**** BEGIN LOGGING AT Fri Jul 15 23:59:56 2005 Jul 16 04:21:06 mickeyl: looks like jbowler and [cc]smart have the go-ahead to put forward a proposal for the populate-var replacement (e.g. an /etc/volatile.d/... style directory where packages can drop a config file to get stuff done in /var) ... ? Jul 16 04:22:29 yes. i didn't have time to answer yet but it gets my full support Jul 16 04:22:58 this way we do it in a fine granular, extensible way without having to worry about doing things too different in the different distributions Jul 16 04:24:05 Cool. Jul 16 04:25:22 I'd like us to come up with something similar in methodology for initscripts - e.g. where you have a table that picks up a number from the available initscripts Jul 16 04:25:27 mickeyl: fully agree, and there can be a very simple base script which sets up /var, and which can be customised for each distro. then packages add to that basic configuration by dropping in files. Jul 16 04:25:43 bbiab Jul 16 04:26:03 mickeyl: table? Jul 16 04:26:31 [g2] Jul 16 04:26:53 <[g2]> dyoung, Morning sunshine! Jul 16 04:26:58 so I have 2.6.12.2 console now Jul 16 04:27:09 <[g2]> dyoung is THE MAN! Jul 16 04:27:19 rwhitby-away: a table or a list to pick from the initscripts rather than having a custom overload of initscripts_1.0.bb Jul 16 04:27:24 * [g2] hugs dyoung Jul 16 04:28:07 unfortunately, my brain is too dead right now to know what to do with this file. Jul 16 04:28:10 help please. Jul 16 04:28:17 its just nslu2-setup.c . Jul 16 04:28:30 * rwhitby-away puts the kids to bed, and then will have time to do stuff Jul 16 04:28:37 dyoung: feel free to email it to me if you like Jul 16 04:28:45 <[g2]> rwhitby-away, give them lots of love Jul 16 04:28:56 [g2]: always do :-) Jul 16 04:29:06 dyoung: have you upgraded to 0.20 ? Jul 16 04:29:20 rwhitby-away, Yeah, I did that a few h ours ago. Jul 16 04:29:26 it even worked. :-) Jul 16 04:29:29 if so, then just put it in the 2.6.12.2 dir and it should override the one in 2.6.11.x Jul 16 04:29:40 I think ... Jul 16 04:29:46 bbiawhile Jul 16 04:29:47 okay, lemme try that and rebuild the sucker and make sure it still works. Jul 16 04:29:53 <[g2]> yeah that's what worked during my testing Jul 16 04:30:22 is that nslu2-foo or openslug-foo ? Jul 16 04:30:37 <[g2]> dyoung, can you nslu2-foo Jul 16 04:30:43 <[g2]> doh... Jul 16 04:30:54 <[g2]> nslu2-foo and can you e-mail me the file ? Jul 16 04:31:04 ah hell Jul 16 04:31:07 my four year old is driving his speed-racer cardboard box into the bedroom :-) Jul 16 04:31:17 Bwaaaa Jul 16 04:31:29 in no time, you'll be teaching him to drive. Jul 16 04:31:31 sigh Jul 16 04:31:45 <[g2]> with is modded firmware for advanced timing :) Jul 16 04:31:49 <[g2]> it's Jul 16 04:32:06 or retarded timing if he has a turbo. :-) Jul 16 04:32:07 he even has a paper plate with the A to G buttons for the steering wheel :-) Jul 16 04:32:22 Man, my cardbox box wasnt that cool. Jul 16 04:32:26 (that one's for the speed racer afficionados) Jul 16 04:32:41 <[g2]> go speed racer, go! Jul 16 04:33:56 okay, testing with patch in place. Jul 16 04:34:06 and trying to remember how to email it.... Jul 16 04:34:45 dyoung: are you gonna fix the irq26 problem while you're on a roll ? Jul 16 04:35:04 I heard a rumour that the irq26 is a red herring. Jul 16 04:35:16 <[g2]> that's what lennert said Jul 16 04:35:35 but I dont fully understand the issue yet. Jul 16 04:35:45 <[g2]> he think we should #ifdef NSLU2... it out Jul 16 04:37:37 [g2] its close to finishing the testing, lemme make sure its really working before I send it to you. Jul 16 04:40:50 flashing Jul 16 04:41:22 while I was at it though, I did start rewriting upslug to use pcap instead of AF_PACKET. Jul 16 04:41:32 but that started to hurt too much. Jul 16 04:42:46 (instead of netpacket