**** BEGIN LOGGING AT Tue Aug 02 23:59:56 2005 Aug 03 01:57:33 good morning all Aug 03 01:57:42 g'day koen Aug 03 02:05:25 Re: the addition of a second "pine" package. We really need to be more careful about modifications to existing packages in OE. It is not acceptable to collide with an existing package, or to modify an existing package without the express permission of the maintainer or the OE core team. Aug 03 02:05:38 'nuff said. Aug 03 02:06:02 and not break a license Aug 03 02:06:32 * rwhitby-away nods Aug 03 02:07:06 * koen stuggles with making a bbclass Aug 03 02:07:08 someone added a pine ? Aug 03 02:07:11 struggles* Aug 03 02:07:42 dyoung: yes, and somebody else added patches for it Aug 03 02:07:51 which turns out to violate the pine license Aug 03 02:08:07 wow, I guess I totally missed that. Aug 03 02:08:09 dyoung: an error was made yesterday. it's fixed now. we move on and learn from it :-) Aug 03 02:09:16 we are now approaching the number of committers where we can get territorial Aug 03 02:11:04 territorial? how do you mean? Aug 03 02:11:35 "or to modify an existing package without the express permission of the maintainer" Aug 03 02:11:53 right. but that's just an open source norm Aug 03 02:12:39 I'd rather call it "considerate" rather than "territorial" :-) Aug 03 02:18:03 Now, what do we need to achieve tonight .... Aug 03 02:28:05 same as every night Aug 03 02:28:24 world domination! Aug 03 02:28:28 Are you thinking what I'm thinking, Koen? Aug 03 02:28:59 :) Aug 03 02:29:40 The Slug Soldiers are waiting... Aug 03 02:30:47 koen, BTW, we have sources.nslu2-linux.org now. I've went ahead and put ctrlproxy-2.6.2.tar.gz there; which is the one with the correct GZIP header. Aug 03 02:31:11 ah, nice Aug 03 02:31:40 I was just working on sourcepkg.bbclass which tars up the patched ${S} Aug 03 02:32:18 ooh nice. Aug 03 02:32:25 prepatched goodness! Aug 03 02:32:38 | tar: bison-2.0/.pc/m4.patch/m4/inttypes-pri.m4: Cannot open: Permission denied Aug 03 02:32:43 stinking quilt Aug 03 02:38:37 NOTE: package bison-native-2.0-r2: task do_tarballing: completed Aug 03 02:38:41 excellent Aug 03 02:46:58 dyoung: pushing the bbclass now Aug 03 02:48:43 Cool. Aug 03 02:49:05 11:48 < CIA-2> koen org.oe.dev * rcb7c16aa... /classes/sourcepkg.bbclass: Aug 03 02:49:05 11:48 < CIA-2> Add sourcepkg.bbclass, a crude way to tar up the patched sources into DL_DIR/sourcepkg and md5sums the result. do INHERIT += "sourcepkg" to get all of this goodness. Aug 03 02:49:09 11:48 < CIA-2> Now I need to find out how to override the Source: field to make it point to said tarballs on your distro's source mirror Aug 03 02:49:10 is there an easy way to tell oe that a specific source file is pre-patched? Aug 03 02:50:07 koen, thats one thing we're noticing. nslu2-mirror.bbclss doesnt seem to work. Probably not licking it right somehow. but I havnt looked myself yet. Aug 03 02:51:23 dyoung: I don't think there's a way to tell it's pre-patched Aug 03 02:54:03 I guess that feature will come eventually. Aug 03 02:54:33 source: and source_patched: or something. Aug 03 02:54:47 source-patched: :-) Aug 03 02:54:57 oh yeah Aug 03 05:21:04 ok, I'm gonna start on Makefile changes for native openslug/ucslugc now ... Aug 03 