**** BEGIN LOGGING AT Sun Apr 16 09:59:56 2006 Apr 16 10:00:01 koen: you want the bitbake -nv ipkg-utils output pastebinned, or is there are single line or two to look at? Apr 16 10:00:07 zecke: g'day Apr 16 10:00:19 rwhitby: pastebin Apr 16 10:02:21 koen: http://pastebin.ca/49492 Apr 16 10:02:57 RP: The cache is not handling removed files ;) Apr 16 10:03:19 RP: I just deleted the task-b-union..bb file and now the cache is broken ;) Apr 16 10:04:05 RP: http://oe.pastebin.com/662825 <- but don't consdier this as an attempt to clone you Apr 16 10:06:33 hey zecke Apr 16 10:06:54 koen: any idea's what we need to update ? Apr 16 10:07:22 or is a tmp-blowaway in order .. Apr 16 10:07:27 rwhitby: this looks like the bitbake heisenbug we have been seeing Apr 16 10:07:43 removing /tmp should fix it Apr 16 10:08:07 ok, we're overdue for a tmp-blowaway, so that's what I will do. Apr 16 10:08:25 RP: could you give a short description (or link) what the multiple framebuffer approach is about? Apr 16 10:11:12 hail zecke Apr 16 10:13:48 cya in a while Apr 16 10:28:23 03frederic 07org.oe.oz354fam083 * r9286cbfa... 10/packages/xmame/ (6 files in 2 dirs): xmame: fix URIs, add 0.100 and xmame-tiny Apr 16 10:28:27 03frederic 07org.oe.dev * rf92f4782... 10/packages/xmame/ (6 files in 2 dirs): xmame: fix URIs, add 0.100 and xmame-tiny Apr 16 10:35:38 03frederic 07org.oe.oz354fam083 * rd467095f... 10/packages/vlc/ (vlc-gpe-0.8.4/pda-interface.patch vlc-gpe_0.8.4.bb): vlc-gpe: add 0.8.4 with PDA interface patch Apr 16 10:35:45 03frederic 07org.oe.dev * re6dd8f24... 10/packages/vlc/ (vlc-gpe-0.8.4/pda-interface.patch vlc-gpe_0.8.4.bb): vlc-gpe: add 0.8.4, with PDA interface patch Apr 16 11:20:22 oh, heh, no wonder bitbake keeps wanting to build this 2.6 kernel. I had KERNEL = "kernel26" in local.conf. Apr 16 11:20:23 doh Apr 16 11:20:26 * pb_ removes that line Apr 16 12:17:59 koen: To the 'Summary screen': We miss firing an Event somewhere Apr 16 12:19:06 the kernelabiversion error also reared its ugly head again Apr 16 12:21:07 koen: well, not bitbake's fault at all ;) Apr 16 12:26:53 03koen 07org.oe.dev * r80830d28... 10/packages/nogravity/nogravity_2.0.bb: Apr 16 12:26:53 nogravity: add nogravity, a space age shooter Apr 16 12:26:53 * this game insists on linking against libvorbis, so it will be quite slow on ARM Apr 16 12:28:25 ~lart libvorbis Apr 16 12:28:25 * ibot farts in libvorbis's general direction Apr 16 12:28:50 koen: you removed gtkmathview from abiwords depends right? Apr 16 12:29:02 CoreDump|home: yes Apr 16 12:29:06 ok Apr 16 12:29:13 CoreDump|home: I don't get why it only built before I checked it in Apr 16 12:29:41 I have one bad ass merge on my hand wth Apr 16 12:33:18 RP: are you around? Apr 16 12:34:09 bloody hell Apr 16 12:34:27 did someone push an ancient ancient altboot or something Apr 16 12:35:17 try with --lca Apr 16 12:36:40 * CoreDump|home tries Apr 16 12:37:11 ahh that worked =) Apr 16 12:39:42 you'll gonna love 0.26's mark-and-merge Apr 16 12:44:37 03coredump 07org.oe.dev * rcfbb957e... 10/packages/altboot/ (6 files in 5 dirs): Apr 16 12:44:37 altboot: - Fix image_reconf to propperly mount /home.flash Apr 16 12:44:37 - Add another bunch of missing redirects for Collie Apr 16 12:44:37 - Work around error message of playbeep.sh on first-boot Apr 16 12:44:37 - Re-enable propper failsafe fallthroughs in init.altboot Apr 16 12:45:13 CoreDump|home, ever considered adding altbooting into chrooted image with no dedicated device? Apr 16 12:46:06 like on /? Apr 16 12:46:35 well, I have, for instance, /dev-image/ on my hdd and altboot into this chroot Apr 16 12:47:16 ah, nope, never occured to me as my rootfs on Akita is kinda small for this heh Apr 16 12:47:27 mount -o bind / /dev-image/mnt/root/ && mount -o bind /lib/modules/2.6.16/ /dev-image/lib/modules/2.6.16/ && mount -o remount,rw / -n && exec /usr/sbin/chroot /dev-image/ $REAL_INIT "$INIT_RUNLEVEL" Apr 16 12:47:46 Yes, It's not even perfectly "legal", but damn convenient :) Apr 16 12:49:07 well, you MicroDrive guys are kinda uniq heh Apr 16 12:49:32 You can blame it on us, it's sharp's guilt. Apr 16 12:49:45 can't Apr 16 12:50:19 What I don't like is that my opie barely survives any reboot... Apr 16 12:50:28 resume, I mean Apr 16 12:50:45 ;) Apr 16 12:52:37 and another holiday tomorrow! Apr 16 12:56:49 ~gettingstarted Apr 16 12:56:50 methinks gettingstarted is http://oe.handhelds.org/cgi-bin/moin.cgi/GettingStarted Apr 16 12:57:36 monotone.vanille.de <- down Apr 16 12:59:21 nm, it's back Apr 16 13:00:42 even sysrq magic doesn't work after suspend/resume... Apr 16 13:00:51 ssh does, though. Apr 16 13:01:28 so your keyboard is dead Apr 16 13:01:40 it's not after I restart opie in ssh Apr 16 13:01:56 suspend / rsume in VT works? Apr 16 13:01:59 something's wrong with vts though Apr 16 13:02:07 Was just going to check. Apr 16 13:02:37 The problem is that is doesn't happen on EVERY suspend... Apr 16 13:03:08 right now suspend on vt worked... Apr 16 13:03:29 and again it did.. Apr 16 13:03:31 .dev image? Apr 16 13:03:35 both Apr 16 13:03:49 it could be the opie kbd patch... Apr 16 13:03:50 now -- .dev, but the same happens on .oz354x Apr 16 13:04:02 you mean -r30 or something? Apr 16 13:04:05 RP: ouch ;) Apr 16 13:04:10 dunno what's it called Apr 16 13:04:19 It was there before, I was hoping it would fix it, but it didn't. Apr 16 13:04:23 ahh ok Apr 16 13:04:26 RP: task-boostrap-union PROVIDES task-bootstrap Apr 16 13:04:40 moreover, sometimes, after it happens I can't chvt 1 from ssh Apr 16 13:04:42 I never had problems with opie resume from SD or NFS Apr 16 13:04:53 it just sits at chvt until I ^C it... Apr 16 13:05:17 not sure where to look for the specifics of the problem now... Apr 16 13:05:20 03coredump 07org.oe.oz354x * rb8a11c7f... 10/packages/altboot/ (altboot_1.0.6-rc3.bb altboot_1.0.6-rc4.bb): altboot: Add 1.0.6-rc4 Apr 16 13:05:25 03coredump 07org.oe.dev * rb2aabdd1... 10/packages/altboot/ (altboot_1.0.6-rc3.bb altboot_1.0.6-rc4.bb): altboot: Add 1.0.6-rc4 Apr 16 13:05:25 RP: and this somehow makes BitBake think that it needs unionfs etc.. for task-bootstrap Apr 16 13:06:08 CoreDump|home, do you know right out of your head how can I chvt before doing suspend? Apr 16 13:06:13 automatically that is Apr 16 13:06:35 /etc/apm ? Apr 16 13:06:36 poli: /etc/apm* place a scipt in there Apr 16 13:06:43 Aha, thank you Apr 16 13:06:46 np Apr 16 13:08:03 I'm going to hit the road this tuesday and I don't fancy the idea of being unable to suspend for the whole month -- I'm afraid it will drain the battery before the month is over :) Apr 16 13:08:45 CoreDump|home: monotone should get restarted within 5 seconds on vanille and ewi Apr 16 13:08:52 neat Apr 16 13:08:55 CoreDump|home: if not, something more serious is going on Apr 16 13:09:43 gcc4 for handheld doesn't build... I have a fix, but I'm not sure if it's the right one.... Will, file it to bugzilla, anyway... Apr 16 13:10:19 s/l,/l/ Apr 16 13:11:35 I butchered gettingstarted a bit Apr 16 13:11:41 it should be a bit clearer now Apr 16 13:12:12 task-bootstrap-1.0unionroot is that a valid name? Apr 16 13:12:38 hey zecke Apr 16 13:12:39 lol Apr 16 13:12:46 PN = "task-bootstrap" Apr 16 13:12:46 PV = "1.0unionroot" Apr 16 13:12:47 PR = "r1" Apr 16 13:12:57 no wonder bitbake chokes majorily... Apr 16 13:14:01 I'm looking in the textbook now - cya Apr 16 13:14:47 looking in the textbook? Apr 16 13:19:56 polyonymous: does it fail in do_package? Apr 16 13:20:15 umm.. don't remember the name, the one that installs Apr 16 13:20:27 likely do_package Apr 16 13:20:56 do_install Apr 16 13:21:18 Actually, MY illiterate fix was: Apr 16 13:21:23 - sed -i -e '/^*cross_compile:$/ { n; s/1/0/; }' ${D}${libdir}/gcc/${TARGET_SYS}/${BINV}/specs Apr 16 13:21:23 + sed -i -e '/^*cross_compile:$/ { n; s/1/0/; }' ${D}${libdir}/gcc/${TARGET_SYS}/${BINV}/specs &>/dev/null || true Apr 16 13:21:28 and Apr 16 13:21:30 - rm ${D}${libdir}/libg2c.la Apr 16 13:21:30 + rm ${D}${libdir}/libg2c.la || true Apr 16 13:21:40 so, just ignore certain errors for inexistant files. Apr 16 13:21:48 As I don't know if the exist sometimes. Apr 16 13:22:04 The resulting package works fine. Apr 16 13:22:07 forme. Apr 16 13:26:47 I'm seeing the same problem over here Apr 16 13:28:28 koen, if you're good with gcc-package internals maybe you know why specs aren't there and if they should be... the same for .la file, but this problem is more likely to be safe to ignore Apr 16 13:28:44 Two '|| true's fixed it for me... Apr 16 13:41:16 same here Apr 16 13:42:19 Then I don't think I should file this bug now. Apr 16 13:42:36 why not? Apr 16 13:42:58 Ah, I thought you'll fix it. Apr 16 13:43:04 Sure, if you want me to I will. Apr 16 13:43:32 the spec file is supposed to do something I guess Apr 16 13:43:40 but lets see what the experts say about it Apr 16 13:43:49 Yes, but since it works without:) Okay, I'll file it. Apr 16 13:50:39 filed it. Apr 16 13:50:47 thanks Apr 16 13:50:59 np Apr 16 13:53:53 hey mithro Apr 16 13:53:55 ehm Apr 16 13:53:57 hey mickeyl Apr 16 13:54:38 RP: what can we do about the RDEPENDS thing? I hesitate to just return "" or None if the file is not found because this would shadow a potential problem. Can't we save RDEPENDS in unexpanded form? Apr 16 13:54:40 hi koen Apr 16 13:56:59 email? Apr 16 13:57:43 mickeyl: tcl/tk doesn't build with eabi, could you have a look at that when you have some time? Apr 16 13:57:57 koen: can't reproduce that. it builds fine here Apr 16 13:58:21 iirc this has nothing to do with eabi, on some systems the configuration file shows an error Apr 16 13:58:28 but not on mien Apr 16 13:58:29 ah Apr 16 13:58:31 mine, even Apr 16 13:58:37 * koen has 'some system' Apr 16 13:59:28 tried removing some of the config file? Apr 16 14:00:11 I've rebuilt from scratch a few times Apr 16 14:01:20 no i mean removing the offending portions of the configure script Apr 16 14:01:59 * koen still doesn't trust the new bitbake Apr 16 14:02:03 it is just too fast Apr 16 14:02:16 the parsing errors are annoying a bit :) Apr 16 14:02:48 * mickeyl touch /local/pkg/oe/c7x0/tmp/staging/c7x0-linux/kernel/kernel-abiversion to make it shut up Apr 16 14:02:51 mickeyl: well, module-base.bbclass Apr 16 14:03:10 mickeyl: feel free to make