that is...) Jul 16 04:43:44 Uncompressing Linux.......................................................... done. Jul 16 04:43:44 Linux version 2.6.12.2 (openslug@openslug.org) (gcc version 3.4.4) #8 Sat Jul 16 05 Jul 16 04:43:49 <[g2]> dyoung, you don't use reflash ? Jul 16 04:44:00 I dotn know how to use reflash Jul 16 04:44:10 <[g2]> reflash -i image Jul 16 04:44:18 <[g2]> reflash -j jffs2 Jul 16 04:44:24 <[g2]> reflash -k kernel Jul 16 04:44:33 that would imply that I have enough space someplace on the thingee to put the image. Jul 16 04:44:47 <[g2]> I think you can turnup ram Jul 16 04:44:59 <[g2]> and cp to a ramdisk Jul 16 04:45:05 <[g2]> it's only 8MB Jul 16 04:45:15 well, no matter, upslug is quite reliable now that jbowler patched the hell out of it. Jul 16 04:45:23 and it still works. Jul 16 04:45:38 so lets see if I can push it. Jul 16 04:46:09 <[g2]> can you e-mail me the nslu2-setup.c so I can diff the changes ? Jul 16 04:46:43 sent Jul 16 04:47:23 its a real minor change. Jul 16 04:47:47 and in fact, you suggested it. :-) Jul 16 04:51:16 okay its pushed... Jul 16 04:53:51 <[cc]smart> mickeyl: still in ? Jul 16 04:54:06 <[cc]smart> mickey|: still in ? Jul 16 04:54:14 yep Jul 16 04:54:18 <[cc]smart> hi Jul 16 04:54:35 <[cc]smart> seems the discussions already went a step further then where i currently are Jul 16 04:54:50 <[cc]smart> that is, the suggestion has been made to use a directory where you can drop i files Jul 16 04:55:03 <[cc]smart> but the current version i have just uses a single file Jul 16 04:55:27 <[cc]smart> the original intention was to get to the possibility to have different setups and be able to modify the configuration. Jul 16 04:55:44 <[cc]smart> my guess is, that this is still an acceptable first step Jul 16 04:56:21 <[cc]smart> if so, how and where should i feed it (i have nslu monotone access, but still need to understand what goes where and how) Jul 16 04:56:56 yeah, i think we can start with the per-file thing and implement the full directory based solution in a 2nd step Jul 16 04:57:14 i think it'd be ok if it comes in via the nslu propagates Jul 16 04:58:16 <[cc]smart> into which apapckge/directory should the script and the basic config go, and i which .bb should it be placed ? Jul 16 04:58:27 <[cc]smart> s/i/in/ Jul 16 04:58:34 hmm good question. Jul 16 04:58:38 i need to see it before i can answer that Jul 16 04:58:44 please send the stuff to me first Jul 16 04:58:45 <[cc]smart> pastebin ? Jul 16 04:58:48 ya Jul 16 04:58:50 pastebin Jul 16 05:00:14 <[cc]smart> script in /etc/init.d: http://pastebin.com/314422 Jul 16 05:00:57 <[cc]smart> configfile in /etc/default/ : http://pastebin.com/314423 Jul 16 05:01:06 i see Jul 16 05:01:35 ok, that needs to go into initscripts, since we already have the 'devices' script there that does some similar things for some machines Jul 16 05:02:13 so, initscripts/initscripts-1.0/populate-fs Jul 16 05:02:16 or so Jul 16 05:09:18 dyoung: did CIA miss your nslu2-setup patch? Jul 16 05:10:00 CIA must be on strike. It should be in there. Jul 16 05:11:28 ( cd openembedded ; monotone update ) Jul 16 05:11:28 monotone: already up to date at cc91df51dc068d2433e770fb9fa673e7aaa95e4e Jul 16 05:11:32 is that your head? Jul 16 05:11:51 d20d02a156ee154e335f3df1c688d3aafd579f45 dyoung@nslu2-linux.org Jul 16 05:14:05 dyoung: and you did a monotone push to monotone.nslu2-linux.org ? Jul 16 05:15:00 I did cd openembedded; monotone add linux/nslu2-kernel/2.6.12/nslu2-setup.c; monotone commit -m "stuff"; monotone push Jul 16 05:15:03 dyoung: your push is referencing 'org' instead of the new 'org.openembedded.* org.nslu2-linux.*" Jul 16 05:15:47 solutiuon? Jul 16 05:16:00 did you do the recommended pull to set your default-include-pattern ? Jul 16 05:16:24 mt unset database default-include-pattern Jul 16 05:16:39 mt pull monotone.nslu2-linux.org org.openembedded.