05:25:54 i'm about to put apex ontop of my redboot. i have jtag working so no problem recovering... Aug 03 05:26:26 i just wanted to check if it is more complex than just moving the apex.bin file into 0x50000000 Aug 03 06:58:08 DaKa2: OE Build Configuration: Aug 03 06:58:08 TARGET_ARCH = "armeb" Aug 03 06:58:08 TARGET_OS = "linux" Aug 03 06:58:08 MACHINE = "nslu2" Aug 03 06:58:08 DISTRO = "openslug-native" Aug 03 06:58:10 TARGET_FPU = "soft" Aug 03 06:58:11 ERROR: Nothing provides ipkg-native Aug 03 06:58:13 ERROR: dependency ipkg-native (for package-index) not satisfied Aug 03 06:58:15 NOTE: no buildable providers for package-index Aug 03 06:58:17 ERROR: dependency package-index (for openslug-native-packages) not satisfied Aug 03 06:58:19 NOTE: no buildable providers for openslug-native-packages Aug 03 07:00:28 uhm.. Aug 03 07:01:01 rwhitby: distro/openslug-native.conf should contain ipkg-native in ASSUME_PROVIDED Aug 03 07:01:29 oops.. Aug 03 07:01:36 wait a minute.. Aug 03 07:04:48 there... Aug 03 07:14:02 DaKa2: got further: NOTE: package package-index-1.0-r0: task do_build: started Aug 03 07:14:02 ERROR: function do_build failed Aug 03 07:14:02 ERROR: see log in /home/slug/openslug/tmp/work/package-index-1.0-r0/temp/log.do_build.937 Aug 03 07:14:02 NOTE: Task failed: /home/slug/openslug/tmp/work/package-index-1.0-r0/temp/log.do_build.937 Aug 03 07:14:02 NOTE: package package-index-1.0-r0: task do_build: failed Aug 03 07:14:41 ok.. whats in the log? Aug 03 07:15:04 it probably want ipkg in staging Aug 03 07:16:54 kg-make-index: not found Aug 03 07:17:02 "/home/slug/openslug/tmp/work/package-index-1.0-r0/temp/run.do_build.937: 437: ip Aug 03 07:17:03 kg-make-index: not found" Aug 03 07:18:12 ok.. what does it use as path in run.do_build.937? Aug 03 07:19:11 ok, we should just let it build ipkg-utils-native Aug 03 07:20:13 probably, if it tries to run it from staging.. Aug 03 07:24:39 rwhitby: actually, it should work, I think.. you do have ipkg-utils installed right? Aug 03 07:30:00 actually, no. Aug 03 07:30:17 it's not in openslug-native.bb :-) Aug 03 07:30:26 rwhitby: bb package-index works for me Aug 03 07:30:50 rwhitby: right, it still needs fixing: look at the wiki Aug 03 07:31:08 ok, package-index doesn't work for me.. Aug 03 07:31:13 ok, building it natively works Aug 03 07:31:15 it needs pythons md5 module Aug 03 07:32:52 rwhitby: python-crypt needs to be added to openslug-native.bb.. and ipkg-utils installed on the slug (and fixed and added to openslug-native.bb) Aug 03 07:34:07 we can leave ipkg-utils off if we like - it's only ever used in the OE builds, and it builds fine natively in-situ. Aug 03 07:34:30 how long does it take to build? Aug 03 07:34:39 seconds. Aug 03 07:34:48 it's just scripts I think. Aug 03 07:34:53 oh, right.. Aug 03 07:34:58 then we can leave it.. Aug 03 07:35:38 still should need python-crypt Aug 03 07:40:26 DaKa2: nod Aug 03 07:40:42 DaKa2: can you fix the python-crypt? Aug 03 07:40:44 should I fix that? Aug 03 07:40:45 ok Aug 03 07:40:53 night all Aug 03 07:40:59 nite Aug 03 07:41:16 we're very close now with openslug-native ... Aug 03 07:41:48 yup Aug 03 09:40:07 make setup on the slug takes 9 minutes, but thats mostly dependant on download speed Aug 03 