**** BEGIN LOGGING AT Tue Oct 02 02:59:57 2007 Oct 02 04:29:08 is anyone able to build wrt56g packages ? Oct 02 06:50:57 sweet, o-hand hired pippin Oct 02 06:55:25 pippin? Oct 02 06:57:50 03koen 07org.oe.dev * ra926e7d2... 10/ (5 files in 3 dirs): midori: add 0.0.8 Oct 02 07:08:43 XorA|gone: Øyvind kolas Oct 02 07:09:21 XorA|gone: the dude we helped find the aussie bar in birmingham Oct 02 07:23:39 koen: ah yeah Oct 02 07:24:53 something for the zaurus geeks: http://linuxdevices.com/news/NS6435117311.html Oct 02 07:27:06 Emblaze make mobile phones so Im better a mobile phone :-D Oct 02 07:27:13 s/better/betting/ Oct 02 07:27:24 and sharp used to make mobile phones Oct 02 07:33:00 XorA: you need to update MAINTAINERS ;) Oct 02 07:34:06 oooh free food at OEDEM :-D Oct 02 07:34:21 free food? Oct 02 07:34:43 ah Oct 02 07:34:48 ~hail rschuster Oct 02 07:34:49 * ibot bows down to rschuster and chants, "I'M NOT WORTHY!!" Oct 02 08:14:44 03koen 07org.oe.dev * r2dc0600c... 10/ (1 conf/machine/hx4700.conf): hx4700: kdrive 1.3.0.0 also has Xw100 Oct 02 08:43:43 heh Oct 02 08:43:59 I do an mtn merge and get a conflict on the file I just edited Oct 02 08:45:09 ~hail XorA Oct 02 08:45:11 * ibot bows down to XorA and chants, "I'M NOT WORTHY!!" Oct 02 08:46:16 libgcrypt? Oct 02 08:46:25 yep Oct 02 08:46:35 heh heh, you must be building gail then Oct 02 08:47:00 gnome-keyring Oct 02 08:47:08 gail was solved by bumping atk :) Oct 02 08:47:22 oh yeah, that fix I pused yesterday Oct 02 08:51:57 03xora 07org.oe.dev * ra1cffa0e... 10/ (1 packages/libgcrypt/libgcrypt_1.2.3.bb): Oct 02 08:51:57 libgcrypt/libgcrypt_1.2.3.bb : remove custom do_stage as its not needed and Oct 02 08:51:57 doesnt actually stage enough. Oct 02 08:52:02 03xora 07org.oe.dev * r34d07eb5... 10/ (1 packages/libgcrypt/libgcrypt_1.2.3.bb): libgcrypt/libgcrypt_1.2.3.bb : PR bump for last change Oct 02 08:52:07 03koen 07org.oe.dev * rcccbcdee... 10/ (1 packages/atk/atk_1.19.3.bb): atk: remove def_pref from 1.19.3 Oct 02 09:25:56 "If you're using the latest OpenMoko software (i.e. in the last week or Oct 02 09:25:57 so), you should find it is becoming a lot more usable. There are quite a Oct 02 09:25:57 number of patches to gsmd being applied in OpenEmbedded which means it Oct 02 09:25:57 appears to be a lot more stable, which in turn means making calls is a Oct 02 09:25:58 lot more reliable. We've been using two Neo1973s to call each other Oct 02 09:25:58 multiple times in the last week without any fiddling on the command Oct 02 09:25:58 line." Oct 02 09:32:04 03xora 07org.oe.dev * rece94f3c... 10/ (3 files in 3 dirs): openmoko2/openmoko-dates2_svn.bb : version bump to latest for thomas Oct 02 09:36:19 03koen 07org.oe.dev * r0e0e3784... 10/ (1 packages/gtk+/gtk-2.10.inc): gtk-2.10.inc: add hack for maemo software Oct 02 09:39:54 XorA: I suspect that needs bumping of libmokoui2 as well, to get the fingerscroll fixes r617 talks about Oct 02 09:40:09 koen: working on it Oct 02 09:40:18 great :) Oct 02 09:54:50 the developer activity on webkit is amazing Oct 02 09:55:50 * koen wonders why ARM ltd sponsors a gecko based browser Oct 02 09:58:45 heh heh Oct 02 10:44:19 Henryk: I will test and push your patch in a couple hours Oct 02 10:45:12 hi everyone Oct 02 10:45:15 hi cbrake Oct 02 10:46:10 cbrake_away: okay, thanks Oct 02 10:46:29 I'm currently building gtk-sharp, which is comparatively easy :) Oct 02 11:36:07 um free food Oct 02 12:26:51 gilligan__: hello Oct 02 12:37:30 morning all Oct 02 12:56:47 guys, is there a switch when using redboot to tell OE to not put the kernel into /boot when do_rootfs task is handled? Oct 02 13:18:20 hi all Oct 02 13:18:42 ibot: seen zecke? Oct 02 13:18:45 zecke was last seen on IRC in channel #oe, 1d 40m 