it use a 'default' value e.g. UNKNOWN Apr 16 14:03:18 I hesitate to just return "" or None if the file is not found because this would shadow a potential problem. Can't we save RDEPENDS in unexpanded form? Apr 16 14:03:35 mickeyl, email? Apr 16 14:03:50 mickeyl: this would require a dependency 'scan' Apr 16 14:04:00 mickeyl: this RDEPENDS is calling a function Apr 16 14:04:11 mickeyl: this function could use data we don't know... Apr 16 14:04:12 Pendalar: eeks. still didn't read the botmails. sorry, doing that atm. Apr 16 14:04:51 hmm Apr 16 14:04:56 ibot is very quet Apr 16 14:04:59 quiet even Apr 16 14:05:06 ibot: you there? Apr 16 14:05:08 Of course I am! Apr 16 14:05:14 ibot: botmail read Apr 16 14:05:22 hmm. no botmail Apr 16 14:05:49 Pendalar: no botmail for me. could you resend your msg via email? Apr 16 14:05:55 Pendalar: mickeyl@handhelds.rog Apr 16 14:05:57 .org even Apr 16 14:07:19 * koen bitbake python Apr 16 14:09:58 hrm Apr 16 14:10:21 obviously I suck at removing the right portion Apr 16 14:11:15 is this shell? Apr 16 14:11:47 configure should be shell code Apr 16 14:12:23 which bash do you have? Apr 16 14:12:49 GNU bash, version 3.1.14(1)-release (i486-pc-linux-gnu) Apr 16 14:12:57 hmm 3.0 here Apr 16 14:13:29 * mickeyl rebuilds on newer machien Apr 16 14:14:00 aha! Apr 16 14:14:11 | checking system version (for dynamic loading)... /local/pkg/oe/c7x0/tmp/work/armv5te-linux/tcl-8.4.11-r2/tcl8.4.11/unix/configure: line 7624: syntax error near unexpected token `)' Apr 16 14:14:11 | /local/pkg/oe/c7x0/tmp/work/armv5te-linux/tcl-8.4.11-r2/tcl8.4.11/unix/configure: line 7624: ` OSF*)' Apr 16 14:14:11 | FATAL: oe_runconf failed Apr 16 14:14:15 bash 3.1 then Apr 16 14:15:13 http://wiki.linuxfromscratch.org/blfs/ticket/1720 Apr 16 14:17:01 * koen taps foot Apr 16 14:19:30 * koen tries google cache Apr 16 14:20:12 NOTE: package tcl-8.4.11-r2: task do_configure: completed Apr 16 14:20:15 ok, i have a patch for it Apr 16 14:20:22 need to do the same w/ tk Apr 16 14:20:24 hold on.... Apr 16 14:23:04 NOTE: package tk-8.4.11-r2: task do_configure: completed Apr 16 14:23:11 ah, cool Apr 16 14:23:20 it was about the oz.pdai.org mirror and my lack of oz.org access to update in a unison/rsync fashion Apr 16 14:23:38 and lack of anyone willing to discuss it with me (or with sufficent access) Apr 16 14:24:26 what kind of access do you needß Apr 16 14:24:27 ? Apr 16 14:24:53 well before I had a sf.net account with oz access Apr 16 14:25:10 and I ran unison/rsync between the boxes in a 1 way fashion Apr 16 14:25:13 for increm. updates Apr 16 14:27:01 can you get a new sf account? Apr 16 14:27:15 i guess I can provide access to the sf.net oz project Apr 16 14:27:29 I suppose so Apr 16 14:27:40 or use my old one Apr 16 14:28:19 mmmk... /usr/lib/mozilla-1.7.11/mozilla-bin: error while loading shared libraries: /usr/lib/libgtk-x11-2.0.so.0: invalid ELF header Apr 16 14:28:24 that can't be good Apr 16 14:28:36 what's your account name @ sf.net ? Apr 16 14:29:23 I assume pendalar@ .. checking Apr 16 14:29:46 koen: pushing the fix.... Apr 16 14:30:17 mickeyl: thanks Apr 16 14:30:45 np Apr 16 14:31:12 i feel kind of responsible since people only cook tcltk because of python :)) Apr 16 14:33:33 heh some monkey changed root pass already this week Apr 16 14:33:36 * Pendalar sighs Apr 16 14:34:49 03mickeyl 07org.oe.dev * rdcff2c4c... 10/packages/tcltk/ (4 files in 3 dirs): Apr 16 14:34:49 tcltk: apply patch to fix a bogus line in the configure scripts Apr 16 14:34:49 bash 3.1 has a more strict syntax checking than 3.0, this is why it bailed out only on some boxes Apr 16 14:35:34 koen: let me know if it works now Apr 16 14:35:53 mickeyl: I think we can modify the ciabot script to sync the mirror servers as well Apr 16 14:36:02 that would save some cpu and bandwidth Apr 16 14:36:07 happy easter to all :) Apr 16 14:36:10 hey gremlin[it] Apr 16 14:36:43 hi koen, mickeyl :) Apr 16 14:38:53 OSError: [Errno 2] No such file or directory: '/home/koen/OE/monotone/org.openembedded.dev/packages/altboot/altboot_1.0.6-rc3.bb' Apr 16 14:38:56 hrm Apr 16 14:39:02 bitbake can't handle deleted files Apr 16 14:40:17 it could. I'm afraid that's a symptom of the new cache Apr 16 14:41:34 koen: have you ordered Letux already? Apr 16 14:41:42 no Apr 16 14:41:48 koen: Deleted Files: Yes I have seen that already... Apr 16 14:41:54 I'm waiting for RP to turn up Apr 16 14:41:58 mickeyl: | 2 out of 2 hunks ignored Apr 16 14:41:58 | Patch fix-configure.patch does not apply (enforce with -f) Apr 16 14:43:31 * koen fixes Apr 16 14:46:28 eek Apr 16 14:46:30 sorry Apr 16 14:46:42 the fix is on its way Apr 16 14:46:46 pnum=2 Apr 16 14:46:59 NOTE: package tk-8.4.11: completed Apr 16 14:47:06 excellent Apr 16 14:48:36 03koen 07org.oe.dev * r4588bfe8... 10/packages/tcltk/ (tcl_8.4.11.bb tk_8.4.11.bb): tcl/tk: add pnum=2 to make the latest patch apply Apr 16 14:48:42 w00t, first boot via g_ether Apr 16 14:49:01 should we start sending out mails about the upgrade to monotone 0.27 to give our ISV time to adapt? Apr 16 14:49:28 ISVs* Apr 16 14:51:41 pretty good idea, would be good to communicate our plan Apr 16 14:52:01 * koen doesn't know 'our plan' Apr 16 14:52:07 heh Apr 16 14:52:22 and I intend to stay out of discussions and just handle the actual changeover Apr 16 14:52:55 if it was me, our plan is to migrate the month when 0.27 comes out and to do the flag thing then. Apr 16 14:53:03 njs: what is the roadmap for monotone 0.27? Apr 16 14:53:14 mickeyl: is a 0.27 planned anyway? Apr 16 14:53:40 mickeyl: that would be problematic for ISVs with branches Apr 16 14:54:18 I'm sure 0.27 is planned Apr 16 14:54:19 they won't close down the project :)) Apr 16 14:54:19 mickeyl: http://frances.vorpus.org/~monotone-cov/net.venge.monotone/#roster.cc Apr 16 14:54:32 koen: ok, what else should we do? Apr 16 14:55:09 mickeyl: maybe it is is monotone 1.0? Apr 16 14:55:23 zecke|study: heh Apr 16 14:55:28 mickeyl: offer them to rosterify their branches by a trusted party (i.e a OE core dev) Apr 16 14:56:01 mickeyl: hmm I think I will get the monotone baseball shirt Apr 16 14:57:37 koen: sounds good to me. i doubt that they would give in to that though. I would think they don't want/can't disclose the things they have in branches. otherwise they would've commited to mainline.... Apr 16 14:57:55 I'd be happy to sign NDAs Apr 16 15:00:08 hmm Apr 16 15:00:36 but as I said, I intend to stay out of discussions Apr 16 15:00:50 well, there's not much to discuss IMO Apr 16 15:00:58 either we switch or we stay @ 0.25 for ever Apr 16 15:01:00 so Apr 16 15:01:14 0.26+ is that much better that it is justified in our opinion Apr 16 15:02:38 and we sincerely hope that this will be last db switch Apr 16 15:02:41 for a long long time Apr 16 15:03:39 together with our svn repository mirror i don't see why people should be angry about that Apr 16 15:03:54 i can't stand all the mt is too slow complaints any longer anyway Apr 16 15:04:07 mickeyl: hehe Apr 16 15:04:09 Pendalar: found your username? Apr 16 15:04:15 mickeyl: alias that Apr 16 15:04:54 the box has already been updated this week (root pass) so I'm trying to break in =-/ Apr 16 15:05:03 my backup admin isn't answering his cell Apr 16 15:05:13 i see. Apr 16 15:05:29 I should get it in a sec Apr 16 15:06:22 on top of that one of my x11 libs broke for some reason Apr 16 15:06:34 so I can't load any new x11 soft ^_^ Apr 16 15:10:26 mickeyl: I wonder when the results of the linuxdevices survey gets published Apr 16 15:12:40 zecke|study: yeah, they're taking long this year.... *cough* Apr 16 15:13:09 mickeyl: I wonder what is up to Harald Welte... Apr 16 15:13:20 mickeyl: wow having soldered two lines to STUART... Apr 16 15:13:35 no wonder netfilter is still experimental :} Apr 16 15:14:14 yes it's pendalar@shell.sf.net Apr 16 15:17:36 koen: have you ever considered using HAL on a handheld? Apr 16 15:21:42 zecke|study: I *am* using hal on a handheld :) Apr 16 15:22:39 koen: which one? Apr 16 15:23:02 the newest in OE Apr 16 15:23:16 koen: can I easily control power status of devices with HAL? Apr 16 15:24:14 zecke|study: no idea Apr 16 15:24:16 do we still need /etc/init.d/hotplug? shouldn't udev handle that? Apr 16 15:24:23 I was busy adding gnome-powermanager to OE Apr 16 15:25:42 zecke|study: it should just be a matter of writing hal metadata to handle such devices Apr 16 15:29:44 well that fixed.. part of my problem Apr 16 15:30:02 koen: what version of poppler do you have staged? Apr 16 15:30:45 XorA: 0.5.1 Apr 16 15:30:48 okay: Did anyone look at task-bootstrap-union as well? Apr 16 15:31:04 zecke|study: I deleted it Apr 16 15:31:11 koen: odd, as abiword is incompatible with that version, how you get it to compile :-) Apr 16 15:31:17 it is playing evil tricks with the PN which leads to the wrong RDEPENDS (unionfs) Apr 16 15:31:21 XorA: I'm asking that myself Apr 16 15:31:22 koen: why didn't you correct the PN and PV and PR usage? Apr 16 15:31:36 zecke|study: because I'm lazy Apr 16 15:31:49 koen: where did you delete it? Apr 16 15:31:50 koen: I will do a patch tomorrow as the fix is one gBool :-) Apr 16 15:36:27 mickeyl, I sent you an email while I was fixing problems Apr 16 15:36:56 mickeyl, just email me back with instructions on what all to mirror etc and the full path to the parts (or I can poke around later) Apr 16 15:37:11 using unison =-) Apr 16 15:37:13 bbl Apr 16 15:38:55 hi Apr 16 15:40:52 hey Bernardo Apr 16 15:52:38 hmm... in perl .bb file there is an SRC_URI for file://config.sh-arm-linux-gnueabi which doesn't exist and in do_configure it does cp config.sh-arm-linux config.sh-arm-linux-gnueabi... Apr 16 15:53:18 polyonymous: monotone update Apr 16 15:53:56 polyonymous: http://ewi546.ewi.utwente.nl/tmp/viewmtn/revision.psp?id=d93900aa884b52d778b478749eeafadb6bef3a21 Apr 16 15:54:38 I see. Thanks. Apr 16 15:55:40 hm... Well, For what I can tell by .diff I don't need it anymore ;-))) But thanks for letting me know my fix was right. Apr 16 16:04:56 bleh Apr 16 16:05:03 this bitbake bug is getting annoying Apr 16 16:05:17 failed build -> go straight to do_rootfs Apr 16 16:06:37 yes, that makes catching errors harder. Apr 16 16:13:12 oh dear Apr 16 16:13:15 bitbake -b broke Apr 16 16:13:31 zecke|study: bitbake