09:40:52 lets see what make openslug takes with the current set of packages Aug 03 09:42:32 hm.. 3 minutes.. but.. I must have done something wrong.. Aug 03 09:44:56 oh fun.. now I get to run make update :) Aug 03 10:03:56 and.. its still going.. so.. rm monotone/*.db && make setup would be faster Aug 03 10:16:44 touch optware/.configured Aug 03 10:16:44 repvik@maimedslug:~/nslu$ Aug 03 10:17:53 hehe 26 minutes to make update with 8 revs i Aug 03 10:17:54 n Aug 03 10:18:23 NAiL: I get an error on do_patch for iperf, anything you know anything about? Aug 03 10:18:46 yes Aug 03 10:19:02 It's because the PATCH_DIRS thing doesn't work Aug 03 10:19:29 ah Aug 03 10:19:48 so I'll just add ./ in the patch Aug 03 10:20:12 hmm Aug 03 10:20:22 might work Aug 03 10:20:31 jbowler came with an alternate solution Aug 03 10:22:54 what the... what is S="${WORKDIR}/iperf-${PV}/cfg" ? Aug 03 10:22:57 why cfg? Aug 03 10:23:22 because that's where the configure file is Aug 03 10:23:34 cd src instead? Aug 03 10:23:35 and do_configure runs ${S}/configure Aug 03 10:23:42 oh.. Aug 03 10:24:11 there was a bb that worked around that Aug 03 10:25:33 oh well.. Aug 03 10:25:34 --- ./../lib/SocketAddr.cpp.orig 2004-10-05 12:10:06.763746800 +0200 Aug 03 10:25:34 +++ ./../lib/SocketAddr.cpp 2004-10-05 12:10:21.079570464 +0200 Aug 03 10:25:37 that works Aug 03 10:25:40 ok :) Aug 03 10:25:43 if you remove pnum=0 Aug 03 10:25:47 yeah Aug 03 10:27:12 NOTE: package iperf-1.7.0-r0: task do_compile: started Aug 03 10:27:48 NOTE: Handling BitBake files: / (0002/0055) [ 3 %] Aug 03 10:27:52 :) Aug 03 10:27:58 thats the meta dir Aug 03 10:28:01 I got it to compile as well, but then it bombed out Aug 03 10:28:42 Im thinking about adding the individual bb-files and not the dir meta to bbfiles Aug 03 10:29:05 Packaged contents of iperf into /home/slug/openslug/tmp/deploy/ipk/iperf_1.7.0-r0_armeb.ipk Aug 03 10:29:20 DaKa2: Neat Aug 03 10:29:55 but.. we have a problem with the native stuff: NOTE: Couldn't find shared library provider for libstdc++.so.6 Aug 03 10:30:18 and all the other libs it uses, so the dependencies are not fixed automatically Aug 03 10:30:26 That is kinda beyond me Aug 03 10:30:40 NOTE: package gzip-1.3.5-r1: task do_configure: started Aug 03 10:30:47 neat Aug 03 10:30:55 How much does it actually build? Aug 03 10:31:13 not much Aug 03 10:31:16 gzip and vlan Aug 03 10:31:24 the two simplest packages Aug 03 10:31:30 hehe Aug 03 10:31:33 tried apache? ;) Aug 03 10:31:38 nope.. Aug 03 10:31:44 is that in nonworking? Aug 03 10:31:52 su -c "ipkg install tmp/deploy/ipk/iperf_1.7.0-r0_armeb.ipk" Aug 03 10:32:10 haha, brilliant Aug 03 10:32:57 and, it works Aug 03 10:33:23 should I fix the package and patch, and move back from nonworking? Aug 03 10:33:29 yes Aug 03 10:33:38 it'll probably work cross now too Aug 03 10:37:37 can't you make monotone remove directories? Aug 03 10:38:27 monotone can move dirs and files Aug 03 10:38:33 you can drop them, but you'll have to rmdir the empty dir yourself Aug 03 10:38:39 although monotone doesn't really know about dirs Aug 03 10:38:42 ah.. Aug 03 10:39:35 it atleast builds the package cross.. Aug 03 10:39:45 havn't tested the package yet. Aug 03 10:46:15 NOTE: package gzip-1.3.5-r1: