-kernel/2.6.12/ and a openembedded/packages/linux/openslug-kernel-2.6.12 ? I'm just trying to find out, which defconfig is actually used... Aug 05 07:57:09 malfi: nslu2-kernel is the standard linux kernel ported to the nslu2. openslug-kernel is a very close derivative of that. Aug 05 07:57:31 it's designed so that someone else can create a different distro based on nslu2-kernel if they like Aug 05 07:58:09 rwhitby: Ah, thanks. Well, is there a patch accept policy for this openslug kernel? Aug 05 07:58:30 rwhitby: in terms of feature patches Aug 05 07:59:10 malfi: if you can convince people here that it's a good idea, then it gets in :-) Aug 05 08:00:32 rwhitby: I'm playing around with the misdn patches, but they are rather large but provide USB ISDN support Aug 05 08:01:04 do the build a module, or increase the size of the kernel unconditionally ? Aug 05 08:01:24 just some more modules, nothing to bzimage Aug 05 08:01:33 should be no problem then Aug 05 08:02:14 hmm, no changes to the kernel at all? Aug 05 08:02:21 Is there any reason why we wouldn't just drop it in, and have the modules in the feed ? Aug 05 08:02:26 rwhitby: okay, but first I have to get it working for myself :) Aug 05 08:02:53 malfi: yeah, around here that's usually the hardest hurdle to overcome .... Aug 05 08:02:56 If there are no kernel changes, I can't see any reason to not include this Aug 05 08:03:51 http://pastebin.ca/19389 Aug 05 08:04:01 that are the changes in the defconfig Aug 05 08:04:34 hmm Aug 05 08:04:35 yep, that looks fine to me Aug 05 08:04:42 yeah Aug 05 08:04:47 night all Aug 05 08:04:50 nite Aug 05 09:47:43 package-index is so slow.. Aug 05 09:48:07 there's a bunch of packages for it to index ;) Aug 05 09:48:10 does it have anything to do with number of packages? if so, there might be a problem when the number of packages rise :) Aug 05 09:48:14 hehe, yes.. Aug 05 09:48:24 its faster on the slug :-) Aug 05 09:48:31 it reads each and every package IIRC Aug 05 09:48:32 not like there are so many packages to index Aug 05 09:48:32 because it has to index 3 packages Aug 05 09:48:44 ? Aug 05 09:49:01 ah Aug 05 09:49:07 there are 3 native packages? Aug 05 09:49:10 on the slug :) Aug 05 09:49:11 yep Aug 05 09:49:15 ok Aug 05 09:49:21 or, actually 0, but I added a few to test with Aug 05 09:49:39 ok Aug 05 09:50:20 DaKa2: Did you get apache to build? ;) Aug 05 09:50:23 * kolla is trying to make some ansi/ascii logo for openslug Aug 05 09:50:31 NAiL: didn't even try, didn't you? :-) Aug 05 09:50:52 I know I once had a program to do 1337-logos, but I cant remember what it was, I suspect it was on the amiga though Aug 05 09:51:06 WMAnsiEd isnt too bad though Aug 05 09:51:41 I remember doing that stuff in DOS many, many years ago Aug 05 09:52:00 The packages creating the most ipks: Aug 05 09:52:03 48 python Aug 05 09:52:03 73 libglib Aug 05 09:52:03 167 kernel Aug 05 09:52:03 338 locale Aug 05 09:52:03 663 glibc Aug 05 09:52:06 859 perl Aug 05 09:52:08 hehe Aug 05 09:52:22 try gst-plugins Aug 05 09:52:31 that's in the 500 range too Aug 05 09:52:35 hm.. I dont have that in my deploy :) Aug 05 09:52:42 ls | cut -f1 -d"-" | cut -f1 -d"_" | uniq -c | sort -n Aug 05 09:52:50 it's not a sluggie package Aug 05 09:53:04 not that correct, but it gest the one with packagename-subname Aug 05 09:53:37 240 gst 257 gpe 260 kernel 338 locale 533 gnome 663 glibc 859 perl Aug 05 09:53:45 :-) Aug 05 09:55:57 Im really starting to wonder why the perl commit is not showing up in oe... Aug 05 09:56:30 http://monotone.vanille.de/viewmtn/revision.psp?id=e4d3d3ab851d4b17d8ffd178d57e2c416ace081a Aug 05 09:56:37 Probably just CIA