doesn't pickup PR anymore... Apr 16 16:13:50 zecke|study: nevermind Apr 16 16:13:56 * koen had a typo Apr 16 16:16:50 automake depends on a bunch of perl modules, but it doesn't depend on them from ipkg point of view... Apr 16 16:16:59 What's the point of using jffs2 guys N Apr 16 16:17:01 ? Apr 16 16:17:17 Ecco: there are no real alternatives to it Apr 16 16:19:59 03koen 07org.oe.dev * ref3ab2ca... 10/packages/gnutls/gnutls_1.3.5.bb: gnutls: ship psktool in 1.3.5 Apr 16 16:21:11 It's not possible to use ext3 on a flash drive ? Apr 16 16:21:28 ext3 needs a block device Apr 16 16:21:37 flash is an mtd device Apr 16 16:23:19 oh, I see Apr 16 16:23:31 I thought flash was handleds as a block device Apr 16 16:23:49 What would you recommend then if I choose to use a CF card as a root fs ? Apr 16 16:24:01 (And by the way, which one would you recommend : CF or SD ?) Apr 16 16:24:05 re Apr 16 16:24:27 Ecco: ext2 with noatime would be the least bad thing Apr 16 16:25:01 ok, thanks for the tip Apr 16 16:25:09 Why not journaling the FS ? Apr 16 16:25:59 that would mean extra writes to the CF/SD card Apr 16 16:26:54 * RP returns Apr 16 16:27:10 * RP will probably regret this judging by scrollback :) Apr 16 16:27:10 RP,zecke|study : wb Apr 16 16:28:02 mickeyl: Still here? Apr 16 16:28:29 RP: hey ;) Apr 16 16:28:29 RP: not too much issues Apr 16 16:28:29 RP: if two files (fn) have the same PN how should we handle it? Apr 16 16:29:10 zecke|study: It becomes a multiple providers issue Apr 16 16:29:25 zecke|study: This is task-bootstrap? Apr 16 16:29:48 zecke|study: Given the unionfs one a negative default preference? Apr 16 16:30:43 RP: worse Apr 16 16:31:09 RP: unionfs just sets PN to task-bootstrap and PROVIDES task-bootstrap Apr 16 16:31:18 zecke|study: Ah, as they have the same name? :-/ Apr 16 16:31:31 zecke|study: I think that should be illegal Apr 16 16:31:41 RP: right Apr 16 16:32:00 zecke|study: It can have the same RPROVIDES but not PN Apr 16 16:32:01 RP: removing that PN makes it a true multiple provider Apr 16 16:32:25 RP: PN, PV, PR should be removed and PROVIDES should be renamed to RPROVIDES Apr 16 16:32:36 zecke|study: Yes Apr 16 16:32:43 RP: or both (PROVIDES and RPROVIDES) Apr 16 16:32:59 RP: e.g. building opie-image could build a unionfs or none unionfs image Apr 16 16:33:34 zecke|study: Probably PROVIDES and RPROVIDES Apr 16 16:33:45 Its PN that will cause the real problems Apr 16 16:34:05 Was this checked in recently? Apr 16 16:37:14 RP: no Apr 16 16:37:34 RP: it is due the change to bin/bitbake Apr 16 16:38:10 zecke|study: So it was just luck that caused task-bootstrap to build before? Apr 16 16:38:13 RP: we take PN and put it in a dict, add add the deps (add_deps) Apr 16 16:38:32 RP: yes, we should have seen multiple providers ;) Apr 16 16:39:11 I guess there should be a check for conflicting PNs in there... Apr 16 16:39:40 Although that won't work as PV and PR will be different and you'd not be able to tell the difference... Apr 16 16:39:43 I just can Apr 16 16:39:58 well it is okay to have conflicting PNs? Apr 16 16:40:13 Two different packages with the same PN is wrong Apr 16 16:40:19 we should fix task-bootstrap first and then consider something to solve such issues more easily Apr 16 16:40:51 Having two different programs both called apache in OE would be equally wrong... Apr 16 16:41:19 RP: it is wrong from a OE point of view Apr 16 16:42:09 zecke|study: Well, bitbake presumably handles them as it should - finds all similar PN's, then picks the highest PV Apr 16 16:42:30 RP: no, it puts all RDEPENDS Apr 16 16:42:38 RP: into the same provider Apr 16 16:42:44 RP: task-bootstrap gets selected Apr 16 16:42:56 RP: but RDEPENDS of task-bootstrap-unionfs get build ;) Apr 16 16:43:12 zecke|study: Ah, that is totally broken :-( Apr 16 16:43:36 The RDEPENDS need to be hashed by fn then? Apr 16 16:43:38 RP: the add_deps is the issue... Apr 16 16:43:50 RP: yes we need one level of indirection more Apr 16 16:44:11 unionfs is wrong regardles Apr 16 16:44:34 zecke|study: Yes, but bitbake is also wrong Apr 16 16:44:47 RP: right, and delete a .bb file and see what happens Apr 16 16:45:21 zecke|study: The deletion business doesn't worry me - we just need to ignore the cache if it doesn't exist :) Apr 16 16:45:46 RP: most things don't worry me ;) Apr 16 16:45:57 zecke|study: I had an insane idea about deepcopy btw Apr 16 16:46:07 RP: I lave insanity Apr 16 16:46:10 love even Apr 16 16:46:25 zecke|study: How about replacing the copies with grafts (joins) ? Apr 16 16:46:51 RP: joins? Apr 16 16:47:20 RP: cooker.bb_cache.loadDataFull Apr 16 16:47:38 RP: why is it returning a 'Cached' version? Apr 16 16:49:16 zecke|study: loaddatafull doesn't return a cached version? Apr 16 16:49:41 RP: will it ever return one? Apr 16 16:49:49 zecke|study: No Apr 16 16:49:55 RP: currently it returns a tuple ;) Apr 16 16:50:41 zecke|study: We can remove that second option - it was from when I had both caches active Apr 16 16:51:15 zecke|study: The idea of joins is instead of doing a copy, we just mark another dict as the parent. If a lookup fails for the copy, we check the parent Apr 16 16:51:38 RP: ah, right Apr 16 16:51:45 If the parent is modified, it knows the children so it can place the unmodified data into each child, then update itself Apr 16 16:51:54 RP: yes currently the dict is not meant to be accurate (it can contain wrong data Apr 16 16:52:43 It would take COW to a new level... Apr 16 16:53:00 but I didn't know the dicts could contain wrong data :-/ Apr 16 16:53:02 RP: hehe, this is our COW implementation already ;) Apr 16 16:53:18 RP: you could have removed an override Apr 16 16:53:25 RP: but that is legal... Apr 16 16:53:43 zecke|study: Then why do we need deepcopy? Apr 16 16:53:52 lazyness Apr 16 16:54:15 zecke|study: The dicts do parent/child tracking? Apr 16 16:56:01 RP: python ones? not that I know Apr 16 16:56:51 zecke|study: So how do you avoid copying all the data? Or does python do that for you somwhow? Apr 16 16:57:43 RP: we d['_parent'] -> going to our parent Apr 16 16:57:43 RP: basicly we need to implement keys() __getitem__ and __setitem__ Apr 16 16:57:53 anyway back to studyinf Apr 16 16:58:09 RP: doing that joins/COW is just a matter of writing a couple of lines python code Apr 16 16:58:43 zecke|study: ok, thanks for the details. I'll understand eventually :) Apr 16 16:59:08 RP: in these two dicts we just save overrides we have seen Apr 16 16:59:25 RP: we might save too many, but we do not skip one Apr 16 16:59:39 RP: so the simple _parenting is almost working perfectly Apr 16 17:00:12 I'll have a look at the code and hopefully it will all become clear :) Apr 16 17:00:30 cya later Apr 16 17:00:37 zecke|study: cya Apr 16 17:25:48 03coredump 07org.oe.dev * r1f312706... 10/packages/altboot/ (6 files in 6 dirs): altboot: Allow usage of USB network (g_ether) for NFS booting Apr 16 17:30:48 ok, I just committed fixed to bitbake for the task-bootstrap issue and the deleted file handling Apr 16 17:32:17 cool Apr 16 17:33:21 is hh.org owned? Apr 16 17:39:39 I'm off for a coffe break some hours. Apr 16 17:39:48 RP: please add changes to ChangeLog Apr 16 17:39:50 (in bitbake) Apr 16 17:39:59 btw. Apr 16 17:40:04 I vote for bumping version to at least 1.4.0 Apr 16 17:40:08 bbi2h Apr 16 17:40:21 mickey|bbi2h: 1.4 would be stable Apr 16 17:40:31 are we linux? :)) Apr 16 17:40:37 go for 1.5 then Apr 16 17:40:40 the changes deserve it Apr 16 17:40:41 (imo) Apr 16 17:40:51 mickey|bbi2h: are we NetBSD? Apr 16 17:40:54 * CoreDump|home plays with kexec Apr 16 17:41:18 mickey|bbi2h: At some point I'd like to talk to you about that kernel RDEPENDS Apr 16 17:45:59 * RP has just noticed zecke's email to oe@ :) Apr 16 17:46:41 * zecke|study crosses the fingers to have not written o-handed.com Apr 16 17:47:05 hehe :) Apr 16 17:48:23 RP: OT: Do you know how big the pxafb framebuffer is? Apr 16 17:48:37 zecke|study: Offhand, no, sorry Apr 16 17:49:39 RP: how would having multiple framebuffers help with videoaccell? Apr 16 17:49:54 koen: One is an overlay for the other Apr 16 17:50:18 koen: So it gives things like mplayer, overlay capability Apr 16 17:50:33 RP: how would you place the overlay? Apr 16 17:50:40 RP: changing the var screen info? Apr 16 17:51:19 zecke|study: Not sure Apr 16 17:52:50 hmm Apr 16 17:52:57 build task-foo doesn't work anymore Apr 16 17:53:02 (did it work before?) Apr 16 17:54:25 RP: how do I pass kernel options with kexec? kexec -l zImage --append="blahh" doesn't change a thing here Apr 16 17:54:40 CoreDump|home: You'll need a kernel without CMDLINE compiled in Apr 16 17:54:44 ahhh Apr 16 17:54:46 well I'm back to my text book... Apr 16 17:54:54 ~lart self Apr 16 17:54:54 * ibot drops a truckload of VAXen on self Apr 16 17:57:31 "unlike OZ, our users are very active" Apr 16 17:57:42 ?? Apr 16 17:57:47 let me guess: pdaX dweeps? Apr 16 17:57:50 who said that ? Apr 16 17:58:09 Ecco: I think koen can guess :) Apr 16 17:58:15 RP: "Then I guess your users will help you with 2.6 integration, goodbye" Apr 16 17:58:31 :-D Apr 16 17:58:40 what is pdaX? Apr 16 17:58:51 polyonymous: moneysucking evil Apr 16 17:58:56 Thanks :) Apr 16 17:59:14 koen: I will remain patient :) Apr 16 17:59:26 :-D Apr 16 17:59:41 RP: ask for a share of the $500 they get each month Apr 16 17:59:46 Indeed, I quite like the way pdaXrom is wired Apr 16 17:59:59 is pdaX == pdaXrom? Apr 16 18:00:03 yep Apr 16 18:00:07 koen: Where does that come from? users? Apr 16 18:00:13 RP: yes Apr 16 18:00:31 RP: they 'demand' $500/month to keep working on pdaX Apr 16 18:00:42 gee Apr 16 18:00:44 I see. Well, it is one of the best, but I preferred OZ over it. Apr 16 18:01:01 i ported Linux to Palm devices fo free !!! Apr 16 18:01:18 I haven't managed to do much for money. Apr 16 18:01:46 RP: CosmicPenguin had a nice quote for that "I could miss $500 a month to lock some russion dude in a closet to work on OE" Apr 16 18:01:58 I have had a few free and open utils for Win32 and by public demand I added the page for donations. Got about 16 EUR in about 2 years :) Apr 16 18:02:01 :-D Apr 16 18:02:44 polyonymous: hehe, and how much did you pay to paypal and your bank? Apr 16 18:02:48 I hid away from donations for a long time with the Zaurus kernels. It