9s ago, saying: 'bbl'. Oct 02 13:20:03 ~seen stelios Oct 02 13:20:05 stelios was last seen on IRC in channel #debian, 562d 19h 11m 47s ago, saying: 'abrotman: cheers mate looking into it'. Oct 02 13:20:20 ~seen steliosk Oct 02 13:20:21 steliosk was last seen on IRC in channel #oe, 9d 22h 8m 43s ago, saying: 'yeah'. Oct 02 13:22:36 Henryk: you have done a lot of work on this mono stuff! What are your reasons for creating mono.bbclass versus just putting the code in an inc file? Oct 02 13:23:21 cbrake: so I can use that class for additional mono packages. Oct 02 13:24:08 Henryk: ok, that makes sense Oct 02 13:24:50 right now I have finished my gtk-sharp recipe which basically only contains the standard stuff (PV, PR, SRC_URI) and the FILES_... variables (semiautomatically generated from debian) and then does an inherit mono which will take care of the dependency stuff Oct 02 13:25:08 Henryk: very nice! Oct 02 13:35:44 cbrake: indeed. i just ran a gtk-sharp hello world on the phone Oct 02 13:35:55 Henryk: :-) Oct 02 13:37:00 what's ${S} ? source dir ? Oct 02 13:37:23 yes Oct 02 13:43:22 hi Oct 02 13:44:11 Henryk: Nice, what phone? Oct 02 13:44:50 CM: GTA01Bv4 Oct 02 13:45:08 hi henryk Oct 02 13:45:15 hi woglinde Oct 02 13:45:27 henryk did you work further on phoneme? Oct 02 13:46:02 woglinde: no, had no further pressing need. Oct 02 13:46:19 okay Oct 02 13:46:25 midpath runs better Oct 02 13:46:30 *g* Oct 02 13:46:43 woglinde: I just fixed up the mono recipes as the project at the university (for which I also did the phoneme) is now going mono Oct 02 13:47:00 oh Oct 02 13:47:02 why mono? Oct 02 13:47:17 Crofton: ping Oct 02 13:48:19 sakoman, pong Oct 02 13:48:42 Henryk: your termcap patch, I think termcap-mini should RPROVIDE termcap, then you can just do PREFERRED_PROVIDER_termcap = "termcap-mini" in a DISTRO Oct 02 13:48:54 woglinde: for the new part of the project we need code download, which is an absoute No-Go with J2ME. I was proposing Python, but unfortunately the chair here is at least partly about middleware, so it kind of needed to be a 'proper' middleware Oct 02 13:49:03 PROVIDE I mean Oct 02 13:49:35 and termcap RCONFLICTS with termcap mini, right? Oct 02 13:49:57 koen: he has used the alternates system to swap them Oct 02 13:49:57 Crofton: having some troubles with installing a wifisix.conf in /etc/modprobe.d in the wifistix recipe Oct 02 13:50:23 urg, this I know little about :( Oct 02 13:50:30 XorA: hmm, okay. I was just thinking about packages that might explicitly want termcap (not mini). But then again, I'm a Gentoo guy and this morning koen so wisely mentioned that I must be missing my USE flags Oct 02 13:51:00 Henryk: I think the number of terminal definitions is probably a DISTRO thing Oct 02 13:51:05 henryk hahah os.system(wget Oct 02 13:51:13 sakoman: what would you prefer: several gumstix--image recipes or one image with some var to select the expansion bard drivers and utils to put int? Oct 02 13:51:25 woglinde: pardon? Oct 02 13:51:42 henryk has j2me no option for executing shell comands? Oct 02 13:51:55 koen, I suspect setting vars would be better Oct 02 13:51:55 koen: I favor the latter approach Oct 02 13:52:08 as there seem to be alot of permutations Oct 02 13:52:15 woglinde: no, of course not. and of course there's no way in hell you could execute any code whatsoever Oct 02 13:52:19 koen: that's basically what I'm inplementing in my collection Oct 02 13:52:30 can't spel :-) Oct 02 13:52:37 woglinde: typical J2ME target doesn't even have a file system, let alone a shell Oct 02 13:52:47 henryk hm okay Oct 02 13:52:53 I'll bbiab, walking ocer to school Oct 02 13:53:07 sakoman: I'm leaning towards vars as well, but them craig will have a fit Oct 02 13:53:36 