task do_configure: started Aug 03 10:46:15 ERROR: function do_configure failed Aug 03 10:46:27 oh? Aug 03 10:46:35 NOTE: Executing autoreconf --verbose --install --force --exclude=autopoint -I /home/repvik/nslu/openslug/tmp/work/gzip-1.3.5-r1/gzip-1.3.5/m4/ -I /home/repvik/nslu/openslug/tmp/staging/armeb-linux/share/aclocal-1.9 -I /home/repvik/nslu/openslug/tmp/staging/armeb-linux/share/aclocal Aug 03 10:46:41 /home/repvik/nslu/openslug/tmp/work/gzip-1.3.5-r1/temp/run.do_configure.2439: 552: autoreconf: not found Aug 03 10:46:44 FATAL: autoreconf execution failed. Aug 03 10:46:59 do you have the entire openslug-native installed? Aug 03 10:49:00 autoconf is in openslug-native I hope.. Aug 03 10:49:00 root@maimedslug:/etc/samba# ipkg install openslug-native Aug 03 10:49:00 Package openslug-native (1.0-r9) installed in root is up to date. Aug 03 10:49:31 hehe Aug 03 10:49:33 wait a minute Aug 03 10:53:09 NAiL: updated openslug-native commited Aug 03 11:06:52 a complete make openslug takes 7 minutes now Aug 03 11:07:03 nice Aug 03 11:13:27 DaKa2, on a slug? Aug 03 11:13:35 yes :D Aug 03 11:13:50 this is native builds of packages? Aug 03 11:14:13 yes :P Aug 03 11:14:39 no need for razzys...I'm just trying to understand what it is. Aug 03 11:19:15 razzys? Aug 03 11:50:09 NOTE: build 200508032030: completed Aug 03 11:50:09 Build statistics: Aug 03 11:50:11 Attempted builds: 4 Aug 03 11:50:18 hehe, yes, I saw :) Aug 03 13:58:53 DaKa2: NOTE: package openssl-0.9.7e-r1: task do_package: started Aug 03 13:59:10 nice Aug 03 14:09:39 ERROR: Build of apache failed Aug 03 14:09:41 bleh Aug 03 14:09:47 Missing sources :P Aug 03 14:10:00 hehe, simple fix atleast Aug 03 14:15:24 not exactly Aug 03 14:15:33 .47 doesn't appear to be around anylonger Aug 03 14:15:38 yes Aug 03 14:15:41 where? Aug 03 14:15:42 http://archive.apache.org/dist/httpd/ Aug 03 14:16:18 NOTE: package apache-2.0.47-r0: task do_configure: started Aug 03 14:16:36 hey! Aug 03 14:16:40 you cheat! :P Aug 03 14:16:47 :) Aug 03 14:19:36 NOTE: package apache-2.0.47-r0: task do_configure: started Aug 03 14:20:02 ugh... Aug 03 14:20:09 this thing gets lots of errors.. Aug 03 14:20:14 not surprised Aug 03 14:20:41 maybe a more recent apache contains less errors? Aug 03 14:20:47 * koen very naive Aug 03 14:20:58 buildconf: checking installation... Aug 03 14:20:58 head: `-1' option is obsolete; use `-n 1' Aug 03 14:20:58 Try `head --help' for more information. Aug 03 14:20:58 buildconf: autoconf not found. Aug 03 14:20:58 You need autoconf version 2.13 or newer installed Aug 03 14:21:00 to build Apache from CVS. Aug 03 14:21:20 I hate when that happens Aug 03 14:21:25 koen: the current patches don't apply to the most recent apache Aug 03 14:21:33 scummvm uses the same obsolote -1 options Aug 03 14:21:37 autoconf -V Aug 03 14:21:37 autoconf (GNU Autoconf) 2.59 Aug 03 14:22:12 it's not the autoconf that freaks out I think. It's the "head -1", which should be "head -n 1" Aug 03 14:22:18 sed -i s:-1:-n\ 1:g * Aug 03 14:22:24 ah Aug 03 14:22:55 koen: Isn't that the same as 'sed s/-1/-n 1/g', only less readable? ;) Aug 03 14:23:20 well.. I'll leave this to you NAiL, have to get home.. its getting late Aug 03 14:23:27 NAiL: sort of Aug 03 14:23:30 I'm going to