getting stuck Aug 05 09:56:48 ah Aug 05 10:00:40 hehe, everything is green in tinderbox :) never seen that before Aug 05 10:05:09 except that openslug just droped off... Aug 05 10:10:22 and there it returned, with a complete perl build done :) Aug 05 10:15:22 the feed is correct? Aug 05 10:16:04 yep :-) Aug 05 10:16:08 fantastic :) Aug 05 10:20:41 Modular ISDN Stack core $Revision: 1.25 , usbcore: registered new driver mISDN_hfcsusb Aug 05 10:20:44 * malfi cheers. Aug 05 10:21:10 does that work with the kernel that is already in the firmware image? Aug 05 10:22:27 oh, well.. doesn't really matter Aug 05 10:34:10 what the... "cp: cannot stat `/build/find_apu.m4': No such file or directory" Aug 05 10:34:20 I can see why the apache bb is in nonworking.. Aug 05 10:34:28 haha Aug 05 10:34:34 Come tell me it's that simple Aug 05 10:35:02 DaKa2: Apu is right here on TV Aug 05 10:35:08 * koen is watching the Simpsons Aug 05 10:35:18 haha :) Aug 05 10:35:55 Thank you, come again! Aug 05 10:35:59 I should probably get going, so I can watch TV too... Aug 05 10:37:12 but its just so nice beeing in a place where you can play music as loud as you want... :) Aug 05 10:37:24 should get a TV and a bed and I can move in at work Aug 05 10:39:38 heh Aug 05 10:39:54 DaKa2: Streaming TV? Aug 05 10:40:12 NAiL: hm.. yeah.. might be an idea :-) Aug 05 10:40:51 get a usb tv-tuner with mpeg encoder, plug it in the slug at home, get it to stream the stuff to work :) Aug 05 10:41:17 I only have one mbit upload at home.. but.. with a bit compresstion that should work Aug 05 10:41:31 'only' Aug 05 10:41:37 * koen has 1Mbit download Aug 05 10:41:37 yes? Aug 05 10:41:39 streaming at ~7-800kbps isn't that bad Aug 05 10:41:47 I've got >1mbit upload Aug 05 10:41:49 I have 24Mbit down :) Aug 05 10:41:57 * NAiL glares at DaKa2 Aug 05 10:42:11 and at work just a tad bit more.. Aug 05 10:42:15 * koen had 100mbit full duplex on campus Aug 05 10:42:21 :) Aug 05 10:42:33 I'm moving to 100mbit this autumn :D Aug 05 10:42:51 Im considering buying 100/100 or atleast 10/10 at home Aug 05 10:43:09 100/100 is like 500SEK/month and 10/10 is 250 I think.. Aug 05 10:43:24 I pay more than that for my 4mbit line Aug 05 10:43:31 :) Aug 05 10:45:18 ah.. apache tries to cp $apu_src_dir/build/find_apu.m4 somewhere.. and I guess $apu_src_dir is not defined Aug 05 10:46:45 * DaKa2 changes to srclib/apr-util Aug 05 11:38:35 Is there any reason (other than bootup scripts) I should care about when switching the runlevel? I've edited so that my slug is running with the default runlevel 3 now. No difference that I can see :) Aug 05 11:42:18 It's just a technicality, but people like me assume (wrongly) that the default runlevel would be 3 :-P Aug 05 13:15:43 NAiL: NOTE: package apache-2.0.47-r0: task do_compile: started Aug 05 13:16:03 native? Aug 05 13:16:06 yep Aug 05 13:16:15 did you get it to work cross? ;) Aug 05 13:16:26 hehe, I'll take that later Aug 05 13:16:38 * NAiL crosses fingers Aug 05 13:39:05 * malfi curses. Aug 05 13:39:19 hmm Aug 05 13:39:27 The isdn adapter I bought should have had the supported hfc chipset, but it doesn't. Aug 05 13:39:27 * NAiL disspels the curse Aug 05 13:39:32 oh Aug 05 13:39:34 damn Aug 05 13:40:08 stupid winbond, I didn't even know that draytek uses this chipset manufacturer Aug 05 13:44:04 but that streaming idea sounds good. Streaming this DVB-T mpeg2 stream should be easily done by a slug. Aug 05 13:47:12 NOTE: package apache-2.0.47-r0: task do_package: started Aug 05 13:47:12 ERROR: function do_install failed Aug 05 13:47:16 close :-) Aug 05 13:47:40 here in Berlin a bouqet of four channels gets 