wasn't why I did it... Apr 16 18:03:02 so they get loads of money for work RP did Apr 16 18:03:09 and insult us at the same time Apr 16 18:03:14 Yeah, I mean, let's keep free software free as in beer as well :-) Apr 16 18:03:35 zecke, nothing, I haven't withdrer the money. Apr 16 18:03:35 koen: Apparently we get "patches and stuff" to look forward to Apr 16 18:03:44 stuff is probably the insults :) Apr 16 18:03:55 yeah, 'patches' Apr 16 18:04:10 their track record for patches is not that good Apr 16 18:04:54 RP: why where you so stupid and started ARM kernel hacking? Apr 16 18:05:12 ?? Apr 16 18:05:22 Well, I can't even fight for when people take credit for what I do. My open sources were stolen and it didn't bother me much. Although I did email those guys and they promised to put the credits back. Never had any release since that, though... Apr 16 18:05:32 RP: what where your reasons? Apr 16 18:05:48 zecke: Initally I got annoyed about the jog wheel on my 760 which didn't work properly Apr 16 18:06:09 :-D Apr 16 18:06:10 RP, you should be paid for being annoyed then. Apr 16 18:06:11 Lol Apr 16 18:06:27 zecke: I then tried to understand the 2.4 kernel's keyboard driver. Things then sort of got out of hand :) Apr 16 18:06:41 RP: which thing worked last with the 2.6er kernel? Apr 16 18:06:45 RP: jog wheel? Apr 16 18:06:49 =D Apr 16 18:06:56 zecke: no, sound Apr 16 18:07:12 zecke: The jog wheel worked with the rest of the keyboard Apr 16 18:07:35 speaking of sound... seems it doesn't work on my .dev image... Apr 16 18:07:44 * koen admires his native EABI toolchain Apr 16 18:07:50 OT: when was the last kdepimpi release? Apr 16 18:09:53 zecke: I recall zautrix saying he's been busy, but has got outlook sync almost finished Apr 16 18:13:06 hmm... no, sound works, but not opie's screen taps/key clicks... I can live without those... Apr 16 18:14:22 polyonymous: Its an opie bug - sometimes sound works, sometimes it doesn't Apr 16 18:14:34 correct Apr 16 18:14:42 I see. It worked in my .oz image, so I used it to test, which was the wrong thing to do. Apr 16 18:14:49 not Opie, QtE :} Apr 16 18:15:00 whatever Apr 16 18:15:03 qtopiaaargh sound 'server'? Apr 16 18:15:16 zecke: Its not the kernel which means its not my problem ;-) Apr 16 18:15:56 RP: I updated my patch to apply to 2.6.16 (really no changes) and the probe function isn't being called when I insert the driver.... Apr 16 18:16:17 opie dies on me all the time... Sometimes after this suspend/resume death it dies so hard that it doesn't communicate with qcop either. Which is good thing because I restart it on resume automatically if it's dead :) Apr 16 18:16:23 JustinP: Did you remember to add the section to arch/arm/mach-pxa/spitz.c Apr 16 18:17:59 polyonymous: is it frozen, or dead (could you get a ps?) Apr 16 18:18:08 polyonymous: I think I know why it is dieing Apr 16 18:18:08 zecke, it is in ps Apr 16 18:18:14 * RP -> food Apr 16 18:18:30 RP: yes, the device is in there Apr 16 18:18:40 RP: and I added the device to the list Apr 16 18:18:55 QtE allocates memory, and uses semaphores from within signal handlers Apr 16 18:19:02 And it clings so hard to keyboard that I can't even switch to vt1 with sysrq magic :) Apr 16 18:22:44 hmmm Apr 16 18:22:58 hh.org mailingslists seem to be broken *again* Apr 16 18:23:13 RP: the patch is essentially the same as it was previously Apr 16 18:25:32 koen: hh.org seems down Apr 16 18:26:19 zecke: scap still works Apr 16 18:26:31 (again) Apr 16 18:29:34 I'm practically ready to do a few "mv"s and switch to .dev image as a main one now... Apr 16 18:34:09 Guess I'm getting close to the point where I will have to dig qte/opie sources deeper. Apr 16 18:36:07 RP: http://oe.reversefold.com/sharpsl-rc.patch Apr 16 19:03:39 holy cow, kernel init w/o flash is _real_ fast Apr 16 19:04:22 what is kernel init w/o flash? Apr 16 19:04:39 as in kernel startup with no mtd crap? Apr 16 19:04:44 a kernel thar does not mount flash Apr 16 19:04:50 correct Apr 16 19:04:54 I see. Apr 16 19:05:05 I wonder what should I do with my flash... Apr 16 19:05:16 you should see the init on akita...... Apr 16 19:06:16 I still wonder why no one came up with the idea to put bootloader into flash for MD models... Apr 16 19:06:34 hmm? Apr 16 19:06:59 CoreDump|home, well, basically, to do the same thing we have on the "computers". So that you don't flash kernel on each upgrade. Apr 16 19:07:25 ah, right. typicallysharp Apr 16 19:07:52 yup. Apr 16 19:08:09 It would be great to delegate this job somewhere else. Apr 16 19:08:09 well, we can use kexec Apr 16 19:08:18 Yes, that's another way to accomplish it. Apr 16 19:08:34 But it's twice as much stuff to do at startup. Apr 16 19:09:12 yep Apr 16 19:10:04 koen: hmm, just tried to openn an openoffice doc in abiword, it just churned away until X crashed, it does however save PDF successfully Apr 16 19:12:11 weird Apr 16 19:13:14 XorA|gone: I guess it emulated OOo Apr 16 19:13:23 taking 600GB of ram and 2000% of cpu Apr 16 19:13:39 koen: also a lot of the plugins seem to be ill, they cant respolve a symbol (it looks like a mangled symbol) Apr 16 19:14:35 XorA|gone: you could try mailing tf@o-hand.com, mallum's abiword minion Apr 16 19:15:42 koen: Well when Im bored in work tomorrow I shall investigate some more, then if I get nowhere, Ill send him an email Apr 16 19:16:04 ah, I can here the food is ready beeper Apr 16 19:27:27 koen: tasks are not broken :) Apr 16 19:27:53 koen: they have almost no DEPENDS though Apr 16 19:28:03 koen: and do not carray a build all rdepends flag Apr 16 19:29:10 that would explain it Apr 16 19:29:23 koen: bitbake task-sdl -> fast Apr 16 19:29:29 koen: bitbake meta-sdl -> slow Apr 16 19:29:43 task-sdk-bare is what I'm after Apr 16 19:32:16 wow if we release a new version of BitBake, I oculd get back to the tinderbox Apr 16 19:33:10 yes, sheriff Apr 16 19:34:25 -fPIC -DPIC -o .libs/SDL_QWin.o Apr 16 19:34:25 | SDL_QWin.h:56: error: extra qualification 'SDL_QWin::' on member 'ApplyKeyRotation' Apr 16 19:34:25 nice gcc4.1 issue Apr 16 19:36:06 gcc 4.1, you're living on the edge Apr 16 19:36:37 koen: hmm if I would live in Nurenberg I could just go to the gcc people Apr 16 19:37:43 but you don't live there Apr 16 19:37:51 right ;) Apr 16 19:44:30 morning Apr 16 19:44:48 hey TheMasterMind1 Apr 16 19:45:02 zecke: I just checked my bank account and I'm still not filthy rich Apr 16 19:45:22 koen: give me your pin and tans numbers, and I can check for you Apr 16 19:45:39 koen: you should apply at TT for the Berlin office Apr 16 19:47:06 zecke: I don't have s SL-5000d, so I can't FUD properly Apr 16 19:48:07 hm 5000? Apr 16 19:50:44 bitbake is so fast. it feels so.. unnatural Apr 16 19:51:09 people are running out of excuses to not use OE Apr 16 19:51:27 its true. i should try the new monotone Apr 16 19:51:38 although some people at TT sill claim bitbake still needs gigs of ram Apr 16 19:51:44 its not in portage yet Apr 16 19:52:58 alright how do i use monotone to create diffs so i can prove i deserve push access Apr 16 19:53:13 zecke, mickey|bbi2h: could one of you send a monotone upgrade plan to oe@? Apr 16 19:53:26 TheMasterMind1: monotone diff Apr 16 19:53:36 koen: 1. upgrade monotone 2. get flithy rich Apr 16 19:53:37 TheMasterMind1: or plain monotone diff Apr 16 19:53:39 koen: or another one? Apr 16 19:53:49 alright Apr 16 19:54:15 zecke: a bit more detail detail on 1) :) Apr 16 19:54:47 koen: 1.1 upgrade monotone 1.2 get rich 2. get flithy rich? Apr 16 19:55:11 1) upgrade monotone 2) ?? 3) Profit! Apr 16 19:55:27 JustinP: I had that in mind as well... Apr 16 19:55:42 JustinP: but we just Copy n' Pasted MontaVistas Business Plan... Apr 16 19:56:02 JustinP: get Venture Capital, get more venture Capital, get more Apr 16 19:56:13 obviously this is not the truth Apr 16 19:56:23 just to avoid missunderstanding Apr 16 19:57:39 which part? Apr 16 19:58:16 TheMasterMind1: I have heard the FUD there is a big company behind OE... e.g. the dark side... Apr 16 19:58:35 TheMasterMind1: I just want to avoid misunderstandings for the forum people of us... Apr 16 19:59:13 gotcha Apr 16 19:59:30 what is mtn? mtsh wants me to run $ mtn pull venge.net net.venge.monotone.contrib.mtsh Apr 16 19:59:39 forums are 1337 lolz0rs OMG Apr 16 19:59:53 TheMasterMind1: the monotone binary got rem Apr 16 19:59:57 renamed to 'mtn' Apr 16 20:00:06 koen: I think one should avoid Venture Capital... Apr 16 20:00:41 ah. is .26 a drop in replacement for .25? Apr 16 20:00:55 TheMasterMind1: it isn't Apr 16 20:01:09 TheMasterMind1: different network protocol, port and storage format Apr 16 20:01:17 hence the dreading to upgrade Apr 16 20:01:21 oh fun Apr 16 20:01:25 can i still use it to create diffs atleast? Apr 16 20:01:31 TheMasterMind1: different port, but IANA assigned one now Apr 16 20:02:45 * JustinP apologizes, the dungeon collapsed Apr 16 20:03:05 someone sent me a msg before I crashed but I didn't see it Apr 16 20:04:52 JustinP: everything has to be about you doesn't it? Apr 16 20:06:24 I just noticed that someone sent me a message before it crashed.... Apr 16 20:07:12 not that i can see Apr 16 20:16:07 TheMasterMind1: how's your gumstix coming along? Apr 16 20:18:02 local dbs and working copies will be broken Apr 16 20:18:02 meh Apr 16 20:19:32 koen: making progress, i've been all over the place. patched the kernel to get mmc hotplug working, patched jikes/classpath to get java working, played with unionfs to get the latest version of that working and wrote some scripts to integrate that with mmc hotplug. working on udev now, the lfs bootscripts has a nice udev script along with a bunch of other ones so i'm adding those to OE Apr 16 20:21:13 sounds good Apr 16 20:21:46 TheMasterMind1: your best bet is to attach diffs to bugzilla, with one bug per package Apr 16 20:22:40 yea. i'm just dreading doing it, so much extra work Apr 16 20:25:35 TheMasterMind1: perhaps you should have been doing it as you went along... Apr 16 20:26:04 perhaps. Apr 16 20:26:39 RP: did you get any patches from the pdaX people yet? Apr 16 20:31:59 koen: What do you think? :) Apr 16 20:32:18 aaaaaadddddeddddd overlay2 support in kernel 2.6! Apr 16 20:32:31 I think no, but please prove me wrong ;) Apr 16 20:32:46 koen: He said he has it