bed! ;) Aug 03 14:23:37 NAiL: except mine doesn't break on the space ;) Aug 03 14:23:40 hehe :) Aug 03 14:23:43 well.. Aug 03 14:24:07 Oh well, look at it tomorrow Aug 03 14:24:10 why bed, when you can stay up all night and fix things, for example apache? :) Aug 03 14:24:14 heh Aug 03 14:24:19 I slept three hours last night Aug 03 14:24:29 because I was hacking openssh/dropbear/etc Aug 03 14:24:35 heh, I got almost 4 hours Aug 03 14:24:52 DaKa2, you europe? Aug 03 14:24:59 and I was going to make up for that tonight Aug 03 14:25:08 ka6sox-office: yep.. 23:25 here now Aug 03 14:25:13 k Aug 03 14:25:14 and Im at work Aug 03 14:25:14 23:24 here :P Aug 03 14:25:30 NAiL is a littl behind Aug 03 14:25:32 I should start getting home, I just got stuck here Aug 03 14:25:32 :) Aug 03 14:25:38 yeah Aug 03 14:25:44 it's due to sleep deprivation ;) Aug 03 14:25:51 yep Aug 03 14:26:07 I'm finally on GMT-7 again...after being GMT+2 for a month. Aug 03 14:26:30 finally? whats wrong with +2? :) Aug 03 14:26:38 +2 is great Aug 03 14:26:40 (in the summer) Aug 03 14:27:01 exactly +1 +DST Aug 03 14:27:59 ouch Aug 03 14:27:59 ouch Aug 03 14:28:00 ouch Aug 03 14:28:04 40 revs in Aug 03 14:28:11 my poor slug is going to work forever Aug 03 14:28:34 oh well, 17 commits for one day is not that bad, and a complete native build enviorment finished Aug 03 14:28:38 haha Aug 03 14:28:43 rm monotone/*.db && make setup Aug 03 14:29:06 nah Aug 03 14:29:17 7 revs in took 20something minutes.. Aug 03 14:29:18 Uses up some server's bandwidth Aug 03 14:29:19 ;) Aug 03 14:29:26 make setup takes like 9.. Aug 03 14:29:29 I'm up to 4 revs now Aug 03 14:29:48 depends.. if the current snapshot hasn't got that merge... Aug 03 14:30:02 Then it'll take me much longer to do it thataway Aug 03 14:30:19 yeah, but isn't it updated like, often? Aug 03 14:30:29 not sure Aug 03 14:31:34 oh well Aug 03 14:31:42 Time to fall asleep Aug 03 14:32:00 yeah, time to leave, bbl Aug 03 14:32:02 once per hour Aug 03 14:32:03 nite nite NAiL Aug 03 14:32:11 nite DaKa2 Aug 03 14:33:02 rwhitby-asleep: oh.. well.. then a make update and go to sleep might not be that bad... Aug 03 14:33:17 or just run make update every ten minutes.. Aug 03 14:33:22 heh, yeah Aug 03 14:33:30 I'll have to do a make update when I wake up Aug 03 14:33:33 then go make breakfeast Aug 03 14:33:52 monotone: bytes in | bytes out | certs in | revs in | revs written Aug 03 14:33:52 monotone: 121.9k | 33.8k | 137 | 40 | 4 Aug 03 14:34:06 monotone: bytes in | bytes out | certs in | revs in | revs written Aug 03 14:34:06 monotone: 121.9k | 33.8k | 137 | 40 | 4 Aug 03 14:34:11 hehe Aug 03 14:34:20 DaKa2: if you already have a monotone database; use make update./ Aug 03 14:35:20 dyoung-zzzz: well, on the slug that will probably take forever with more than 50 revs in.. Aug 03 14:35:21 the setup target that downloads a recent monotone db is a feature for first time setup folk, so it doesnt take a day to get up to date. Aug 03 14:35:47 Master Makefile? Aug 03 14:35:49 For native dev, you should pull it from a local server (ie: your primary build box) Aug 03 14:36:15 yeah, I should probably set that up.. Aug 03 14:36:42 is there some nice script for doing that somewhere? Aug 03 14:37:11 I suspect