14.5mbit/s, that's 3-3.5mbit/s per channel. So 4mbit/s upstream would do the trick for recodingless streaming. Aug 05 13:47:49 DaKa2: Neat Aug 05 13:49:22 could someone tell me whats wrong with: Aug 05 13:49:22 + sed -e 's,/usr/sbin/,/usr/sbin/,g; \ Aug 05 13:49:22 s,/usr/bin/,/usr/bin/,g; \ Aug 05 13:49:22 s,/usr/lib,/usr/lib/,g; \ Aug 05 13:49:22 s,/etc/apache/,/etc/apache/,g; \ Aug 05 13:49:25 s,/etc/,/etc/,g; \ Aug 05 13:49:27 s,/usr/,/usr/,g;' Aug 05 13:49:30 sed: -e expression #1, char 165: unterminated address regex Aug 05 13:50:26 I always use several -e '' Aug 05 13:50:40 like sed -e "s#a#b#g" -e "s#c#d#g" Aug 05 13:50:45 yep.. but apache doesn't :-) Aug 05 13:51:09 is there a space after the trailing backslash somewhere? Aug 05 13:52:05 hm. no.. Aug 05 13:52:43 you aren't using busybox sed, are you? Aug 05 13:52:47 actually, this is in the bb, so I can change it.. Aug 05 13:52:48 nope Aug 05 13:52:54 real sed Aug 05 13:55:21 why don't you delete that sed stuff and just insert a /bin/true, it does the same... :) Aug 05 13:56:09 hehe, well :-) this is just incase someone uses different dirs.. not that I know of any dists in OE doing that.. Aug 05 13:57:40 sed -e 's,/usr/sbin/,/usr/sbin/,g;s,/usr/bin/,/usr/bin/,g;s,/usr/lib,/usr/lib/,g; s,/etc/apache/,/etc/apache/,g;s,/etc/,/etc/,g;s,/usr/,/usr/,g;' /dev/null Aug 05 13:57:44 works for me Aug 05 13:58:07 so it's either your sed version or some problems with the linebreaks. Aug 05 13:58:19 hm.. I just don't understand why they did it this way.. Aug 05 13:58:57 malfi: thanks.. I'll probably just test without the linebreaks.. and then scrap that part completly. Aug 05 13:59:31 woho, I did several -e '' of it, and that worked and: Aug 05 13:59:33 NOTE: package apache-2.0.47: completed Aug 05 13:59:36 NAiL! Aug 05 13:59:40 yes, I see Aug 05 13:59:42 wohooo :) Aug 05 13:59:49 ipkg install && httpd :) Aug 05 13:59:56 hm.. do I dare? Aug 05 14:00:03 yes Aug 05 14:00:04 * NAiL grins Aug 05 14:00:11 * NAiL watches DaKa2's slug catch fire Aug 05 14:00:15 watches his slug burn Aug 05 14:00:17 :-) Aug 05 14:00:30 actually, my slug is at home.. Aug 05 14:01:04 phew.. I got hundreds of errors.. because I tried ipkg install as normal user :-) Aug 05 14:01:22 haha Aug 05 14:04:52 # httpd Aug 05 14:04:53 Aborted Aug 05 14:04:56 grr Aug 05 14:05:12 strace to the resque! Aug 05 14:05:19 rescue Aug 05 14:05:23 way ahead of you :-) Aug 05 14:05:41 you never know ;) Aug 05 14:06:58 uh.. I don't understand whats happening... Aug 05 14:07:18 output in logs? tracing children? Aug 05 14:07:29 it dies almost instantly Aug 05 14:08:38 -X ? Aug 05 14:08:47 -X Run httpd in debug mode. Only one worker will be started and the Aug 05 14:08:48 server will not detach from the console. Aug 05 14:09:01 a strace or gdb might help too Aug 05 14:09:14 -X makes no difference Aug 05 14:09:25 strace apache2 -X Aug 05 14:09:26 the end: http://pastebin.com/330274 Aug 05 14:09:46 SIGABRT oh Aug 05 14:09:56 yes... Aug 05 14:11:28 what does gdb say? Aug 05 14:11:53 lets see.. Aug 05 14:11:56 ipkg install gdb Aug 05 14:12:44 hm.. (no debugging symbols found) :-) Aug 05 14:13:21 http://www.fiveanddime.net/apache-web-server-files-and-documentation-2.0.54/CHANGES..html say *) CoreDumpDirectory directive directs where the core file is Aug 05 14:13:21 written when a SIGSEGV, SIGBUS, SIGABORT or SIGABRT are Aug 05 14:13:21 received. Aug 05 14:13:58 neat Aug 05 14:13:59 it doesn't even read the config-file.. Aug 05 14:14:00 sudo works Aug 05 14:14:06 (almost) Aug 05 14:14:45 DaKa2: why don't you try 2.0.54, maybe they fixed that problem already... Aug 05 14:15:01 malfi: yeah... might