at OEDEM I'll write a pygtk GUI for that Oct 02 13:53:42 (or zenity based) Oct 02 13:53:45 XorA: should I modify the patch and resubmit or will you do the changes? Oct 02 13:53:50 koen: I think craig will understand Oct 02 13:53:55 Henryk: let me ponder it a bit longer Oct 02 13:54:02 ok Oct 02 13:54:27 the model should work for other hardware also Oct 02 13:54:31 poof Oct 02 13:54:49 koen: He has 2 distinct sets of customers: hobby and commercial. And th needs are widely divergent Oct 02 13:55:37 koen: but to survive the support nightmare for the hobby guys he really needs to keep it simple Oct 02 13:57:18 koen: perhaps you can help with my install issue. In the wifistix recipe do_install I added a "install -m 0755 -d ${D}${sysconfdir}/modprobe.d" and then a simial install for the wifistix.conf Oct 02 13:58:22 The conf file doesn't make it into the install image though. It is in the work wifistix image directory though Oct 02 14:03:36 Henryk: your mono recipes built for me -- I have a meeting for a couple hours now, but I'll try to finish reviewing and push this afternoon. Looks great! Oct 02 14:03:40 hmm.. having some problems with TARGET_OS here.. it is empty for some reason. linux.inc checks the value and enabled EABI/OABI support in the kernel accordingly.. but since it is empty... Oct 02 14:03:47 cbrake: cool Oct 02 14:04:15 Henryk: Nice, I'd happily try mono on my GTA01 too :) Oct 02 14:05:22 sakoman, FILES_something += the file Oct 02 14:05:47 I am wondering if we could make a task per gumstix Oct 02 14:06:00 to load the right stuff for each card Oct 02 14:06:10 Crofton: thanks Oct 02 14:06:48 I think :) Oct 02 14:07:51 Crofton: I'm playing with a task-gumstix that takes MACHINE_FEATURE inputs from the machine conf and "does the right thing" Oct 02 14:08:23 cool Oct 02 14:08:40 I am just wondering if it would be easier to make a task for each peripheral Oct 02 14:09:23 Crofton: perhaps. I'm such a newbie I don't really know which is right :-) Oct 02 14:09:57 heh Oct 02 14:10:03 I am not sure either Oct 02 14:10:24 I've done most of my work on very static hw Oct 02 14:10:32 and not for mass consumption Oct 02 14:11:05 Crofton: I'll keep playing with my current approach and see where it leads Oct 02 14:11:51 morning Oct 02 14:13:44 anyone got any ideas on why TARGET_OS could be empty here ? it seems to be set in distro/include/angstrom-uclibc.inc .. but I don't quite get the statements there Oct 02 14:14:20 this is slowly but surely driving me mad Oct 02 14:17:56 Crofton: your suggestion worked :-) Thanks again. Now I get to figure out why OE gnuplot pulls in libstdc++ when buildroot gnuplot doesn't Oct 02 14:18:27 Crofton: at that int I'll have image size roughly the same for OE and buildroot! Oct 02 14:18:37 int = point Oct 02 14:18:51 need more coffee :-) Oct 02 14:19:07 sakoman hm Oct 02 14:19:15 try an ldd on the target Oct 02 14:20:54 sakoman, excellent Oct 02 14:22:06 these guys on the gumstix list speak in some kind of mysterious code Oct 02 14:22:33 woglinde: don't have ldd on my target system :-( Mine is a uclibc build if that matters Oct 02 14:23:12 Crofton: not sure what you mean about code-speak Example? Oct 02 14:23:51 the latest email about the netwifimicroSD card Oct 02 14:24:00 patchs, build root revs Oct 02 14:24:12 netCF-vx Oct 02 14:24:25 * Crofton is a little frustrated :) Oct 02 14:27:12 Ah, yes that is a doozy! Oct 02 14:27:54 Basically, I think I will do something else, before trying to get that board working, hopefully Craig updates buildroot with the stuff Oct 02 14:28:06 sakoman: is ${sysconfdir} in FILES_${PN} ? Oct 02 14:28:15 Yes, a good idea! Oct 02 14:28:47 koen: thanks to Crofton's suggestion it now is :-) And it works too! Oct 02 14:30:02 don't