there are several folks getting started now; somewhere to the tune of a gig a day. Aug 03 14:37:17 for JUST that file. Aug 03 14:37:32 ow.. Aug 03 14:37:46 no need script actually. Aug 03 14:38:15 scp buildbox:/home/slug/monotone/nsl2-linux.db -O monotone/nsl2-linux.db Aug 03 14:38:25 will work. Aug 03 14:39:47 hm.. well.. yes.. I was more thinking about making a keystripped compressed one and put that on a server.. Aug 03 14:40:21 DaKa2: http://ewi546.ewi.utwente.nl/OE/OE.db.bz2 Aug 03 14:40:55 koen, we have one too. :-) Aug 03 14:41:15 yeah Aug 03 14:41:30 I'm glad that we have 4 public monotone server Aug 03 14:41:37 with a max difference of 20 minutes Aug 03 14:41:40 servers* Aug 03 14:41:53 DaKa2: I can mail you the script I'm using. if youre doing this with regularity though, you'll want to use a server on your local link. Aug 03 14:41:58 a local scp from cross build box will be faster too than a new download Aug 03 14:41:59 that file is getting rather large. Aug 03 14:42:56 bitbake:/var/www/OE# du ../tmp/cvsexport/cvs.db.bz2 -h Aug 03 14:42:56 638M ../tmp/cvsexport/cvs.db.bz2 Aug 03 14:42:57 dyoung-zzzz: please do (daka@thg.se) Aug 03 14:43:05 the old bk export Aug 03 14:43:32 imagine pulling that in the first time Aug 03 14:45:25 now I really should get home and get some sleep, bbl Aug 03 14:45:48 sent Aug 03 14:45:57 thanks :) Aug 03 14:46:30 here's my script: scp monotone/nslu2-linux.db root@openslug:/home/slug/monotone/ :-) Aug 03 14:48:24 Yup, thats pretty much it. Aug 03 15:34:18 DaKa2: looks like we need diffstat for quilt in openslug-native.bb Aug 03 15:37:30 rwhitby: oh, quilt doesn't repend on diffstat? Aug 03 15:37:55 depend* Aug 03 15:40:28 well, I did an ipkg upgrade, and quilt wasn't installed. Aug 03 15:40:41 so openslug-native is not bringing it in Aug 03 15:41:12 oh.. Aug 03 15:41:45 that's with openslug-native-1.0r7 Aug 03 15:42:19 r10 is in there now.. Aug 03 15:42:46 Hmm - wonder why the feed hasn't been updated with that ... Aug 03 15:43:07 That's even in the stable feed :P Aug 03 15:43:20 (or was that just r9= Aug 03 15:44:15 hm.. I think you added the lastes thing I did, so you should have r10 in the stable feed Aug 03 15:44:41 autobuild broke with requiring a merge Aug 03 15:44:51 I've recopied the db and restarted Aug 03 15:45:14 that was probably my fault for making a makefile change in the autobuild account yesterday Aug 03 16:13:28 rwhitby: couldn't conf/topdir.conf get created by the master makefile on make setup somehow? Aug 03 16:15:31 thats the only annoyance I have left with the native stuff, except having to run bash.. Aug 03 16:16:04 couldn't bash u-a itself as /bin/sh? or would that be wrong? Aug 03 16:16:12 oh well.. /me needs sleep Aug 03 16:16:13 bbl Aug 03 17:09:54 DaKa2: unstable feeds should be synced by now Aug 03 17:26:53 hm, when my pine was rejected.. what happens when I sync now? Aug 03 17:27:32 I dont want to re-commit, so to speak Aug 03 17:27:36 it will be rejected at your end too I think Aug 03 17:29:16 I do a sync now, we'll see :) Aug 03 17:29:37 monotone is annoyingly slow :P Aug 03 17:32:37 the setup-env is broken Aug 03 17:32:49 you cant have exits in a file you source Aug 03 17:33:10 well, you can, but it's pointless Aug 03 