as well update the bb to that... Aug 05 14:15:57 the patches break then though Aug 05 14:16:22 DaKa2: pray for the hungs :) Aug 05 14:17:03 DaKa2: can you post your bitbake file and other modifications somewhere? I'd try to help... Aug 05 14:17:24 right now Im basically at the one in nonworking/apache Aug 05 14:17:57 but Im working on stripping it down to the bare minimum since this is not meant to be cross built right now Aug 05 14:37:34 what's wrong if I get ERROR: Nothing provides openembedded/packages/apache/apache_2.0.54.bb, when trying ". setup-env; bitbake openembedded/packages/apache/apache_2.0.54.bb" ? Aug 05 14:38:02 ah, I forgot the -b Aug 05 14:38:04 nevermind Aug 05 14:38:19 :) Aug 05 14:39:13 the only problem is that testing takes so much time.. just do_configure takes 20 minutes on the slug.. Aug 05 14:39:25 shouldn't it use ccache? Aug 05 14:39:52 well, why don't set up distcc with your crosscompiler, it should work Aug 05 14:40:06 compiling nativly supported by your power horse Aug 05 14:40:15 hm.. Aug 05 14:40:43 or, just hack on something else while waiting :) Aug 05 14:40:52 hehe Aug 05 14:44:29 yasy Aug 05 14:44:30 argh Aug 05 14:44:33 yay :) Aug 05 14:44:34 8204 -rw-r--r-- 1 repvik repvik 8388608 Aug 5 23:37 openslug-nslu2-20050805195306.flashdisk.img Aug 05 14:44:43 That's a 2.5 firmware image (more or less) Aug 05 14:44:59 but NAiL was right, patches are broken with 2.0.54 Aug 05 14:45:08 of course I am Aug 05 14:45:11 I tried :-P Aug 05 14:45:25 2.5? is 2.4 frozen? Aug 05 14:45:36 haha, Im debugging why my extra 32mb sdram doesn't work :-) seems like I didn't solder some pins :-) I get 00 00 f5 c0 all over.. Aug 05 14:46:04 malfi: what used to be 2.4 has been tagged. If you do an update, you'll see you're suddenly working on 2.6 ;-) Aug 05 14:46:26 NAiL: when did this happen? Aug 05 14:46:30 It still in the process of being released Aug 05 14:46:32 hmm Aug 05 14:46:43 21:37 < jbowler-zzz> Ok, OpenSlug-2.5-beta == 16b882cb36841a02c3d2f489b8c299a90062c821 Aug 05 14:46:51 two hours and 9 minutes ago Aug 05 14:47:02 oh, thanks Aug 05 14:47:49 NAiL: has anybody tried compiling nativly using distcc with the crosscompiler? Aug 05 14:48:06 I think someone was going to set it up Aug 05 14:48:15 I don't know if anyone actually succeeded Aug 05 14:48:55 It should make all these build vs host system tool problem go away. Aug 05 14:49:37 maybe. That's what gurus do behind the scenes when it comes to me :P Aug 05 14:50:12 I'm just watching in awe as the environments (native and cross) are developing, and try to make as much use of them as possible :) Aug 05 14:50:28 which reminds me.. I have a feed to update Aug 05 14:50:42 hehe Aug 05 14:51:15 svn commit ./ -m "Update" && ssh nudi make -> feed updated :) Aug 05 14:52:44 apropos svn, they should have take that instead of monotone which makes me feel, that I'm working on slug and not on some amd64... Aug 05 14:53:13 heh Aug 05 14:53:23 svn is not distributed Aug 05 14:53:40 I was just about to say I couldn't remember the reasoning Aug 05 14:53:46 but there part of it was ;) Aug 05 14:53:46 * koen|tv likes to work his CPUs Aug 05 14:54:08 oh, wait Aug 05 14:54:15 it aren't my CPUs :) Aug 05 14:54:19 lol Aug 05 14:54:20 s/my/mine/ Aug 05 14:54:22 hehe ;) Aug 05 14:54:39 it's the 'spare' dual opteron at the uni Aug 05 14:54:43 hahah Aug 05 14:54:46 bastard ;) Aug 05 14:54:50 but is montone that ahead feature-wise? Aug 05 14:55:35 svn will never be distributed Aug 05 14:55:46 There's two things bugging me with monotone. The speed, and the 'too-old-ancestor' stuff when merging branches Aug 05 14:55:55 But I can live with both