working but I've not seen the patch Apr 16 20:32:55 monotone diff doesn't show files that i added.. do i need to monotone add them first? Apr 16 20:32:59 where is that chat channel anyways? Apr 16 20:33:07 JustinP: probe never gets called? Apr 16 20:33:08 TheMasterMind1: yes Apr 16 20:33:14 TheMasterMind1: monotone add --unknown Apr 16 20:33:21 CoreDump|home: The boot time is why I like Spitz btw ;-) Apr 16 20:33:38 no kidding, it's pratically instant heh Apr 16 20:33:58 CoreDump|home: Its also why boot from microdrive rules and why I'll never be using the flash ;-) Apr 16 20:34:31 mtsh has a tab for 'File Comments' on each file, how does that work from commandline. can i create comments for each change i made and have them show up in the monotone diff output Apr 16 20:34:31 oh I can understand that perfectly, flash stinks heh Apr 16 20:35:29 no idea, I never used mtsh Apr 16 20:35:46 RP: yes, probe is never called. init and exit are, though Apr 16 20:36:41 well i'm not using mtsh, just noting that functionality. basically is there a way for me to embed comments in the diff or should i put that in the bug report Apr 16 20:37:50 Bug report I guess Apr 16 20:39:09 mickey|bbi2h: ping Apr 16 20:39:16 pöng Apr 16 20:39:50 JustinP: It looks fine - are you sure the patch applied ok to the kernel? Apr 16 20:39:57 wb mickeyl Apr 16 20:40:01 hi Apr 16 20:40:05 hey mickeyl Apr 16 20:40:17 mickeyl: http://oe.pastebin.com/663712 Apr 16 20:40:24 mickeyl: am I stupid or what is the issue? Apr 16 20:41:07 RP: well....pretty sure... I suppose perhaps I could be running the wrong kernel...I'll try reflashing with my remote kernel agian Apr 16 20:41:19 RP: I know that the patch applied, yes Apr 16 20:41:35 JustinP: It should work... Apr 16 20:41:46 will make sure I flashed the right kernel Apr 16 20:41:57 mickeyl: why am I getting a KeyError: 0 Apr 16 20:42:05 mickeyl: When you have a moment, we could discuss this RDEPENDS issue Apr 16 20:42:24 zecke: heh, that's a funny one Apr 16 20:42:31 lets see.... Apr 16 20:43:05 zecke: when you are using getitem and setitem you should derive from Object Apr 16 20:43:18 try if that changes anything Apr 16 20:43:24 the new class object model is a bit more sane Apr 16 20:44:11 it would be probably better to not use __getitem__/__setitem__ directly on the _dict Apr 16 20:44:32 mickey@gandalf:/tmp$ python foo.py Apr 16 20:44:32 init Apr 16 20:44:32 setitem 10 20 Apr 16 20:44:32 getitem 0 Apr 16 20:44:33 hmm Apr 16 20:44:37 with print output Apr 16 20:44:47 http://oe.pastebin.com/663716 Apr 16 20:46:46 mickeyl: hmm for 'i' in o Apr 16 20:46:52 mickeyl: what do I need to implement? Apr 16 20:47:00 mickeyl: only __getitem__(self, name) right? Apr 16 20:47:00 an iterator Apr 16 20:47:01 hrm, 300 files with diffs Apr 16 20:47:11 btw., your __getitem__ should return something :) Apr 16 20:47:22 mickeyl: isn't None returned anyway? Apr 16 20:47:34 zecke: don't bother with __getitem__, you really want to implement the iterator protocol Apr 16 20:47:36 mickeyl: or return NotEvenNone Apr 16 20:47:49 mickeyl: I want to have [] and 'in' working Apr 16 20:47:58 RP: ah...oops, sorry, looks like I flashed the wrong kernel Apr 16 20:48:05 mickeyl: well, after it choked on me, I decided to use add and get Apr 16 20:48:13 mickeyl: as my semantic is different from a dict anyway Apr 16 20:48:16 RP: I'm getting probe again Apr 16 20:48:20 heh Apr 16 20:48:32 JustinP: ok, that's better :) Apr 16 20:48:44 zecke: __getitem__ seems to be given an integer Apr 16 20:48:54 that's where the 0 comes from Apr 16 20:49:10 mickeyl: that is KeyError(0) isn't it? Apr 16 20:49:30 yeah Apr 16 20:49:45 iirc you need to raise an index exception if you don't find something Apr 16 20:49:59 all nasty things.... which is why prefer the iterator : Apr 16 20:50:00 9 Apr 16 20:50:02 :) Apr 16 20:50:02 mickeyl: I need to look into it at some point... Apr 16 20:50:10 RP: go ahead Apr 16 20:51:02 RP: at do_package time, we need to evaluate the RDEPENDS statement. Apr 16 20:51:08 RP: it's not necessary before Apr 16 20:51:09 RP: same as before, though, I get 0-5 for press and 254-255 for release.... Apr 16 20:51:22 mickeyl: http://oe.pastebin.com/663723 Apr 16 20:51:34 mickeyl: add was __setitem__, get was obviously __getitem__ Apr 16 20:51:58 hmm Apr 16 20:52:00 mickeyl: With the runtime dependency code in bitbake, it has to evaluate RDEPENDS before do_package Apr 16 20:52:11 OT: Developing bitbake is like developing a 16bit, real mode operating system Apr 16 20:52:13 RP: oh that's nasty. right Apr 16 20:52:28 RP: will it get reevaluated at build time ? Apr 16 20:52:32 RP: if not, then we have a problem Apr 16 20:52:35 mickeyl: yes Apr 16 20:52:35 mickeyl: yes Apr 16 20:52:40 oh Apr 16 20:52:43 in that case Apr 16 20:52:56 i could perhaps live with it returning "" or None Apr 16 20:52:59 thing is though... Apr 16 20:53:04 mickeyl: and it is guranteed that at least one kernel has been built Apr 16 20:53:16 in general it really would hide potential problems Apr 16 20:53:35 i guess i can't find out whether i'm in dependency tracking state, right? Apr 16 20:53:41 (from base class) Apr 16 20:53:48 mickeyl: The cache is only used for the depencency calculations and to work out what to build and in what order - each build actually has its data reparsed just before it builds Apr 16 20:54:18 mickeyl: I'm not sure having variables that depend on things in staging is a good idea really Apr 16 20:54:36 mickeyl: In a way, I'd prefer it if this was handled more like the shlibs code Apr 16 20:54:50 RP: ya, something to think about. Apr 16 20:55:09 mickeyl: This about the case of a multithreaded bitbake/OE Apr 16 20:55:15 s/This/Think/ Apr 16 20:55:28 * mickeyl patches base class Apr 16 20:56:28 blah Apr 16 20:56:43 mickeyl: something like a USE flag on stereoids Apr 16 20:56:44 All this sitting in front of computer starts to take it's toll Apr 16 20:56:54 nooooo Apr 16 20:56:58 not USE flags Apr 16 20:57:06 mickeyl: have a graph and dependencies on the nodes.. if a node linux-2.6 exists, create node FOO_FEATURE Apr 16 20:57:07 I think they're banned :) Apr 16 20:57:14 USE flags are only usefull if you aren't going to make binary packages Apr 16 20:57:19 and update the graph until nothing happens Apr 16 20:57:34 this way packages can be configured dependand on others :} Apr 16 20:57:53 We are going to need to consider this at some point... Apr 16 20:57:53 e.g. one could create a 'profile' this way and store it Apr 16 20:58:05 If X is going to be built, configure Y with X enabled Apr 16 20:58:06 versioned dependencies Apr 16 20:58:16 we *need* those Apr 16 20:58:37 - raise bb.build.FuncFailed("can't read from file '%s' (%s)", (filename,reason)) Apr 16 20:58:37 + return "" # can't raise an error because of the new runtime dependency handling Apr 16 20:58:42 should i commit that? Apr 16 20:58:54 what can we do about the method already seen stuff now? Apr 16 20:59:03 surpress it? Apr 16 20:59:18 mickeyl: can't we give a default argument to it? Apr 16 20:59:19 yeah. for once it would be nice to have it as a -D warning Apr 16 20:59:24 mickeyl: to suppres the exception? Apr 16 20:59:48 mickeyl: well, with func failed it will not land in the cache Apr 16 20:59:52 mickeyl: That fix is ok with me for now. Long term, I think it should be more shlibs like - something like if we see a kernel depends, add a version depedency if the app has said it needs one Apr 16 21:00:19 mickeyl: 2 feature req for the shell: build and build -k Apr 16 21:00:27 mickeyl: See debian.bbclass for examples of the insane things you can do to package names/dependencies Apr 16 21:00:30 ah the method already seen things occur only on first parse Apr 16 21:00:37 koen: can you put these to the bugtracker please. Apr 16 21:00:42 koen: that way I remember :) Apr 16 21:01:25 mickeyl: I rewrote debian.bbclass due to similar "we don't know the field until build time" problems Apr 16 21:01:45 RP: ok, i will look into that when I have more time Apr 16 21:02:37 mickeyl: I think we will have to stipulate that certain core fields should be determined at parse time. This kernel one is borderline as bitbake doesn't care about the version field (yet) :) Apr 16 21:02:43 please someone test r444 Apr 16 21:02:43 specially initial parsing Apr 16 21:02:59 RP: i agree Apr 16 21:04:37 mickeyl: done :) Apr 16 21:05:01 good :) Apr 16 21:05:12 ok, I'll commit that bit and then i need to leave again Apr 16 21:05:16 easter days are family days... Apr 16 21:05:19 cya Apr 16 21:05:40 cya mickeyl Apr 16 21:06:08 zecke: still method already seen errors Apr 16 21:06:14 zecke: Somehow, these Error Method seen issues don't raise an error and still get cached... Apr 16 21:06:28 I also still see them... Apr 16 21:06:47 zecke: Should they all see each other? Apr 16 21:07:08 RP: koen : compare the time from 443 to 444 Apr 16 21:07:19 glibc_2.4.bb shouldn't see the contents of glibc_2.2.5.bb ... Apr 16 21:07:34 .incs.... Apr 16 21:07:36 RP: well, we have a global method scope Apr 16 21:07:50 RP: but both files specify a method with the same name Apr 16 21:07:55 zecke: Initial parse is 3m13 vs. 4m+ for r443 Apr 16 21:07:59 koen: the point is, it is not in a .inc Apr 16 21:08:23 03mickeyl 07org.oe.dev * r60b952e9... 