17:34:54 actually, it's not pointless, cause it stops the vars further down being defined, which stops the build Aug 03 17:35:54 set -e Aug 03 17:36:08 so one is not supposed to do a "source openslug/setup-env" before running bitbake? Aug 03 17:36:30 sorry, I don't get what you're saying? Aug 03 17:36:40 make openslug Aug 03 17:37:02 I just the env set in the shell I'm in Aug 03 17:37:10 not to rebuild openslug :) Aug 03 17:37:41 but I now for some reason miss ./conf/topdir.conf Aug 03 17:38:10 everything has changed - you need to make setup at the top-level Aug 03 17:38:10 which brings me to an "exit 1", which ofcourse logs me off the machine :) Aug 03 17:38:20 hm, again? sheesh Aug 03 17:38:42 don't delete anything, just make setup Aug 03 17:38:49 developing for openslug is like putting up tents in a storm :) Aug 03 17:40:31 didnt make any difference, I still have no topdir.conf Aug 03 17:40:48 setup-env should create it Aug 03 17:41:09 oh, sorry, top-level makefile creates it Aug 03 17:41:14 make openslug Aug 03 17:41:50 mrf.. Aug 03 17:43:00 ah, it's actually openslug/Makefile which creates it Aug 03 17:52:15 oh, my problem was that I sourced the setup-env from wrong directry Aug 03 17:54:02 too bad bash/unix has no equivalent for progdir: Aug 03 17:57:15 kolla: latest changes to the Makefile were to support building openslug packages natively on the slug Aug 03 18:11:54 ok Aug 03 18:13:21 UcSlugC just made it's debut on Tinderbox Aug 03 18:13:55 and autobuilds should be back to normal now Aug 03 18:52:07 rwhitby-away: how about ". ${BASH_SOURCE[0]%/*}/conf/topdir.conf" instead of just . ./conf/topdir.conf" ? :) Aug 03 18:52:55 then we can source it from wherever Aug 03 18:54:07 kolla: yuck. just cd into the directory. Aug 03 18:54:22 oh, hang on, I misread what you wrote Aug 03 18:54:41 It doesn't work because there is no way to know which topdir is required. Aug 03 18:55:04 Somehow the user has to say 'I want a foo env' (foo == openslug, ucslugc, unslung) Aug 03 18:55:44 right - and we do that as part of the setup from the top-level Aug 03 18:55:47 I have a command 'slugenv', which takes the name of the thing, but that isn't sourced - it gives me a new interactive shell. Aug 03 18:56:48 What I'm saying is that for an interactive shell, where I can type 'bitbake', I have to first decide what the conf is - and it is by chosing the correct topdir.conf that this is done. Aug 03 18:56:49 ok, no big deal :) Aug 03 18:57:48 bitbake is weird anyway - first time I've ever seen a build system which 'knows' where it is - you can bitbake from anywhere and you get the same thing. Aug 03 18:58:03 I think that's confusing... Aug 03 19:00:13 well, I'm used to that Aug 03 19:00:53 it's how gentoo works too Aug 03 19:01:49 I'm just annoyed that I have to "cd openslug" before I can "source setup-env", and if I dont, I get kicked out Aug 03 19:02:04 Well, yes, that should be fixed I think. Aug 03 19:02:21 which is what I suggested fixes :) Aug 03 19:02:32 In fact test -r conf/topdir then test -r ${..etc would work. Aug 03 19:02:58 that wont help Aug 03 19:03:08 It still fails? Aug 03 19:03:34 I dont get it.. test -r only tests if a file is readable Aug 03 19:03:44 and exists ofcourse Aug 03 19:04:07 if I