Aug 05 14:56:32 I hope the merging stuff get's fixed Aug 05 14:56:34 what do you mean with distributed? distributed repositories, that are spread all over the world? Aug 05 14:57:07 If both nslu2-linux and oe's servers go down, even I can serve the repo Aug 05 14:57:37 and offline commits Aug 05 14:57:43 NAiL: ah, that's nice Aug 05 14:57:46 yes! :) Aug 05 14:57:50 offline commits are nice Aug 05 14:58:02 * NAiL has done some 12 commits offline, then done a sync Aug 05 14:58:14 looks rather nice when CIA chokes (That was before throttling) Aug 05 14:59:12 oh, yes. That was the third thing with monotone.. CIA doesn't *instantly* show up with the revision ;-) Aug 05 14:59:30 with svn, the channel actually knows it *before* the svn process has exited on my box Aug 05 14:59:31 that can be done Aug 05 14:59:39 Yeah, I know it can be done now Aug 05 14:59:45 but noone has implemented it :P Aug 05 14:59:49 :) Aug 05 14:59:59 And don't look at me, because I wouldn't know how to ;) Aug 05 15:00:00 find a lua wizard and make it happen Aug 05 15:00:02 oh, do post-commit hooks actually work now? Aug 05 15:00:13 monotone 0.21 introduced those iirc Aug 05 15:00:33 neato Aug 05 15:00:54 lua? shudder Aug 05 15:00:55 NAiL: I found cogito (the linus scm) very funny Aug 05 15:01:19 cogito? Is that what became of git? Aug 05 15:01:19 "push to repo"..... *curse*.... *read manual*.... *curse again* Aug 05 15:01:25 *rofl* Aug 05 15:01:34 it doesn't have a push function Aug 05 15:01:39 whaaa? Aug 05 15:01:48 you can only pull and merge Aug 05 15:01:54 haha Aug 05 15:02:07 and put your repo online for someone to pull Aug 05 15:02:23 omg Aug 05 15:02:24 the putting online part was also severely undocumented Aug 05 15:02:35 hahah, cool Aug 05 15:03:13 so monotone looked very impressive after that experience Aug 05 15:03:53 yeah Aug 05 15:03:57 monotone isn't too bad Aug 05 15:04:29 I like monotone. But then again, I do not have much experience with SCM Aug 05 15:06:00 I abused svn as a backup system at the company, I job at... works really nice... :) Aug 05 15:06:32 hehe Aug 05 15:06:55 svn can be distributed if you use svk as a wrapper Aug 05 15:07:23 but svk is an ugly mess of perl scripts Aug 05 15:08:00 monotone needs a user friendly front end - as it is it is a very good database abstraction, but I want to see a window with 'monotone annotate' merged with the log messages... Aug 05 15:08:14 I just do a 'svn commit' (and some magic foo) every hour. I'm not aware of any backup solution that provides that much features. Aug 05 15:09:22 But if you want to hack on something for a while without killing everyone elses build you have to svn cp the whole database, then svn cp it back afterward (apparently). Aug 05 15:10:07 And you have to make a complete svn cp for every branch, and for every sub-branch - the database ends up with thousands of copies of everything. Aug 05 15:10:50 It doesn't even have the bk 'link' option so far as I can see (monotone, of course, doesn't need that). Aug 05 15:11:38 And monotone is blindingly fast compared to svn or bitkeeper. Aug 05 15:11:46 pardon me? Aug 05 15:11:55 It is so much faster. Aug 05 15:12:11 It's practically instant for me to check what has changed - try that with svn! Aug 05 15:12:20 yeah Aug 05 15:12:22 svn has to go back to the server... oops. Aug 05 15:12:33 bk was ok, but still slower than monotone. Aug 05 15:12:51 okay, I admit that I used svn with file:/// repositories... :) Aug 05 15:13:01 that was quite okay speedwise Aug 05 15:13:15 And monotone update is incredibly much faster than bk - where you had to 'pull' to update (the two were not separated). Aug 