* org.nslu2-linux.* Jul 16 05:17:00 then your next mt push should reference those branch glob patterns for the push, instead of "org" Jul 16 05:17:30 or use the latest master makefile Jul 16 05:17:47 but you still have to reset the default-include-pattern Jul 16 05:19:40 okay I did those 2 command then a push. Jul 16 05:21:45 dyoung: what branches did it say it was pushing? Jul 16 05:22:37 all of them, .dev, .dreambox .nslu2-linux... etc Jul 16 05:23:27 dyoung: looks like it worked - let's see if CIA picks it up Jul 16 05:24:09 There are push hooks for CIA to use now, right? Jul 16 05:24:36 so do I need to do that each time, or can I jus tmonotone push now? Jul 16 05:24:45 (*since I did it once already) Jul 16 05:25:38 you can just say "make push" now ;-) Jul 16 05:44:38 dyoung: it's building now Jul 16 05:45:39 ok I was hoping to see someone else had success. Jul 16 05:45:43 before crashing out Jul 16 05:46:13 it got nslu2-setup.c from the right place ... Jul 16 06:08:14 <[cc]smart> rwhitby: should one use all three patterns to fix monotone as you suggested above or should one like me only take one of the patterns ? Jul 16 06:09:22 mt pull monotone.nslu2-linux.org org.openembedded.* org.nslu2-linux.* Jul 16 06:11:35 <[cc]smart> wow Jul 16 06:11:42 <[cc]smart> distant hunderstorm here Jul 16 06:11:44 <[cc]smart> in summer Jul 16 06:12:07 <[cc]smart> those thunder rolls are REALLY low freq. base deep hums Jul 16 06:12:11 <[cc]smart> never heard it like that Jul 16 06:12:46 <[cc]smart> astonishing Jul 16 06:18:34 you know I would really like root's home directory to be in /root instead of /home/root .... Jul 16 06:19:10 cause I usually mount /home from somewhere else, and that can stuff up root's ssh key existence ... Jul 16 06:59:00 <[g2]-away> good nite sunshine sweet dreams Jul 16 06:59:43 [g2]-away: NOTE: package perl-native-5.8.4-r0: task do_compile: started Jul 16 07:00:10 <[g2]-away> DaKa2, I saw that awhile ago Jul 16 07:00:17 :) Jul 16 07:00:24 <[g2]-away> you also mentioned you didn't think it would work Jul 16 07:00:36 well, Im starting to think it will Jul 16 07:00:45 its more than half through the compile Jul 16 07:01:23 after this there is just the perl package left to try Jul 16 07:02:11 <[g2]-away> I think perl-native would work no ? Jul 16 07:02:21 does it build an ipk? Jul 16 07:02:32 or just place stuff in staging? Jul 16 07:02:33 <[g2]-away> I think everything builds an ipkg Jul 16 07:02:40 ahh, then it should work Jul 16 07:02:52 if it places the right stuff in the ipk Jul 16 07:03:04 havn't really looked at it Jul 16 07:03:33 <[g2]-away> well maybe not on the ipkg :( Jul 16 07:05:08 <[g2]-away> check tmp/deploy/ipk for natives Jul 16 07:05:32 hm.. none Jul 16 07:05:45 <[g2]-away> oh well Jul 16 07:05:54 well, if this works, bb perl should work to Jul 16 07:05:56 maybe :-) Jul 16 07:05:57 <[g2]-away> so how did you get db3 to go ? Jul 16 07:06:03 I did nothing Jul 16 07:06:13 it worked with my ipkg installed autoconf Jul 16 07:06:33 <[g2]-away> hmmmm Jul 16 07:06:38 :) Jul 16 07:06:53 <[g2]-away> I'd want to sort that out Jul 16 07:08:22 <[g2]-away> did you notice dyoung push the fix for the 2.6.12.2 kernel messages ? Jul 16 07:08:30 yep Jul 16 07:09:03 I only have one slug with serial, and I left that one at work :-( Jul 16 07:39:27 NOTE: package perl-native-5.8.4-r0: task do_compile: completed Jul 16 07:41:42 Congratulations Jul 16 07:42:07 if just the staging works I can start on perl Jul 16 07:43:51 and since I'm for some reason is doing this on my NormalSlug, its going to take some time Jul 16 07:44:21 NOTE: package perl-native-5.8.4: completed Jul 16 07:51:31 <[g2]> DaKa2, we should be able to build perl directly Jul 16 07:57:55 yes.. Jul 16 