forget is to add to CONFFILES as well Oct 02 14:30:31 CONFFILES_${PN} that is Oct 02 14:31:42 sakoman: if you could summarize your goals and methods in a short document/email, we can discuss it at OEDEM to see how OE can make stuff easier for you Oct 02 14:31:55 koen: something like: CONFFILES_${PN} = "/etc/modprobe.d/wifistix.conf"? Oct 02 14:32:16 CONFFILES_${PN} = "${sysconfdir}/modprobe.d/wifistix.conf Oct 02 14:32:40 koen: of course! I knew that :-) Oct 02 14:33:15 I read it in the manual! Oct 02 14:33:29 Even used it in the recipe! Oct 02 14:34:22 koen: on the goals summary -- sure no problem Oct 02 14:36:31 sakoman: thanks! Oct 02 14:37:04 florian: I think I fixed all hildon stuff to build against a stock gtk+ Oct 02 14:38:01 koen: oh cool, that's good news! Oct 02 14:38:28 hi florian Oct 02 14:39:01 * florian looks around... Oct 02 14:39:08 florian: maemo-mapper fails with unresolved symbols, but I suspect that's is a matter of headers missing IFDEFs Oct 02 14:39:39 I have been offline for some days... moved to a new(er) house. Oct 02 14:39:41 bye till later Oct 02 14:39:52 we are going shopping a little bit Oct 02 14:41:31 florian: I recreated parts of the ubuntu-mobile patches, maybe we can use other patches from them as well Oct 02 14:42:05 koen: that might be interesting Oct 02 14:42:35 hmmm Oct 02 14:42:59 is anyone else failing to build procps with current metadata? Oct 02 14:43:28 Crofton: polyonymous|note's fix from a few days ago worked for me Oct 02 14:43:42 hrm... how do I go to Berlin for OEDEM? Oct 02 14:43:53 there is a post on the ml Oct 02 14:44:20 I suppose I should ask when he updated, these hashes as rev's are not so helpful Oct 02 14:44:43 is there an easy to convert one to a date? Maybe using viewmnt? Oct 02 14:44:55 Crofton: http://monotone.openembedded.org/revision/info/fa850c805eca306aba96aedcbfface81cfd404d8 Oct 02 14:45:15 Crofton: mtn log --no-merges --no-graph packages/procpc --last 5 | more Oct 02 14:45:25 s/procpc/procps/ Oct 02 14:46:07 Crofton: in cleaning up the wifistix recipe you checked in I noticed that it calls make directly rather than using oe_runmake. Was there a reason for this? Oct 02 14:47:09 That is what I got, I need to look over this some more Oct 02 14:47:17 thanks for reminding me Oct 02 14:47:38 sakoman: I tried it with oe_runmake and could never get it to work Oct 02 14:47:46 03pfalcon 07org.oe.dev * rd2f47efb... 10/ (1 conf/machine/htcblueangel.conf): htcblueangel.conf: Remove PREFERRED_VERSION for kernel. Oct 02 14:47:54 03koen 07org.oe.dev * rdd9786d5... 10/ (4 files in 3 dirs): libhildonfm: add 1.9.41 works against a stock gtk+! Oct 02 14:48:02 03koen 07org.oe.dev * r71333c75... 10/ (1 packages/maemo3/libhildonhelp_1.9.1.bb): libhildonhelp: correct DEPENDS Oct 02 14:48:07 03koen 07org.oe.dev * r6d0b0c40... 10/ (1 packages/fontconfig/fontconfig_2.4.1.bb): fontconfig: stage nokia device table header Oct 02 14:48:17 03koen 07org.oe.dev * r8f2107c8... 10/ (1 packages/maemo3/hildon-1_svn.bb): hildon-1: update PV Oct 02 14:48:21 scruggs: Ah, what was the issue? Oct 02 14:48:25 03koen 07org.oe.dev * r9f9ee853... 10/ (1 conf/distro/include/sane-srcrevs.inc): sane-srcrevs: bump hildon-1 Oct 02 14:48:31 koen, looks the the ml problem is acouple of revs b4 the fiux went in Oct 02 14:48:32 03koen 07org.oe.dev * rb739d287... 