05 15:14:16 'running consistency check..' Aug 05 15:14:20 Well, that's not completely true - I had local bk clones (lots of them), but I didn't automate the parent clone pull because it just seemed to dangerous. No such worry with monotone. Aug 05 15:15:10 Yeah, consistency check == verifying revisions - but with monotone that's already been done because it happens automatically (i.e. I have a script run every 10 minutes). Aug 05 15:15:34 every twenty minutes over here Aug 05 15:15:42 push/pull to all 4 repos Aug 05 15:15:49 Four? Aug 05 15:15:57 I only push/pull to three... Aug 05 15:16:23 Actually, two - because I chose ewi or vanille. Aug 05 15:16:39 dominion.kabel.utwente.nl serves too Aug 05 15:16:45 but that's my cable connection :) Aug 05 15:16:48 ta-da. feed updated :) Aug 05 15:17:21 * NAiL just got a mail stating that the build succeeded, and which packages were pushed. Gotta love this ;-) Aug 05 15:17:39 All in all I really like monotone - like NAiL said there are some big problems (the merge issue) but the big advantages tip the scale. Aug 05 15:18:38 svn trunk has OpenSlug-2.5-beta openembedded (but not the .dev structure yet) Aug 05 15:19:55 Hmm tinderbox says openslug is currently broken Aug 05 15:20:52 indeed it does Aug 05 15:22:23 woohoo - builderrors.log has " openslug" in it :-) Aug 05 15:22:32 neat :) Aug 05 15:22:46 when is that info gonna show up in #nslu2-linux? ;) Aug 05 15:22:47 Must be something I propagated from .dev Aug 05 15:22:51 NOTE: package perl-5.8.7-r14: task do_compile: started Aug 05 15:22:51 ERROR: function do_compile failed Aug 05 15:23:15 huh? Aug 05 15:23:21 why is it building perl *again*= Aug 05 15:23:24 NAiL: as soon as someone write a command line cia notifier script for me to run from the master makefile. Put it in ./scripts Aug 05 15:23:38 rwhitby-asleep: haha, yes. of course :P Aug 05 15:23:52 -r14 was built and pushed to the feed already Aug 05 15:25:25 Hum, my from scratch builds haven't got to perl yet... Aug 05 15:26:40 not mine either Aug 05 15:27:28 jbowler-away: how far has your build gotten? Aug 05 15:27:57 ucslugc is at monotone-4, openslug is at mysql Aug 05 15:28:16 (This isn't the release build - this is the monotone head) Aug 05 15:29:38 ah Aug 05 15:30:02 I just installed distcc from OE, and it works. Can anybody include that line in conf/distro/openslug-packages + packages/meta/openslug-packages.bb, or should I file a feature request at slugbug? Aug 05 15:30:04 perl built for me earlier... I did a clean build right after daka's perl commits Aug 05 15:30:36 OE supports icecream for distributed builds too Aug 05 15:30:39 I can't do that right now, since my tree is wildly rearranged ;) Aug 05 15:30:49 that has te added advantage it uploads the toolchain too Aug 05 15:31:18 Oh, I never heard of that before, me -> google Aug 05 15:33:48 * NAiL crosses fingers Aug 05 15:33:57 I'm currently waiting for microcom to download Aug 05 15:35:33 make[1]: Leaving directory `/home/repvik/nslu2/openslug' Aug 05 15:35:33 make: *** [build-openslug] Error 2 Aug 05 15:35:56 It couldn't fetch the sources from port5, and gave up directly Aug 05 15:38:30 'night all Aug 05 15:39:27 port5 is down for quite some time Aug 05 15:40:25 the dns entry no longer exists Aug 05 15:42:14 hmm - so our nslu2-mirrors.bbclass isn't working. Aug 05 15:42:23 jbowler-away: can you check the logic on that one? Aug 05 15:43:36 * NAiL looks at how to interact with cia using python. Utterly confused ;) Aug 05 15:44:09 there is a openembedded/classes/icecc.bbclass, but how do I use it? Aug 05 15:47:52 rwhitby-asleep: the extensive documentation of MIRRORS doesn't