07:58:01 NOTE: package perl-5.8.4-r2: task do_compile: started Jul 16 07:58:15 <[g2]> DaKa2, way cool Jul 16 07:58:38 <[g2]> DaKa2, can you post a little howto with the steup you are doing ? Jul 16 07:58:44 but, its going to take a few hours Jul 16 07:58:57 in wiki? or just here Jul 16 07:59:06 things are a bit sketchy right now Jul 16 07:59:46 <[g2]> well we've got the install.sh script after building everything right ? Jul 16 08:00:12 <[g2]> that gets the native compile environ all setup Jul 16 08:00:26 hm.. yes Jul 16 08:00:37 <[g2]> then it's build perl and python Jul 16 08:00:58 <[g2]> then bb setup Jul 16 08:01:23 when we have perl and python packages built nativly people can just install them Jul 16 08:01:55 <[g2]> absolutelty, but we need to build it on the build machine and deploy from there Jul 16 08:02:05 yep Jul 16 08:02:13 <[g2]> hence the recipe Jul 16 08:02:51 ah, yep Jul 16 08:03:44 <[g2]> I'm gonna miss you when you go back to work Jul 16 08:04:27 hehe, well, I have a computer there too you know :-) Jul 16 08:04:57 <[g2]> well maybe you can gw out or just irc out from there :) Jul 16 08:05:16 I am ircing from work right now so Jul 16 08:05:21 <[g2]> cool Jul 16 08:05:36 I always have a terminal with a screened irssi on my work computer Jul 16 08:05:40 on the other monitor Jul 16 08:06:11 this irssi is running on one of our servers Jul 16 08:09:53 Just like this irssi is running on my mirror ;) Jul 16 08:10:51 I'm kinda hoping to sneak a slug into the server room and use that for everything ;) Jul 16 08:11:23 I'm going to put a slug into the server room Jul 16 08:11:42 when things are so stable I dont need physical access Jul 16 08:13:22 could probably find a spare serial port on a server and connect that too Jul 16 08:16:46 and I could probably reprogram my always on PIC to take a few io pins for reboot and power off/on Jul 16 08:16:56 hm.. Jul 16 08:25:45 [g2]: a tip.. don't install psyco.. Jul 16 08:30:47 DaKa2, what are you seeing with psyco? Jul 16 08:31:55 on the slug: Jul 16 08:31:56 Segmentation fault Jul 16 08:32:00 :) Jul 16 08:32:17 maybe I did something wrong Jul 16 08:32:42 <[g2]> I hope it gets to the server room soon :) Jul 16 08:34:09 [g2]: the devices you are making, will they be binary compatible with the slug, could they be used for native package builds? Jul 16 08:34:34 <[g2]> sure Jul 16 08:34:46 fantastic :-) Jul 16 08:34:55 <[g2]> just the kernel and bootloader will be tweaked a little Jul 16 08:35:13 <[g2]> different USB host chip Jul 16 08:35:19 <[g2]> but should work the same Jul 16 08:35:48 <[g2]> plus with the CF one could install a microdrive Jul 16 08:36:05 ahh, much better i/o Jul 16 08:36:44 <[g2]> there will need to be testing Jul 16 08:37:30 <[g2]> I'm pretty excited about it Jul 16 08:38:07 hm.. I wonder how my Sandisk Extreme III 2Gb would perform.. Jul 16 08:39:30 <[g2]> I think the II is supposed to support a minimum of 20MB/sec Jul 16 08:40:28 I have a 1Gb Extreme II, too iirc Jul 16 08:42:37 <[g2]> DaKa2 has lots of toys :) Jul 16 08:42:50 :) yup Jul 16 08:45:32 I started working when I was 16, and at that time my expences every month were close to 0 :-) Jul 16 08:46:32 <[g2]> was that a long time ago ? Jul 16 08:47:11 well, no.. Im 21 now Jul 16 08:47:31 <[g2]> at that age it's a lifetime Jul 16 08:47:39 :) Jul 16 08:49:31 * [g2] is probably older than your dad :) Jul 16 08:50:15 well, I think he is like, 65.. so... Jul 16 08:50:33 <[g2]> heh Jul 16 08:50:51 <[g2]> I'm not quite *that* old :) Jul 16 08:52:40 <[g2]-away> bbl Jul 16 08:55:11 :) Jul 16 08:59:00 Im guessing there is 30-40 minutes left of my perl compile Jul 16 09:19:54 uhm.. the perl package seems to be doing do_install in do_compile ... Jul 16 09:22:15 NOTE: package perl-5.8.4-r2: task do_package: started Jul 16 10:07:44 