10/classes/base.bbclass: Apr 16 21:08:23 base class: make base_read_file not raise an exception if the file to read is not present Apr 16 21:08:23 This is necessary due to the new RDEPENDS handling. I don't feel 100% comfortable with it, Apr 16 21:08:23 but until we rework the complete staging-shouldn't-contain-dependency-information, this is Apr 16 21:08:23 our way. Apr 16 21:08:26 zecke: Should we have a global method scope? Apr 16 21:08:35 RP: that is a different question Apr 16 21:08:47 zecke: Does anything make use of it? Apr 16 21:08:56 RP: form a performance point of view, it makes sense to have one Apr 16 21:09:07 RP: no need to compile the same code over and over again Apr 16 21:09:11 (base.bbclass etc) Apr 16 21:09:14 Have we always had it? Apr 16 21:09:28 yes Apr 16 21:09:40 can someone look at http://www.themastermind1.net/oe/patches/ipkg_offlinedir_handling.patch real quick and make sure i did the diff right Apr 16 21:09:42 interesting Apr 16 21:10:09 RP: yes it was :} Apr 16 21:11:14 TheMasterMind1: The diff looks fine Apr 16 21:11:21 zecke: http://oe.pastebin.com/663764 Apr 16 21:11:40 TheMasterMind1: the IPKGLIBDIR - > /usr/lib is a bit bogus, isn't it? Apr 16 21:12:02 TheMasterMind1: and didn't pigi fix offline rootstuff a while ago? Apr 16 21:12:38 koen: Which version? I remember seeing it but perhaps the version in OE is old? Apr 16 21:13:10 zecke: 2m57 to do a full scan here with a hot disk cache :) Apr 16 21:13:17 "006-03-30 pigi ( pigi@frumar.it) * Version update to 0.99.160 * Fix for #1585 after the bug introduced with the modify for offline root. Apr 16 21:13:20 " Apr 16 21:13:26 from http://handhelds.org/cgi-bin/cvsweb.cgi/familiar/dist/ipkg/C/ChangeLog?rev=1.220&content-type=text/x-cvsweb-markup Apr 16 21:13:30 koen: that wasn't my patch, i just moved it from 154/ into files Apr 16 21:13:39 koen: Is 160 in OE and does it work? Apr 16 21:13:46 koen: i reported that bug, its not fixed in .159 so i put a patch in, and it doesn't work in .169 Apr 16 21:13:48 160 Apr 16 21:13:53 i reopened the bug a couple of days ago Apr 16 21:13:58 TheMasterMind1: I see Apr 16 21:14:19 RP: with some more testing Apr 16 21:14:29 RP: I would be happy to release a new version of bitbake soon Apr 16 21:14:37 i wish i could embed comments Apr 16 21:14:47 zecke: That sounds good to me Apr 16 21:15:01 TheMasterMind1: Embed comments in what? Apr 16 21:15:23 in the diff, like you can with changesets Apr 16 21:15:45 TheMasterMind1: Just add them at the top Apr 16 21:16:14 alright, thats easier than putting them in the bug report Apr 16 21:16:30 zecke: could you bump the version in svn? Apr 16 21:16:46 RP: could you give the cache a version? Apr 16 21:16:52 zecke: We need to add a cache version :) Apr 16 21:16:53 RP: I would be fine with starting with 666 Apr 16 21:17:03 I was just typing that as you said it :) Apr 16 21:17:06 hehe Apr 16 21:17:35 I will bump both versions Apr 16 21:17:47 'night all Apr 16 21:18:00 'night koen|sleep Apr 16 21:19:15 zecke: Do we store the bitbake version anywhere accessible? Apr 16 21:19:41 RP: I would use a separate version for the cache Apr 16 21:19:58 import bb Apr 16 21:20:02 bb.__version__ Apr 16 21:20:07 zecke: I want to compare bitbake version, libbitbake and cache versions Apr 16 21:20:20 zecke: What about __version__ in bin/bitbake ? Apr 16 21:21:48 we are now at 1.3.9 Apr 16 21:21:56 so only stabilisation now :) Apr 16 21:22:24 I'm currently building oz354fam083 with bitbake 1.3.9 Apr 16 21:24:09 zecke: very good Apr 16 21:24:53 pb_: and then it failed, libtool-native couldn't find i686-linux-libtool Apr 16 21:25:06 zecke: very bad Apr 16 21:25:55 pb_: but it is on the todolist for 1.4.0 Apr 16 21:26:57 great Apr 16 21:27:29 hm, drat, I see my gpe-image build failed again while I was away Apr 16 21:27:49 pb_: on libgcrypt? Apr 16 21:27:57 no, xstroke Apr 16 21:28:00 pb_: gnupg.com is so secure, you can't even connect to it anymore Apr 16 21:28:03 the SRC_URI is wrong Apr 16 21:28:06 zecke: heh Apr 16 21:28:08 pb_: oh another secure site Apr 16 21:28:35 right Apr 16 21:29:04 ERROR: function do_stage failed Apr 16 21:29:04 ERROR: log data follows (/home/ich/hacking/embedded/oe/builds/fam83/tmp/work/libtool-native-1.5.10-r3/temp/log.do_stage.16023) Apr 16 21:29:08 | install: Aufruf von stat für „i686-linux-libtool“ nicht möglich: Datei oder Verzeichnis nicht gefunden Apr 16 21:29:11 NOTE: Task failed: /home/ich/hacking/embedded/oe/builds/fam83/tmp/work/libtool-native-1.5.10-r3/temp/log.do_stage.16023 Apr 16 21:29:13 mm, oh well, I guess I will try again in the morning Apr 16 21:29:14 any idea? Apr 16 21:29:35 pb_: xstroke.org vanished Apr 16 21:29:36 that is rather weird Apr 16 21:29:46 zecke: indeed it did Apr 16 21:29:55 the SRC_URI needs to point to a copy of the tarball on some other site Apr 16 21:29:57 pb_: weird because it is german? Apr 16 21:30:07 pb_: do we have that tarball? Apr 16 21:30:25 zecke: well, that too, but I was thinking more of i686-linux-libtool being missing. Apr 16 21:30:34 did it build libtool under some other name? Apr 16 21:30:54 zecke: I guess we do. it hasn't changed for a while, so it ought to be in the familiar 0.8.0 sources (for example) Apr 16 21:31:08 ich@bigboy:~/hacking/embedded/oe/builds/fam83$ ls tmp/work/libtool-native-1.5.10-r3/libtool-1.5.10/libtool Apr 16 21:31:09 likewise openzaurus 3.5.1 or some such Apr 16 21:31:12 tmp/work/libtool-native-1.5.10-r3/libtool-1.5.10/libtool Apr 16 21:32:37 how strange Apr 16 21:33:04 pb_: I could have a package mission... I don't know which one though :} Apr 16 21:33:50 zecke: what is bin_SCRIPTS set to in your libtool-native's Makefile? Apr 16 21:34:26 # The standalone libtool script, and the libtool distributor. Apr 16 21:34:26 bin_SCRIPTS = libtool libtoolize Apr 16 21:34:49 that is indeed no good Apr 16 21:34:52 how about Makefile.am? Apr 16 21:35:10 pb_: the same version with .dev bin_SCRIPTS = $(host_alias)-libtool libtoolize Apr 16 21:35:11 hmm Apr 16 21:35:36 pb_: without $(host-alias) already Apr 16 21:35:56 how can this happen :} Apr 16 21:36:28 sounds like your computer has been sabotaged Apr 16 21:37:04 pb_: I should have not used that 'Linux'... Apr 16 21:37:19 too true Apr 16 21:37:23 * pb_ go to sleep now Apr 16 21:37:24 night night Apr 16 21:37:30 I should have stayed with Windows for Workgroups Apr 16 21:37:59 pb_: sleep well Apr 16 21:38:08 zecke: will do Apr 16 21:40:16 ~lart busybox Apr 16 21:40:17 * ibot hooks into a hydrant and hoses busybox down Apr 16 21:41:12 Hmm - OE's perl-5.8.7 build is failing: Apr 16 21:41:16 NOTE: Unpacking /home/slug/openembedded/packages/perl/perl-5.8.7/config.sh-arm-linux-gnueabi to /home/slug/debianslug/tmp/work/perl-5.8.7-r15/ Apr 16 21:41:16 cp: cannot stat `/home/slug/openembedded/packages/perl/perl-5.8.7/config.sh-arm-linux-gnueabi': No such file or directory Apr 16 21:41:38 rwhitby: out of space? Apr 16 21:41:46 zecke: nope - 26 Gig free Apr 16 21:41:55 I should be enough Apr 16 21:41:58 rwhitby: your OE tree is too old Apr 16 21:42:08 koen|sleep: updated 5 minutes ago Apr 16 21:42:30 http://ewi546.ewi.utwente.nl/tmp/viewmtn/revision.psp?id=d93900aa884b52d778b478749eeafadb6bef3a21 Apr 16 21:42:41 ah, a recent change? Apr 16 21:43:25 2 days ago... Apr 16 21:43:45 Hmm - how often do the syncs from ewi to nslu2-linux.org happen? Apr 16 21:43:56 could someone check the md5sum of libtool's tar gz? Apr 16 21:43:59 rwhitby: in theory every 15 minutes Apr 16 21:44:21 rwhitby: but as I said, you have to do your own syncing, since you're not serving all OE branches Apr 16 21:44:26 e2093a85f6d48f1562c36920087502d6 ? Apr 16 21:44:41 (libtool-1.5.10.tar.gz.md5) Apr 16 21:45:07 hrm Apr 16 21:45:14 it seems someone botched a merge Apr 16 21:46:12 some one else on libtool 1.5.10? Apr 16 21:46:27 zecke, didn't like my md5? ;-) Apr 16 21:46:33 polyonymous: it matched mine Apr 16 21:46:38 ah, sorry about that :) Apr 16 21:46:39 polyonymous: and my bin_SCRIPTS is wrong Apr 16 21:46:44 e2093a85f6d48f1562c36920087502d6 /home/koen/OE/downloads/libtool-1.5.10.tar.gz Apr 16 21:46:45 e2093a85f6d48f1562c36920087502d6 downloads/libtool-1.5.10.tar.gz Apr 16 21:46:52 uhh Apr 16 21:46:59 hmm Apr 16 21:47:00 I think my libtool works okay... Apr 16 21:47:26 pb_: some odd is happening then, I will ping you tomorrow Apr 16 21:47:30 maybe it's patched wrongly? Apr 16 21:47:37 03koen 07org.oe.dev * r42ed9228... 10/packages/perl/perl_5.8.7.bb: perl: someone seems to have botched a merge, so I'm reapplying http://ewi546.ewi.utwente.nl/tmp/viewmtn/revision.psp?id=d93900aa884b52d778b478749eeafadb6bef3a21 Apr 16 21:47:51 koen|sleep: thx Apr 16 21:48:15 * koen|sleep needs to harass njs about stuff like that Apr 16 21:48:32 and go to sleep as well Apr 16 21:48:37 koen|sleep: I beleive that was the nlus to oe merge mickey|bbl had to do Apr 16 21:48:42 manual merges - just say no. Apr 16 21:48:57 just have $DISPLAY set up right Apr 16 21:52:16 okay Apr 16 21:55:53 hrm whats the best way to do this.. include the patch as an attachment, as a URL in the url field or in the bug report text? Apr 16 21:56:18 pb_: oz354fam083 should apply the prefix.patch from .dev on libtool as well Apr 16 21:56:20 bug report text is messy, attachment means two emails are sent for each bug report Apr 16 21:57:38 attachment Apr 16 21:58:23 k Apr 16 22:02:01 * koen|sleep gives sleeping another try Apr 16 22:03:20 The cache now has a version number Apr 16 22:04:41 nice Apr 16 22:04:48 Does anyone have something (maybe a script) which cleans out the previous obsolete versions of stuff in tmp/work and tmp/stamps? Apr 16 22:05:33 rwhitby: hrw, had once Apr 16 22:05:33 rwhitby: Not me but if you find/write one, could you add it to contrib please? :) Apr 16 22:05:55 rwhitby: RP: actually one could automatically remove tmp/work after the build Apr 16 22:06:13 Our nslu2-linux autobuilder runs every 15 minutes, and it has over 100GB in old versions in work and stamps Apr 16 22:06:21 zecke: There is already a bbclass to do that as you build Apr 16 22:06:29 RP: My plan is: release 1.4.0 Apr 16 22:06:35 RP: work on 1.5.0 in trunk Apr 16 22:06:42 RP: make bugfixes in 1.4.1 as well Apr 16 22:07:16 zecke: ok with me. We have a few bugs to fix before 1.4.0 though Apr 16 22:08:02 rwhitby: I guess the problem is you can't automatically say the highest version number is the latest :-/ Apr 16 22:08:36 RP: what numbers do you have in your mind? Apr 16 22:09:00 zecke: The recursive cleaning for a start Apr 16 22:09:05 ~curse busybox Apr 16 22:09:07 May the fleas of a thousand camels infest your most sensitive regions, busybox ! Apr 16 22:09:17 RP: true - perhaps a successful build could remove all other versions than the one which just built Apr 16 22:09:19 rwhitby: you can always do a find -ctime +/-DAYS | xargs rm Apr 16 22:09:22 zecke: and the cached build failure if it still happens Apr 16 22:09:36 RP: how should it happen? Apr 16 22:09:50 only by a broken stamp... or a bög in Cache.py Apr 16 22:10:07 TheMasterMind1: I want to keep the latest built version (no matter how old) - I just want to remove all the other versions. Apr 16 22:10:57 rwhitby: Really you want bitbake to dump out all the packages it would build for targets X, Y, Z, make a script that removes anything but those versions Apr 16 22:11:07 