help in this case, but I believe it does a 'first match', so it will always use www.oesources.org Aug 05 15:48:25 in our conf/distro/foo.conf dont we need to INHERIT += "nslu2-mirrors" someplace ? Aug 05 15:48:40 Look in machine/nslu2.conf Aug 05 15:50:30 Are there plans to mirror the sources as well? Aug 05 15:50:41 NAiL, yes Aug 05 15:50:45 neat Aug 05 15:52:08 well, the load on sources is much smaller, as not many people build from source Aug 05 15:53:26 no, and the mirrors should only be used when the original site fails, right? Aug 05 15:53:34 fetch.py does iterate through all the mirrors, until 'fetch_uri' succeeds. Aug 05 15:54:16 why didn't it do this on my build right now? Aug 05 15:54:29 So is http://sources.nslu2-linux.org/sources/ actually being hit? Aug 05 15:54:52 I've hit it manually a couple of times Aug 05 15:54:56 for m102.tar.gz Aug 05 15:54:59 Then it must be working. Aug 05 15:55:13 I.e. the nslu2-mirrors.bbclass must be working. Aug 05 15:55:28 uh.. Aug 05 15:55:53 I thought you meant if anyone was actually using that mirror. And yes, I have been using it. with wget, not bb. Aug 05 15:56:07 I've manually fetched the m102 sources Aug 05 15:58:41 lets just change the .bb to get it from sources.nslu2-linux.org Aug 05 15:58:41 Hum... something is reseting the time of day by 1 hour somewhere. Aug 05 15:59:23 rwhitby-asleep: I thought about that. Either that or don't build it by default. It'll be in the feed anyways. Aug 05 15:59:49 change the .bb - it's not a temporary failure Aug 05 15:59:58 The problem is that it will happen to something else... Aug 05 16:00:15 we need to fix the mirrors stuff too Aug 05 16:00:30 but we should change anything that is really dead too Aug 05 16:02:20 I don't really think we need to have more than one mirror for the sources .... dyoung, what do you think (since sources is your machine) Aug 05 16:02:41 I think its probably unneccessary. Aug 05 16:03:01 the sources are a backup anyways. Aug 05 16:03:37 yeah, and the number of developers is small enough that it probably could have been left on nugabe if we didn't have your sweet connection to use :-) Aug 05 16:03:58 how big is your pipe again? Aug 05 16:04:11 I dunno, its ridiculous. Aug 05 16:04:46 Can I enlist your help to dig a ~50 meter tunnel? :D Aug 05 16:04:46 "ridiculous" <- that's a nice number when you're talking about bandwidth ... Aug 05 16:05:08 I know I can do 2 100Mbit transfers at the same time; so at least 200Mbit on the Level3 cnnection; but it was specified as a 622Mbit link. Aug 05 16:05:37 multi-homed too, dunno what the other 2 are. Aug 05 16:06:09 kolla: ping? Aug 05 16:06:24 NAiL: "If you can see this, it means that the installation of the Apache web server software on this system was successful. You may now add content to this directory and replace this page." Aug 05 16:06:40 * NAiL shakes DaKa2's hand :) Aug 05 16:07:29 still some things need fixing Aug 05 16:08:11 and.. this package is now made just for native building Aug 05 16:11:10 This doesn't make sense because the fetch isn't trying any of the mirrors. Aug 05 16:12:49 Eh... Aug 05 16:13:01 http://microcom.port5.com/m102.tar.gz Aug 05 16:13:18 Does not match the pattern: http://.*/.*/ Aug 05 16:13:32 There's a missing path. Aug 05 16:13:40 aha Aug 05 16:14:16 It's a dumb pattern (in all cases). Try http://.*/ Aug 05 16:20:08 autobuild is working now - dunno what happened Aug 05 16:21:10 Two builds in parallel? (That would explain perl getting built twice). Aug 05 16:21:33 shouldn't happen - autobuild locks itself out Aug 05 16:21:46 (there is a very small window) Aug 05 16:22:27 NAiL: pong? Aug 05 16:22:44 jbowler-away: can you fix the nslu2-mirrors.bbclass if that's the root of the problem? Aug 05 16:22:47 .