dont "cd openslug", the file "./conf/topdir.conf" will never exist Aug 03 19:04:46 however, the file ${BASH_SOURCE[0]%/*}/conf/topdir.conf always exists Aug 03 19:04:47 Well - try your fix, then try "cd openslug; . setup-env" Aug 03 19:05:02 Eh... no it doesn't Aug 03 19:05:26 does here Aug 03 19:06:08 No it doesn't Aug 03 19:06:20 Try what I said Aug 03 19:06:34 what did you say? Aug 03 19:06:55 ah Aug 03 19:07:03 I see what you mean Aug 03 19:07:22 . ./setup-env Aug 03 19:07:35 well.. Aug 03 19:07:37 What I suggested works, but it means "cd openslug; . ../unslung/setup-env" does something weird. Aug 03 19:07:59 I guess the tests could be reversed, also the pattern can probably be improved, or use dirname. Aug 03 19:11:18 Why did they add all this crazy stuff to bash and not add ${@D}? Aug 03 19:12:48 indeed Aug 03 19:13:04 Anyway, bash runes are moot because this has to work on an NSLU2 as well, and there it needs to work with ash. Aug 03 19:13:30 right Aug 03 19:13:35 oh well.. Aug 03 19:14:04 Must be some way of finding the source file name from ash... Aug 03 19:15:04 Hum, it's on file descriptor 11 - ls -l /proc/$$/fd Aug 03 19:17:33 . ${BASH_SOURCE[0]/setup-env/conf\/topdir.conf} Aug 03 19:17:36 :P Aug 03 19:20:25 source the argument replacing setup-env with conf/topdir.conf Aug 03 19:20:31 Syntax error: Bad substitution (from ash) Aug 03 19:20:35 if it's a choice between putting something like that in the Makefile and just asking the developer to "cd" into the right directory first, then I will lean towards the latter Aug 03 19:20:59 Yes, I think so too. With the exit 1 people don't get it wrong more than once Aug 03 19:21:04 if we change the exit 1, then kolla won't get kicked. Aug 03 19:21:28 ash has no BASH_SOURCE ofcourse Aug 03 19:21:28 will changing it to return 1 work ? Aug 03 19:21:42 No Aug 03 19:21:57 goto ? ;-) Aug 03 19:22:01 change it to if test -r conf/topdir.conf; then doit; true; else false; fi Aug 03 19:22:19 cool Aug 03 19:22:31 Then change the Makefile line to . ./setup-env && bitbake Aug 03 19:22:36 That'd stop my putty-session from closing if I'm in the wrong dir Aug 03 19:22:47 which confused me "a bit" Aug 03 19:23:05 NAiL: especially if it closes so fast that you dont see what's happening Aug 03 19:23:08 In fact I suspect any failure code in that setup-env will immediately stop the command - isn't it executed -e ? Aug 03 19:23:38 no Aug 03 19:23:41 kolla: of course it did ;) Aug 03 19:24:17 jbowler-away: if it was, then I shouldnt have been kicked out, since the . ./conf/topdir.conf should have stopped the script :) Aug 03 19:24:33 uless ofcourse, bash treats . and "source" differently Aug 03 19:25:19 anyways.. it waaaaaay to late Aug 03 19:25:22 Ok, make doesn't seem to set -e, but it's pretty easy to set -e; . ./setup-env; bitbake Aug 03 19:25:57 That works. Aug 03 19:26:08 mm Aug 03 19:26:12 time for bed Aug 03 19:26:47 3 hours of sleep, heh Aug 03 19:27:09 at least I'll sleep on the train to stockholm tomorrow - that's fo sure Aug 03 19:29:01 night Aug 03 19:29:55 jbowler-away: did you see that UcSlugC is now fully autobuilding, with mirrored feed updates and in the tinderbox ? Aug 03 19:36:40 back later tonight **** ENDING LOGGING AT Wed Aug 03 23:59:56 2005