rwhitby: bitbake's dry-run option might be useful Apr 16 22:11:09 damn... Apr 16 22:12:04 RP: Hmm - I'd like to do it as bitbake runs, on a package by package basis. Just in case there is a dependencies problem. The autobuilder needs to be conservative in deleting things :-) Apr 16 22:13:18 RP: there is a bug with my COW Apr 16 22:13:29 * zecke gets the insecticid... Apr 16 22:13:34 rwhitby: You can have a class with event notifications like tinderbox.bbclass Apr 16 22:13:49 rwhitby: Maybe you could abuse them? Apr 16 22:14:20 rwhitby: package X builds successfully, nuke all other versions? Apr 16 22:14:27 yep Apr 16 22:14:46 I would think that could even default to on. Apr 16 22:16:32 rwhitby: It just won't nuke the last copy of things you don't normally build but perhaps that could be a manual cleanup script now and again Apr 16 22:16:48 zecke: What's the COW infected with? Apr 16 22:17:19 RP: yeah, the safest thing is only touch other versions of the package which just built sucessfully. Apr 16 22:17:35 RP: hmm, one bbfile affecting another one Apr 16 22:18:04 task do_build: completed, task do_cleanup: started ... Apr 16 22:20:02 zecke: At least the build failure caching bug doesn't appear to be here anymore (not that I'm sure its possible with the new cache) Apr 16 22:28:09 how fast is usbnet? Apr 16 22:28:45 probably as fast as usb1/usb2 Apr 16 22:28:58 Which is? comparing to 11mbps? Apr 16 22:29:56 no networking technology runs at mas speed Apr 16 22:29:58 max Apr 16 22:30:13 I know, but you know what I mean Apr 16 22:30:15 usb1 is ~10mbps, usb2 is 40-something IIRC Apr 16 22:30:24 Actually, it doesn't matter. Apr 16 22:30:25 easily findable on google Apr 16 22:30:40 It just seemed to me that ipkg update takes too long, but then I found out it was networking problem :) Apr 16 22:30:51 RP: I'm reverting the COW patch now... Apr 16 22:32:43 zecke: :-( Apr 16 22:33:00 RP: somehow it forgets some 'keys' Apr 16 22:33:29 okay good nite Apr 16 22:33:34 I will make it work tomorrow Apr 16 22:35:25 'night zecke Apr 16 22:37:03 hrm how do i do simple if then logic in a bb file Apr 16 22:37:48 TheMasterMind1: python and lambda expressions Apr 16 22:37:54 yuck Apr 16 22:38:03 RP: did you eat yet? http://handhelds.org/~zecke/pics/privat/belly_left.jpg Apr 16 22:38:21 http://handhelds.org/~zecke/pics/privat/belly_right.jpg Apr 16 22:38:40 this was one day before CeBit 2003 or 2004 I don't remember correctly Apr 16 22:38:43 I think it was 2003 Apr 16 22:38:53 *shouder* Apr 16 22:39:40 * rwhitby looses his breakfast Apr 16 22:40:08 ouch/yuck :-/ Apr 16 22:43:33 umm... I just beamed some files from palm to opie's irda thingie and I'm excited -- it seems to be putting files in / ;-) Apr 16 22:44:09 TheMasterMind1: don't know so much on bb syntax but in python it is something like this print "The loop ran %d time%s" % (i, i != 1 and 's' or '') Apr 16 22:44:26 ~curse busybox Apr 16 22:44:28 May you be reincarnated as a Windows XP administrator, busybox ! Apr 16 22:46:58 ah, that would have bee easier. i just did a normal if else like i'm used to Apr 16 22:52:13 TheMasterMind1: well, if else is simpler but if you need a ternary operator, then 'a and b or c' can be useful :) Apr 16 22:55:12 yea that's pretty nifty Apr 16 22:55:15 nicer than ? : Apr 16 23:13:31 * Philippe is away: visual contact - melancholic dreams Apr 16 23:13:45 this is getting tiring Apr 16 23:15:34 (yawning) good night ! Apr 16 23:15:50 night Apr 17 01:21:29 37 emails to the oe mailing list- i'm a regular spammer now Apr 17 01:43:31 Curious Apr 17 01:43:35 How goes the status of the OE book? :) Apr 17 01:49:01 there's an OE book? Apr 17 01:49:54 well, user's manual Apr 17 01:49:59 http://oe.handhelds.org/manual.php Apr 17 01:56:01 ah Apr 17 01:56:03 i don't know Apr 17 01:56:09 developers usually don't like documentation Apr 17 01:58:10 Right, but when I'm just starting to hack at OE with effort Apr 17 01:58:15 And hopelessly lost Apr 17 01:58:17 It kinda sucks ;) Apr 17 02:01:25 The wiki, quite frankly, sucks Apr 17 02:01:49 I got the tree pulled (after hours), and I'm attempting bitbake nano now just for kicks Apr 17 02:04:14 shouldn't have taken you hours Apr 17 02:04:22 are you using the getting started page? Apr 17 02:04:33 grab a copy of a recent db file and use that Apr 17 02:04:37 also use bitbake from svn Apr 17 02:04:58 its about 50 times faster Apr 17 02:05:27 monotone should be better in a couple weeks as well as we transition to .26, which is atleast 8x faster Apr 17 02:06:59 i understand where you're coming from though, getting started is a pain, there's too much involved to get to the point where you can actually start doing stuff. i'm trying to build a vmware player image with an OE environment all setup already to make it easier to get right into it Apr 17 02:07:13 Last I heard, the SVN was a bit slower to update than the monotone (not the transfer speed, the "updated" level) Apr 17 02:07:27 huh? Apr 17 02:07:31 bitbake is in svn Apr 17 02:07:34 OE is in monotone Apr 17 02:07:46 Sorry, I'm not being clear Apr 17 02:08:06 isn't the SVN tree of the OE database slower to update (from the point where a developer pushes an update) than monotone? Apr 17 02:08:12 yes. Apr 17 02:08:16 That's all I meant Apr 17 02:08:22 there are two different things we're talking about here, bitbake and OE Apr 17 02:08:24 bitbake is in SVN Apr 17 02:08:28 right Apr 17 02:08:30 OE is in monotone Apr 17 02:08:33 use bitbake from SVN Apr 17 02:08:34 its faster Apr 17 02:09:12 Yeah, see, a page with tips like this would be *immensely* useful. Apr 17 02:09:16 bitbake is the program used to build packages using the files in OE Apr 17 02:09:21 Yeah, I know. Apr 17 02:09:23 ScytheBlade1: well, its a wiki. add one Apr 17 02:09:31 I don't know enough to do so :) Apr 17 02:09:53 Otherwise I would have, a while ago Apr 17 02:10:17 please? heh. you're starting out so your prespective is a good one to have writing it. put up what you discover and others will come correct and add to it Apr 17 02:10:37 That's a different way of looking at it Apr 17 02:10:41 And a horribly good way at that, too Apr 17 02:11:35 All right Apr 17 02:11:46 using crosstool, I've created a few cross compilers Apr 17 02:11:57 How do I tell bitbake to use that compiler instead? Apr 17 02:13:06 hrm i'm not sure how easy that will be. its best to let bitbake build the cross compiler because it sets it up in a very particular way which makes stuff a lot simpler and organized Apr 17 02:13:42 * ScytheBlade1 makes mental note of that Apr 17 02:13:49 That was a big thing that was throwing me off :P Apr 17 02:14:41 All right. I'm building a distro for my soekris embedded board (http://www.soekris.com). I'm sure you guys have seen them mentioned in here Apr 17 02:14:50 I need to compile all of this for an i486 Apr 17 02:15:05 look at conf/machine/* and conf/distro/* Apr 17 02:15:22 set up a local.conf with TARGET_MACHINE= one of the machines and DISTRO =one of the distros Apr 17 02:15:31 or copy one of the distros to a custom one Apr 17 02:15:36 i'm sure a machine for your board is in there already Apr 17 02:15:38 hey, look at that Apr 17 02:15:55 Hmm, there isn't actually Apr 17 02:16:05 not for machine/, anyways Apr 17 02:16:52 hmm well i've heard of that board and i dont' know where i would have outside of here Apr 17 02:17:15 It's rather generic PC hardware Apr 17 02:17:22 anyway, if its just a i486 its not that hard Apr 17 02:17:24 133MHz AMD Elan proc, 64MB RAM, etc. Apr 17 02:17:40 koen|sleep: the problem is recurring after a build from scratch: Apr 17 02:17:51 ERROR: Nothing provides runtime dependency glibc-gconv Apr 17 02:17:51 NOTE: no buildable providers for bogofilter Apr 17 02:18:23 ScytheBlade1: look at x86.conf in machine, that should work. you can copy that to soekris.conf and edit the variables. look at the other machine files to see what other vars are available to be set Apr 17 02:18:38 Right, I'm already half-way through that Apr 17 02:18:47 there is a build for linux-x86-2.6.11.6 Apr 17 02:18:54 so you can use that as kernel Apr 17 02:19:04 Eh, I hand-roll the kernel Apr 17 02:19:39 well go to packages/linux/ copy linux-x86-2.6.11.6.bb to linux-x86-2.6.versionyouwant.bb Apr 17 02:19:50 create linux-x86-2.6.versionyouwant/defconfig Apr 17 02:19:55 with your .config Apr 17 02:19:59 and it should build it the way you want Apr 17 02:21:06 if you look at that .bb file you can see what it does. takes the version from the filename, thats ${PV}, grabs that kernel source from kernel.org, before running configure it does configure_prepend which copies the defconfig file to .config and runs oldconfig Apr 17 02:21:24 and the rest is standard kernel compiliation functions defined by the 'inherit kernel' Apr 17 02:22:35 * ScytheBlade1 continues hacking at this Apr 17 02:22:56 OLDEST_KERNEL? Apr 17 02:23:04 That the oldest minimum version of the kernel that it'll run? Apr 17 02:23:19 yea Apr 17 02:23:51 so if you created a 2.6.16.bb like i described before Apr 17 02:23:56 set oldest kernel to 2.6.16 Apr 17 02:24:01 right Apr 17 02:24:03 and then set PREFERRED_PROVIDER_virtual/kernel = "linux-x86" Apr 17 02:24:08 it would use that kernel Apr 17 02:24:38 TARGET_ARCH will decide what compiler you need and build a cross compiler for you Apr 17 02:24:54 TARGET_ARCH = "i486" Apr 17 02:24:55 Simple enough Apr 17 02:25:14 yep Apr 17 02:25:41 Is there any good documentation for the distro/ files? Apr 17 02:26:18 look at conf/bitbake.conf Apr 17 02:26:46 those are most of the pertinent variables Apr 17 02:27:09 haha Apr 17 02:27:09 export CC = "${CCACHE}${HOST_PREFIX}gcc ${HOST_CC_ARCH}" Apr 17 02:27:12 Simple enough Apr 17 02:27:50 yea you don't need to do any of those Apr 17 02:27:53 I know Apr 17 02:28:49 look at wrt54oe.conf for example, you probably want to set the three gcc preferred providers, and then three preferred versions for whatever versions of gcc you want it using Apr 17 02:29:33 PREFERRED_PROVIDERS += " virtual/${TARGET_PREFIX}libc-for-gcc:uclibc" if you want uclibc instead of glibc Apr 17 02:30:34 nah, glibc is fine Apr 17 02:31:14 All right Apr 17 02:31:22 I believe I have the distro and machine files setup correctly Apr 17 02:33:02 along with my local.conf Apr 17 02:33:19 What