* is greedy, try http://([^/]+/)+ Aug 05 16:23:09 * rwhitby-away goes out to enjoy the day Aug 05 16:23:58 .* is meant to be greedy in this context. The problem is that fetch.py apparently expects two wildcard patterns and doesn't work otherwise. Aug 05 16:24:13 So the microcom path cannot be handled. Aug 05 16:24:29 kolla: Can't you pull a cable from teknostallen over here? Aug 05 16:24:39 ah, hehe Aug 05 16:25:24 What's uninett's pipe nowadays anyway? 2.5gbit, innit? Aug 05 16:27:20 yes Aug 05 16:27:53 and 10gbit with nordunet iirc Aug 05 16:28:27 bastards ;) Aug 05 16:28:37 "Memory: 62976KB available (1587K code, 151K data, 80K init)", thats all nice and stuff, but since then panics just after mounting jffs2... I need [g2]... Aug 05 16:28:38 Now, pull some cables my way Aug 05 16:30:04 I'd like to sneak a server in there somewhere atleast ;-P Aug 05 16:30:11 ugh, I was supposed to go to sleep Aug 05 16:31:37 OpenSlug-2.5-beta is now 90381a674a1aead7590d9fe88885d9a57dea4e9e Aug 05 16:31:59 (microcom fix) Aug 05 16:34:41 all these usb lcds are quite pricy (more expensive than a slug) :-( Aug 05 16:35:20 Basic LCDs are quite pricey Aug 05 16:35:54 Im thinking of hooking up a nokia 3310 lcd to a slug Aug 05 16:36:13 really cheap, and graphical, and simple to code for Aug 05 16:37:37 :) Aug 05 16:38:44 A basic 128x64 pixel monochrome (yellow) module is $28.60 Aug 05 16:40:03 http://tinyurl.com/cbxtu looks quite nice (it's a german shop, just look at the pictures...) Aug 05 16:40:38 * jacques just spent more than that on motor oil. :-\ Aug 05 16:42:29 you could add speex console support in the kernel along with usb sound support ;) Aug 05 16:43:39 rotfl Aug 05 16:44:37 openslug tinderbox is green again Aug 05 16:44:57 what is this tinderbox anyways? Aug 05 16:45:47 http://ewi546.ewi.utwente.nl/OE_qa/OpenEmbeddedBuild/status.html Aug 05 16:46:36 aha Aug 05 16:46:50 rwhitby-treo: is nudi generating this? Aug 05 16:47:21 hey that's neat! Aug 05 16:47:34 jacques! Aug 05 16:48:54 hi ka6sox-office :-) I'm in and out of the kitchen for a bit Aug 05 16:49:20 nudi gives info to #oe Aug 05 16:50:22 all depends on the autobuilds Aug 05 16:53:53 l Aug 05 16:53:56 er k Aug 05 17:01:22 ssh babysmurf.kolla.no Aug 05 17:01:24 kinky, huh? Aug 05 17:45:00 NAiL: how is the build from scratch going? Aug 05 17:47:03 the /etc/syslog.conf and /etc/init.d/syslog* creates options that the syslogd doesnt support.. so it doesnt start :P Aug 05 17:47:59 I thought jbowler fixed that in latest? Aug 05 17:54:50 ah, perhaps I'm not on latest Aug 05 18:15:29 /etc/syslog.conf is a CONFFILE now, so a package upgrade won't overwrite it - and the bad options are commented out. Aug 05 18:23:01 yes, I commented them out here as well Aug 05 19:33:47 slugbug is broken? Aug 05 19:37:36 in what way? the page comes up for me Aug 05 19:37:48 NAiL: how did the build from scratch go? Aug 05 19:37:59 if I'm logged in, I see a bunch of sql-messages, if I query a bug Aug 05 19:38:37 http://slugbug.nslu2-linux.org/bug.php?op=show&bugid=243 works for me Aug 05 19:38:58 as does http://slugbug.nslu2-linux.org/query.php?projects=&versions=&components=&order=severity.sort_order&sort=asc&savedqueryname=&op=doquery Aug 05 19:39:12 http://pastebin.ca/19432 Aug 05 19:39:23 that's what I get, I can hit reload as often as I want Aug 05 19:39:57 http://slugbug.nslu2-linux.org/bug.php?op=show&bugid=241 works for me, sorry :-( Aug 05 19:41:15 then it must be my bad karma. :) Aug 05 19:43:00 night **** ENDING LOGGING AT Fri Aug 05 23:59:56 2005