10/ (1 conf/distro/include/sane-srcrevs.inc): sane-srcrevs: bump openmoko icon themes Oct 02 14:50:03 sakoman: not too sure exactly - it kept pulling in userspace headers instead of kernel ones during the compile, but I'm a bit too new at this to really know why Oct 02 14:50:10 * florian -> home, bbl Oct 02 14:50:47 scruggs: hmm . . . just tried it with oe_runmake and it built just fine! Oct 02 14:51:40 sakoman: strange... but good! Oct 02 14:52:09 sakoman: how is the autoloading stuff coming along? Oct 02 14:52:13 scruggs: I'll send you a copy of what I have -- see if it works for you now Oct 02 14:52:24 sakoman, copy me Oct 02 14:52:25 excellent! Oct 02 14:52:25 scruggs: everything seems to work now Oct 02 14:52:36 Crofton: sure thing Oct 02 14:52:43 I have a meeting at 2 Oct 02 14:52:54 and after I think I will go shopping and hide and work on thesis Oct 02 14:53:04 trying to kill some little stuff b4 then Oct 02 14:53:15 Crofton: I'll send it now Oct 02 14:53:24 still looking for someone who might help me figuring out how TARGET_OS can be empty when building angstrom .. anyone ? Oct 02 14:57:01 koen, I have some autobuilder bugs from things I do not know about libglade for example Oct 02 14:59:00 Crofton: the idea is that people that know about libglade see the bug in bugzilla and fix it :) Oct 02 14:59:06 ok Oct 02 14:59:24 I'm just trying to close autobuilder hiccups Oct 02 14:59:46 that's why I said "We have an autobuilder that files bugs now, it would be nice if people looked at those bug Oct 02 14:59:47 reports." two weeks ago Oct 02 15:02:24 scruggs, Crofton: sent the wifistix package. let me know if you see issues Oct 02 15:03:10 sakoman: thanks - I'll check it out this morning Oct 02 15:03:13 sakoman, what changes did you make? Oct 02 15:03:24 can you send me a mtn diff? Oct 02 15:04:22 sakoman: what daughtercards do you use with your stix for your testing? Oct 02 15:14:48 Crofton: just tried to make it more "OE" : no hard coded paths to etc & lib, used oe_runmake, and then added the wifistix.conf install into /etc/modprobe.d Oct 02 15:15:27 scruggs: just wifistix and my custom hardware Oct 02 15:15:40 I am adding to what I have and will compile check and push later Oct 02 15:15:49 I had wiped tmp again :) Oct 02 15:15:56 Crofton: taking my car in for service right now, I'll send the diff later Oct 02 15:15:59 koen: did LTG monotone just die? Oct 02 15:16:16 don't worry about it Oct 02 15:16:21 I have diffed by eye Oct 02 15:16:30 pretty simple, eh? Oct 02 15:16:33 the changes look good to me Oct 02 15:16:36 yeah Oct 02 15:16:58 I am goign to try and get teh args to oe_runmake into EXTRA_OEMAKE Oct 02 15:17:47 Crofton: sounds good to me. I didn't want to touch that in case it turned into a tar baby :-) Oct 02 15:18:16 Crofton: I think thats what I was trying - oe_runmake with the opts in EXTRA_OEMAKE - mabye it will work for you Oct 02 15:18:26 thats how I thought it should be in the first place Oct 02 15:19:24 is monotone.openembedded.org down? Oct 02 15:20:07 chouimat: yup Oct 02 15:20:46 XorA: fun ... the day I have no paperwork to do :) Oct 02 15:38:24 sigh Oct 02 15:52:39 Crofton: just sent you a small patch to the gum kernel Oct 02 15:56:11 XorA: I restarted it Oct 02 15:56:27 koen: still connection refused Oct 02 15:57:22 and now? Oct 02 15:58:07 ah Oct 02 15:58:12 the ssh tunnel collapsed Oct 02 15:58:36 heh heh Oct 02 15:58:50 working now Oct 02 15:59:15 are there restrictions as to where I can put an oenote command ? I am trying to print something inside do_configure_prepend() in packages/linux/linux.inc but I don't get any output (the file is being parsed though because when I type some random characters I get a parse error) Oct 02 16:07:31 re Oct 02 16:22:19 gilligan__: last I tried, oenote did exactly nothing Oct 02 16:23:13 i see ;/ Oct 02 16:26:10 oh for F's sake... Oct 02 16:35:58 03xora 07org.oe.dev * rb53d918f... 10/ (1 conf/distro/include/sane-srcrevs.inc): sane-srcrevs.inc : update some openmoko revs Oct 02 16:36:05 03xora 07org.oe.dev * r1e1adf51... 