next? ;) Apr 17 02:34:28 anyone want any socket 7 boards or rs422 gear Apr 17 02:34:35 in a few months when I finish sorting and testing Apr 17 02:37:53 hello....i keep getting this error when trying to build different pacakges with bitbake anyone know the fix? Apr 17 02:38:08 ERROR: Exception:exceptions.OSError Message:[Errno 2] No such file or directory: '' Apr 17 02:38:08 ERROR: Printing the environment of the function Apr 17 02:38:08 ERROR: 0136: if target[0] != '/': Apr 17 02:38:08 ERROR: 0137: target = os.path.join(root[len(inst_root):], target) Apr 17 02:38:08 ERROR: 0138: dangling_links[pkg].append(os.path.normpath(target)) Apr 17 02:38:11 thank you Apr 17 02:44:44 i have no idea what socket 7 boards or rs422 gear is Apr 17 02:45:06 tyz: what version of bitbake? Apr 17 02:45:43 looks liek he is missing a usable local.conf Apr 17 02:46:57 wow, what a cryptic message for that Apr 17 02:48:25 not really Apr 17 02:48:36 it say target doesnt exist Apr 17 02:48:42 and doesnt know what to set it to Apr 17 02:49:12 well path to target Apr 17 02:50:55 * emte doesnt think system arch is my forte Apr 17 02:51:06 PCI arch anyway Apr 17 02:59:41 TheMasterMind1: rs422 is an improved serial protocol (as compared to say, rs232) widely used on Apple branded computers Apr 17 03:00:18 i belive rs422 is the usb1 protocol Apr 17 03:01:01 i would have to read next weeks notes to be sure tho Apr 17 03:02:13 i doubt it Apr 17 03:02:15 no Apr 17 03:02:17 have to make a USB Data Aq system Apr 17 03:02:20 * fishhead bites emte Apr 17 03:02:24 you stupid canuck Apr 17 03:02:25 no? Apr 17 03:02:31 rs422 is a serial protocol, Apr 17 03:02:40 usb protocol is usb protocol Apr 17 03:02:41 :P Apr 17 03:02:54 wonder what we are using the 422 chip for then with usb ... Apr 17 03:03:11 you can have usb to rs232 Apr 17 03:03:14 you can have usb to rs422 Apr 17 03:03:17 you can have usb to rs485 Apr 17 03:03:27 so that a pc with usb can talk to serial only devices Apr 17 03:04:06 maybe ... its not interfacing aserial device tho Apr 17 03:04:07 fishhead: i've an adapter made by (Belkin) that interfaces USB <==> (RS232 || RS422) Apr 17 03:04:33 you can think of RS422 as RS232 with more capabilities Apr 17 03:04:38 yep Apr 17 03:04:49 and a different termination package (i.e. connector ends) Apr 17 03:04:50 damnit ... now you made me pull out my course notes Apr 17 03:05:36 TheMasterMind1 hi, it's: BitBake Build Tool Core version 1.3.9, bitbake version 1.3.9 Apr 17 03:07:45 my head still hurts from building a PCI memory modlue all afternoon Apr 17 03:08:42 TheMasterMind1: when is the 0.26+ mtn transition scheduled? Apr 17 03:09:00 can't believe i got yelled at for asking about it on the mailing list Apr 17 03:09:14 lol Apr 17 03:09:21 jnc: i have no idea Apr 17 03:09:30 yeah i get that for sedning things i really shouldnt Apr 17 03:09:56 i need to write an "emte no mail mailinglist" feature Apr 17 03:10:03 does that mean i'll get yelled at for the 37 emails i generated earlier tonight Apr 17 03:10:14 trying to run after and catch electrons is too hard Apr 17 03:11:05 but from what i understand from ~a week ago we are not shifting to monotone 26 we are waiting for monotone 27 Apr 17 03:11:17 to avoid the big nasty bugs Apr 17 03:11:34 i was going off comments in the channel earlier tonight Apr 17 03:11:48 oh? has it changed? Apr 17 03:12:07 hmm Apr 17 03:12:27 * emte seems to have misplaced my embedded systems notes somewhere ... Apr 17 03:12:51 i wish i could misplace quartus Apr 17 03:13:46 zecke, mickey|bbi2h: could one of you send a monotone upgrade plan to oe@? Apr 17 03:13:58 hmm Apr 17 03:14:04 goodnight Apr 17 03:14:04 njs: what is the roadmap for monotone 0.27? Apr 17 03:14:32 guess i lied. they were talking about .27 not .26 Apr 17 03:15:07 for anyone who wants some advice NEVER use Quartus unless its a last resort Apr 17 03:17:31 stupid hunk of junk with "mystery" file and libray linking Apr 17 03:18:37 could you tell me what version of bitbake most of your are using? Apr 17 03:18:56 ~ 3weeks ago Apr 17 03:19:11 before all the major recent changes Apr 17 03:19:34 but your error seems more to me like you do not have a correct enviroment tyz Apr 17 03:21:18 tyz, you have rm'd your whole tmp and can build dropbear ? Apr 17 03:22:11 yeh rmed everything from when i developing before with oe. i'll try again and see if i can do dropbear Apr 17 03:28:50 tyz: newest bitbake from svn works just fine Apr 17 03:31:13 gah Apr 17 03:31:18 does anyone understand autoconf syntax? Apr 17 03:32:28 ok yeh that's the one i'm using actually. Apr 17 03:32:30 satan perhaps Apr 17 03:32:49 AS_IF([test "$uint64_t" = no], Apr 17 03:32:49 dnl Check for endianness Apr 17 03:32:49 [AC_C_BIGENDIAN()], Apr 17 03:32:49 [AC_DEFINE([HAVE_64BIT_TYPES], [1], Apr 17 03:32:49 [Defined when the compiler supports 64 bit integer math])]) Apr 17 03:35:39 what exactly is the logic there? Apr 17 03:36:21 testing for flags i think Apr 17 06:11:07 CoreDump|afk: altboot rc4 is acting different....pretty sure that rc3 didn't allow me to "press a key to start altboot" after a loop image was booted Apr 17 06:11:18 CoreDump|afk: rc4, however, still lets me use altboot Apr 17 06:11:22 (on the image) Apr 17 07:54:32 morning Apr 17 07:57:19 good morning all Apr 17 08:02:12 hey koen Apr 17 08:04:01 night Apr 17 08:04:49 'night TheMasterMind1 Apr 17 08:05:40 hmmm Apr 17 08:06:06 I haven't gotten any mail from oe@ the past day Apr 17 08:06:29 the last one I got was teh EABI mail from pH5 Apr 17 08:06:42 you should have gotten 37 from me Apr 17 08:06:56 I know Apr 17 08:07:02 http://handhelds.org/hypermail/oe/78/date.html Apr 17 08:07:34 ooh sorted by date, i like that Apr 17 08:07:39 anyway, i must sleep. adios Apr 17 08:30:07 ~lart win2k server Apr 17 08:30:08 * ibot stuffs win2k server into a shiny new tin can and vacuum seals it Apr 17 08:31:02 waste of a shiny new tin can if you ask me... Apr 17 08:32:59 heh heh, me servers rebooted last night for essential patches and win2k being win2k you need to hand hold them to come up again in the right order or the network ceases to work Apr 17 08:33:37 sounds fun Apr 17 08:34:56 has anybody gotten mail from oe@ yesterday? Apr 17 08:35:28 and more specifically: the patch queue from TheMasterMind1 Apr 17 08:35:53 The last piece of mail I got was 21 hours ago Apr 17 08:36:14 johnX: ok, then it's not just me Apr 17 08:36:22 ~lart hh.org for being broken *again* Apr 17 08:36:22 * ibot executes killall -KILL hh.org for being broken *again* Apr 17 08:42:17 morning all Apr 17 08:42:47 morning RP Apr 17 08:43:02 koen: got the same depends problem after a build from scratch Apr 17 08:43:34 I building now with make -k and bitbake -k so I can at least progress past the package which fails for no good reason. Apr 17 08:44:29 koen: I added some changes to bitbake last night which should solve the recursive cleaning problem. It should also not be caching build failures anymore Apr 17 08:44:47 koen: If you could confirm that, that'd be great Apr 17 08:48:58 good morning all Apr 17 08:49:49 Due to the monotone-0.26 internal rewrites, it seems that OE has to rebuild its history. Does this mean there will be two OE branches: one with monotone-0.25 and one with monotone-0.26 ? Apr 17 08:54:52 cromain: If/when we switch, it will have to be all at once Apr 17 08:55:17 RP: when do you plan to switch ? Apr 17 09:00:05 RP: I mean, is it a real concern for OE, or have you more important plan for now, like a new bitbake release or... ? Apr 17 09:00:42 cromain: It is a real concern and we will probably wait for 0.27, then switch Apr 17 09:01:05 Waiting for 0.27 is to try and ensure we don't hit any nasty bugs Apr 17 09:01:34 RP: ok, thanks for the info :) Apr 17 09:05:30 good morning Apr 17 09:07:20 RP: I already updated bitbake :) Apr 17 09:20:50 good day Apr 17 09:21:02 Simple question: Apr 17 09:21:19 ipaq + (i)vorbis = works? Apr 17 09:21:21 :) Apr 17 09:21:31 in terms of gstreamer plugins Apr 17 09:26:08 bug #868 is nice Apr 17 09:28:22 RP: did I read correctly that the pdaX people just used the patches from phytec.com fro the overlay? Apr 17 09:35:11 koen: I think so Apr 17 09:36:43 We're starting to see a few interesting targets for OE :) Apr 17 09:37:09 koen: Did the bitbake update solve the issues I mentioned? Apr 17 09:37:34 RP: I'm trying to fail a build, but OE is too good ;) Apr 17 09:37:58 koen: "exit 1" in do_compile Apr 17 09:42:29 what version of bitbake is best to use? newest? Apr 17 09:46:56 rm_you: The latest svn is fastest - currently being tested for stability Apr 17 09:47:15 RP: k... I'm using gentoo's version right now, 1.3.3 Apr 17 09:47:24 RP: will I run into any problems? Apr 17 09:47:38 rm_you: We hope not :) Apr 17 09:48:21 RP: good. new to the whole bitbake thing, wasnt sure if it was one of those things where version *REALLY* matters Apr 17 09:49:14 RP: so when do I get a fully functioning 2.6 for collie!? eh!? :P j/k Apr 17 09:49:34 rm_you: When Sharp open up the SD controller specs? Apr 17 09:49:51 yeah :/ prolly after I buy a c1000 :P Apr 17 09:50:03 when someone besides kergoth gets the documents under NDA? Apr 17 09:50:05 my guess is *right* after. Apr 17 09:51:03 Yeah, not sure what sharp has to loose... they arent selling the OS, just the device, and OZ prolly helps them sell more zaurii than any other marketing... :/ Apr 17 09:51:17 s/OZ/OE/ Apr 17 09:51:55 I doubt they're too bothered about promoting sales of collie anymore. Apr 17 09:52:16 yeah, since they dont actually sell any anymore :) Apr 17 09:52:32 indeed Apr 17 09:54:52 so if I do "bitbake openzaurus-sa", it'll BUILD the kernel, right? how do I set it up so I can tweak at the kernel source before it builds it? Apr 17 09:55:27 rm_you: with -c unpack Apr 17 09:55:37 rm_you: see bitbake manual Apr 17 09:55:42 XorA: k, thanks Apr 17 09:56:00 rm_you: or maybe -c patch depending on what stage you want to tweak the source Apr 17 09:56:06 hi all Apr 17 09:56:16 hi cedric__ Apr 17 09:56:25 hi RP Apr 17 09:56:27 i wish bitbake had a man page... viewing an online manual just feels wierd Apr 17 09:57:53 cedric__: Sometime, you might like to try enabling the LOCOMO LEDs in your defconfig and see if they work on poodle **** ENDING LOGGING AT Mon Apr 17 09:59:56 2006