rwhitby-asleep: the FHS says that /root is the root home directory. openslug already has a /etc/passwd patch for the root passwd (so it could change /home/root too). Jul 16 10:08:44 IMO OE should probably use that too, but changing it is a significant step because it may break some systems. Jul 16 10:09:02 It's most important on a file server because /home is often mounted (so not available during boot). Jul 16 10:27:06 damn.. perl takes forever to do_package Jul 16 10:27:17 of course.. it makes 800 packages or something.. Jul 16 10:50:53 <[g2]-away> NAiL, Jul 16 10:50:55 <[g2]-away> ping Jul 16 10:51:20 <[g2]-away> DaKa2, Jul 16 10:51:22 <[g2]-away> ping Jul 16 10:51:24 yes? Jul 16 10:51:46 <[g2]-away> do you know what the line is in repo fix the cvs date for gnu-config ? Jul 16 10:52:03 hm. yes... Jul 16 10:53:14 uhm. no.. Jul 16 10:53:21 <[g2]-away> heh Jul 16 10:55:05 CVSDATE_gnu-config ?= "20050713" Jul 16 10:55:17 <[g2]-away> that's the one I think Jul 16 10:55:33 nslu2.conf Jul 16 10:56:00 Packaged contents of perl into /home/slug/openslug/tmp/deploy/ipk/perl_5.8.4-r2_armeb.ipk Jul 16 10:56:11 <[g2]-away> COOL! Jul 16 10:56:19 it just have 700 packages more to package Jul 16 10:56:24 then its done... Jul 16 10:56:29 perl is way to big Jul 16 10:57:22 <[g2]-away> heh feature creep Jul 16 10:57:33 <[g2]-away> over a decade Jul 16 10:57:40 10 seconds / package... Jul 16 10:58:25 actually, more.. Jul 16 10:58:45 <[g2]-away> THX for the CVSDATE Jul 16 11:33:29 [g2]-away: pomng Jul 16 11:33:31 pong, even Jul 16 12:12:22 hm.. bb perl seems to take around 6 hours on a 133mhz slug Jul 16 12:12:55 or 7.. Jul 16 12:13:24 298 ipk built, of 700something Jul 16 12:13:39 700? Jul 16 12:13:53 ls /home/slug/openslug/tmp/deploy/ipk/perl* | wc -l Jul 16 12:13:54 726 Jul 16 12:14:12 thats on my cross build Jul 16 12:14:13 ls /home/slug/openslug/tmp/deploy/ipk/perl* | wc -l Jul 16 12:14:14 298 Jul 16 12:14:20 that is my slug Jul 16 12:14:22 so far Jul 16 12:14:34 aaah Jul 16 12:56:53 511 packages done... Jul 16 12:57:39 but.. I have a perl-lib package, but on the cross build I have one called libperl5.. strange... Jul 16 13:20:36 <[g2]-away> I think things on the fatturbo will go faster Jul 16 13:21:08 <[g2]-away> I've got a 533Mhz IXP425 with 256MB on loan for a couple more days Jul 16 13:21:23 <[g2]-away> If I can get that fired up I'm sure it'd crank Jul 16 13:22:57 Oooooo.. fire it up then :-) Jul 16 13:24:03 thats four times faster than this one, and 8 times more memory Jul 16 13:24:49 <[g2]-away> I can fire up FatTurbo Jul 16 13:25:32 <[g2]-away> That's to 2x the speed and 2x the memory Jul 16 13:25:37 well, perl is done in 20 minutes Jul 16 13:25:55 someone has to make an apache bb Jul 16 13:26:08 or is there one somewhere? Jul 16 13:26:41 <[g2]-away> well once we get it some everyone can run little bitbake's easily things will get rock'n Jul 16 13:26:54 yup Jul 16 13:27:02 <[g2]-away> speaking of witch..... Jul 16 13:27:04 Im trying to write down what is needed Jul 16 13:27:07 <[g2]-away> which ;) Jul 16 13:27:27 and somehow Im not sure of whats needed Jul 16 13:27:39 <[g2]-away> well lets try right now Jul 16 13:27:41 Ive done so much strange things to try and make this work Jul 16 13:27:49 well: Jul 16 13:27:53 ipkg install ipkg-utils libtool quilt pkgconfig Jul 16 13:28:05 (probably needs some overrides) Jul 16 13:28:18 local.conf: Jul 16 13:28:18 ASSUME_PROVIDED = "libtool-cross libtool-native automake-native autoconf-native \ Jul 16 13:28:21 quilt-native ipkg-utils-native gnu-config-native pkgconfig-native \ Jul 16 13:28:24 virtual/armeb-linux-gcc virtual/libc update-rc.d" Jul 16 13:28:25 <[g2]-away> I've got everything in the shell script Jul 16 13:28:27 BUILD_ARCH = armeb Jul 16 13:28:40 <[g2]-away> lets start from there and a clean install Jul 16 13:29:16 that is all I have besides DevInst.sh Jul 16 13:29:26 <[g2]-away> Ok... Jul 16 13:29:40 Ive probably missed one or two things Jul 16 13:30:14 I dont know if this is needed, probably not: Jul 16 13:30:14 cd /usr/bin Jul 16 13:30:14 ln -s libtool armeb-linux-libtool Jul 16 13:30:26 but dor some reason I did that Jul 16 13:30:34 but we should try without Jul 16 13:30:57 and this is in my file too: Jul 16 13:30:57 # is this really needed? Jul 16 13:30:58 #mkdir -p /home/slug/openslug/tmp/cross/armeb-linux/lib Jul 16 13:30:58 #ln -s /usr/lib/libstdc++.la /home/slug/openslug/tmp/cross/armeb-linux/lib/libstdc++.la Jul 16 13:31:39 <[g2]-away> I suggested that Jul 16 13:31:57 hm.. no, that was another one, I think.. Jul 16 13:32:20 that was something with armeb-linux-libtool in the staging Jul 16 13:32:49 but that was just one package (jpeg), that really should be fixed instead Jul 16 13:33:30 <[g2]-away> ok... let's backup a second Jul 16 13:33:36 :) Jul 16 13:33:58 <[g2]-away> the goals are a) a native perl package build with bitbake Jul 16 13:34:25 <[g2]-away> b) a native python with bb Jul 16 13:34:36 <[g2]-away> all packaged up so we can release them Jul 16 13:34:46 yep, a) is known to work without changes to perl.bb Jul 16 13:34:52 <[g2]-away> and we need the recipe for the build machine to run Jul 16 13:35:20 well, we need a whiped build-machine, to test what is really needed Jul 16 13:35:23 <[g2]-away> a) isn't known to run yet is it ? Jul 16 13:35:28 heh, no Jul 16 13:35:38 but in a few minutes we will know Jul 16 13:35:41 <[g2]-away> the other perl builds... but doesn't run Jul 16 13:36:18 <[g2]-away> is the build almost finished ? Jul 16 13:36:25 712 of 726 packaged Jul 16 13:36:42 <[g2]-away> cool Jul 16 13:37:15 Im just interested if its going to rename perl-lib to libperl5... Jul 16 13:37:35 because I have no perl-lib in the deploy on my crossbuilt.. Jul 16 13:40:27 <[g2]-away> nod, so should be build python first ? Jul 16 13:40:49 <[g2]-away> s/be/we Jul 16 13:41:02 well, seems like the next logical step Jul 16 13:41:17 <[g2]-away> but most of the python runs from the oe install Jul 16 13:41:27 <[g2]-away> or did you have to native build it Jul 16 13:41:54 well, I built it, but maybe isn't needed Jul 16 13:41:59 maybe should try without.. Jul 16 13:42:09 but ipkg install python* Jul 16 13:42:55 <[g2]-away> hmmm... I'm a little conflicted Jul 16 13:43:07 hm? Jul 16 13:43:25 <[g2]-away> I know the native builds of python and perl pass all but 1 test Jul 16 13:43:32 <[g2]-away> which is expected to fail Jul 16 13:44:01 hm.. bad news... perl doesn't want to play.. Jul 16 13:44:05 <[g2]-away> but we haven't verified that with the cross built Jul 16 13:44:14 <[g2]-away> what's the error ? Jul 16 13:44:23 same as crossbuilt Jul 16 13:44:28 perl: symbol lookup error: /usr/lib/libperl.so.5: undefined symbol: __eqdf2 Jul 16 13:44:56 <[g2]-away> I think that's the exact same error from the cross ipkg Jul 16 13:45:10 yup.. Jul 16 13:45:15 so.. something is wrong.. Jul 16 13:45:30 maybe one of the patches, or something else Jul 16 13:45:42 <[g2]-away> well it's consistenly broken Jul 16 13:45:56 <[g2]-away> which is actually a good thing in my mind Jul 16 13:46:14 hm.. we are building without patches nativly, maybe we should remove the patches and try again Jul 16 13:46:25 Ill better look at what the bb actually is doing Jul 16 13:46:59 <[g2]-away> or just create a simple new bb Jul 16 13:47:09 hm.. good idea Jul 16 13:47:27 but.. Jul 16 13:47:42 does any packages depend on all theese perl-module-* ? Jul 16 13:48:01 <[g2]-away> we can package it up the same Jul 16 13:48:09 do we have to? :-) oh well Jul 16 13:48:12 <[g2]-away> dunno if we want to Jul 16 13:48:31 <[g2]-away> I think it's more than fine to start-off with a full-perl and a full python Jul 16 13:48:36 Ill try making something simple first, and if that works we should try more strange things with it Jul 16 13:49:42 <[g2]-away> I'd like to understand what we should install Jul 16 13:50:08 for native development? Jul 16 13:50:17 <[g2]-away> doing a native build of perl with ./configure --prefix=/usr/bin Jul 16 13:50:40 <[g2]-away> then packaging up the make instal/test stuff would be the "full perl" right ? Jul 16 13:50:51 well.. actually Jul 16 13:50:57 sh Configure -de Jul 16 13:51:07 make Jul 16 13:51:10 make install Jul 16 13:51:11 done Jul 16 13:51:26 <[g2]-away> Configure -de ? Jul 16 13:51:41 just have to make it put stuff in the image dir Jul 16 13:51:52 d = default, e = something else.. Jul 16 13:53:25 <[g2]-away> Oh.. you mean build perl locally and then as part of the .bb scripts have it install in to the tmp/work/perl-.... area right ? Jul 16 13:53:36 <[g2]-away> ditto python Jul 16 13:53:47 yes, something like that Jul 16 13:54:11 Ill try making something with perl 5.8.7 Jul 16 13:54:33 <[g2]-away> except for the chicken-egg thing with python that makes sense Jul 16 13:54:35 perl-openslug_5.8.7.bb ? Jul 16 13:55:42 <[g2]-away> actually since it's all local to the slug we can just have a perl_5.8.7.bb and it'll use the latest right ? Jul 16 13:56:28 well, if it should go back into the rep? Jul 16 13:56:44 <[g2]-away> ok... this is starting to make a bunch of sense to me. Jul 16 13:56:54 <[g2]-away> A) devinst.sh Jul 16 13:57:06 <[g2]-away> b) add user Jul 16 13:57:10 <[g2]-away> c) su user Jul 16 13:57:23 <[g2]-away> d) from user natively build perl/python Jul 16 13:57:58 <[g2]-away> e) install perl/python on system from root Jul 16 13:58:19 <[g2]-away> f) install extra's to setup bitbake Jul 16 13:58:44 <[g2]-away> g) run bitbake to patch up perl/python install into bb's work area Jul 16 13:58:48 "install extra's" == what? Jul 16 13:59:03 <[g2]-away> h) put those packages to the feed and we've boot strapped Jul 16 13:59:35 <[g2]-away> ipkg install ipkg-utils libtool quilt pkgconfig Jul 16 13:59:39 ahh Jul 16 14:00:07 <[g2]-away> after that's done, Jul 16 14:00:15 and, they have to be built first, I think they are all in oe-symlinks, but not openslug-packages Jul 16 14:00:20 <[g2]-away> then users can ipk install a full perl and python Jul 16 14:00:47 <[g2]-away> that's trivial Jul 16 14:00:58 yep, but has to be done :-) Jul 16 14:01:31 <[g2]-away> btw I think [cc]smart pushed the turnup fix earlier today Jul 16 14:01:41 yup, saw that Jul 16 14:02:01 things are moving forward :-) Jul 16 14:02:17 <[g2]-away> it's been like this for a year Jul 16 14:02:41 :) Jul 16 14:02:49 <[g2]-away> ~slugtime Jul 16 14:02:57 <[g2]-away> no bot in here Jul 16 14:03:03 <[g2]-away> nslu2-linux Jul 16 14:03:04 hehe, 40h/day Jul 16 14:03:12 <[g2]-away> yup Jul 16 14:04:21 <[g2]-away> is it 10PM or 11PM there ? Jul 16 14:04:36 11PM here Jul 16 14:04:44 or, 23:04 Jul 16 14:04:49 <[g2]-away> purrfect Jul 16 14:05:09 <[g2]-away> I think you, I and dyoung are each 1/4 globe away Jul 16 14:05:22 :) Jul 16 14:05:48 <[g2]-away> and rwhitby-asleep is about 3.5/4.5 from you Jul 16 14:06:01 and I think I can make a working perl bb Jul 16 14:06:10 that compiles it and all Jul 16 14:09:19 <[g2]-away> I'm on the fence about the default prefix for perl and python being /usr/local/bin Jul 16 14:09:37 <[g2]-away> i'd be really nice to make the update-alternatives work properly Jul 16 14:09:41 my perl installed into usr/bin without anything.. Jul 16 14:10:01 <[g2]-away> yeah that's the one that didn't work right ? Jul 16 14:10:12 no, my real native build Jul 16 14:10:23 not the bb one Jul 16 14:10:42 <[g2]-away> nahh... Jul 16 14:10:49 <[g2]-away> type which perl Jul 16 14:11:00 could be because I have no /usr/local the Configure put it in /usr/bin Jul 16 14:11:01 <[g2]-away> except you install the ipkg Jul 16 14:11:58 I had to compile perl manually first and that got into /usr/bin Jul 16 14:12:29 <[g2]-away> you didn't use the default config Jul 16 14:12:45 well.. I did: Jul 16 14:12:49 sh Configure -de Jul 16 14:12:50 make Jul 16 14:12:51 make install Jul 16 14:13:18 <[g2]-away> that not the same as sh ./Configure Jul 16 14:13:42 no.. sh Configure asks 200 questions.. Jul 16 14:14:02 -d : use defaults for all answers. Jul 16 14:14:02 -e : go on without questioning past the production of config.sh. Jul 16 14:14:05 <[g2]-away> ./Configure doesn't ask any Jul 16 14:14:11 uhm.. not? Jul 16 14:14:22 <[g2]-away> maybe it's ./Configure.gnu Jul 16 14:14:55 I have no Configure.gnu.. Jul 16 14:15:03 perl-5.8.7? Jul 16 14:15:10 <[g2]-away> YEP Jul 16 14:15:44 ahh, ./configure.gnu Jul 16 14:15:57 <[g2]-away> yeah that'd be it Jul 16 14:16:05 it runs sh Configure -ds -e Jul 16 14:16:14 the same I run basically.. Jul 16 14:16:23 <[g2]-away> but a diffent prefix Jul 16 14:16:33 -s : silent mode, only echoes questions and essential information. Jul 16 14:17:46 ah, well, ./configure.gnu has --prefix Jul 16 14:18:23 and that runs sh Configure -ds -e -Dprefix=/usr Jul 16 14:19:30 then we just need to find the -D option for installing in one dir, but later running it from another Jul 16 14:19:52 (installing into tmp/work, later to be ipkg installed to /usr) Jul 16 14:25:10 <[g2]-away> So the script to build perl/python is really just a add user; su user; wget perl; tar ; ./configure.gnu; make && make test; Jul 16 14:25:47 yes.. you could skip make test, and make install instead Jul 16 14:26:03 <[g2]-away> no we want to test Jul 16 14:26:12 <[g2]-away> and then exit and make install from root Jul 16 14:26:16 ah Jul 16 14:26:17 right Jul 16 14:26:41 <[g2]-away> with procps and the new turnup we should have 100% success Jul 16 14:27:06 <[g2]-away> except for the expected failure in python Jul 16 14:27:11 hopfully yes Jul 16 14:28:34 <[g2]-away> I'll setup the script and run it later tonight Jul 16 14:28:42 ahh, make install DESTDIR=${D} in the bb Jul 16 14:29:09 I'll try and get a bb that somehow makes a working ipk of perl Jul 16 14:30:11 <[g2]-away> ok cool Jul 16 14:30:27 <[g2]-away> so the ./configure.gnu is ok with you ? Jul 16 14:30:32 yes Jul 16 14:30:56 <[g2]-away> time to run out Jul 16 14:31:00 maybe with --prefix=/usr Jul 16 14:31:01 ok Jul 16 14:31:15 see you Jul 16 14:31:35 <[g2]-away> I could do a ./configure.gnu --prefix=/usr Jul 16 14:31:56 <[g2]-away> something similar on python Jul 16 14:31:59 just to be sure Jul 16 14:32:12 mine installed in /usr/bin, but that could just have been me Jul 16 14:32:50 <[g2]-away> ok... back later cheers Jul 16 15:46:04 I should probably get some sleep, perl is compiling, so far so good Jul 16 15:46:12 bbl Jul 16 15:48:07 [g2]-away: DaKa2: none of our ipks should put stuff in /usr/local. /usr/bin/perl is the right choice. Jul 16 15:48:37 good work on the perl BTW. Jul 16 16:27:27 I just hope it'll work Jul 16 16:27:47 a compile takes 2 hours, and packaging it another 2... Jul 16 16:35:12 yes, I remember when we were doing the native compilation of crosstool for the unslung native build environment .... Jul 16 16:36:53 DaKa2, are you turbo-ified? Jul 16 20:10:08 dyoung: yes, on all slugs but this one... used the wrong one when I started seting things up :) Jul 16 20:11:39 Bummers. Jul 16 20:12:01 well, you could say that :-) Jul 16 20:12:58 oh well, 533 of 726 ipkgs built this time Jul 16 20:17:41 hm. and it seems to be working, but the packaging is wrong, perl and the lib ended up in the package perl-misc Jul 16 20:17:50 better redo that part Jul 16 20:18:28 but, really should get some more sleep, bbl **** ENDING LOGGING AT Sat Jul 16 23:59:56 2005