10/ (2 files in 2 dirs): Oct 02 16:36:05 xorg-xserver/xserver-kdrive-1.2.0/kdrive-imageon.patch : drop duplicate old Oct 02 16:36:05 patch Oct 02 16:43:53 !seen zecke Oct 02 16:43:54 zecke (n=ich@gprs-pool-1-016.eplus-online.de) was last seen quitting from #oe 1 day, 3 hours, 56 minutes ago stating (Remote closed the connection). Oct 02 16:44:06 D'OH Oct 02 16:44:27 are most people in here using bash or something else like zsh or whatever ? Oct 02 16:45:02 yes Oct 02 16:45:11 I use bash Oct 02 16:45:21 yes to an "or" question ain't that good :) Oct 02 16:45:38 yes means the first Oct 02 16:45:43 ;] Oct 02 16:45:48 like lambda calöculus Oct 02 16:46:00 calculus even Oct 02 16:46:37 in packages/linux/linux.inc : if [ "${TARGET_OS}" == "linux-gnueabi" -o "${TARGET_OS}" == "linux-uclibcgnueabi" ]; then Oct 02 16:47:00 '==' should be '=' Oct 02 16:47:17 let me see Oct 02 16:47:22 at least only works when changed to = here Oct 02 16:47:24 looks rather python code Oct 02 16:47:27 than shell Oct 02 16:47:46 do_configure_prepend() { Oct 02 16:47:47 nope Oct 02 16:47:48 its shell Oct 02 16:47:49 scruggs, you saw I updated the patches yesterday Oct 02 16:48:37 Crofton: yes - sorry I didn't get that to you earlier - there's no rush (or even need) to really get that applied Oct 02 16:48:48 I think I'm the only user using that lcd display Oct 02 16:48:50 woglinde, and looking at the manpage of test ... "STRING1 = STRING2 the strings are equal" Oct 02 16:48:57 gilligan then write a bug please and attach a fixing patch Oct 02 16:49:03 hm yes its shell code Oct 02 16:49:05 scruggs, I think I am going to sit and wait for the patches to flow into buildroot Oct 02 16:49:33 woglinde, took me hours to figure out hehe Oct 02 16:49:48 Crofton: I don't blame you - sounds like Craig has a lot to commit here in the shortterm Oct 02 16:49:58 yeah Oct 02 16:50:10 I hope this happens soon, I want to get my netwifimicroSD going Oct 02 16:50:42 woglinde, will report it in a sec.. Oct 02 16:50:56 There have been some user patches posted to the gum list to get it going, but they look a bit ugly Oct 02 16:51:07 yeah Oct 02 16:51:20 Craig said he would commit his patch yesterday, but I haven't seen it Oct 02 16:51:44 gilligan ah Oct 02 16:51:50 its posixshism Oct 02 16:52:00 string1 == string2 True if the strings are equal. = may be used in place of == for strict POSIX compliance. Oct 02 16:52:09 thats from the bash manpage Oct 02 16:52:19 == is a bashism Oct 02 16:52:32 koen yeah I meant = Oct 02 16:53:00 so weg should a fast grep to spot other Oct 02 16:55:18 scruggs, do you have a binary u-boot for the verdex that works with the netwifimicroSD card? Oct 02 16:55:28 and can load files via ethernet? Oct 02 16:55:46 I don't have a verdex :) Oct 02 16:55:59 bummer Oct 02 16:56:06 but the u-boot is supposed to be the same for connex/verdex iirc Oct 02 16:56:27 I am running u-boot 1.2.0 on my connex, built from buildroot late last week Oct 02 16:57:07 are you trying to get mmcinit to work? Oct 02 16:57:30 no Oct 02 16:57:55 I just want a u-boot that has working tftp so I do not need to use serial downloads Oct 02 16:58:12 ahhh yes... Oct 02 16:59:57 looking back at the list it appears tftp should be working for verdex - at least it was back in August Oct 02 17:00:04 not that that really helps you, but... Oct 02 17:01:21 yeah, there is a problem Oct 02 17:01:32 gumstic are a "hobby" for me Oct 02 17:01:40 oh wait - Oct 02 17:02:00 I have to be careful with time management Oct 02 17:02:03 I remember now - u-boot had to be patched to work with the verdex smc driver Oct 02 17:02:17 the netwifimicro thingy Oct 02 17:02:23 its just a hobby for me too Oct 02 17:02:41 * Crofton has too many hobbies :) Oct 02 17:02:52 if you want I can try to build the latest u-boot and send it to you Oct 02 17:02:53 crofton hehe Oct 02 17:03:02 ok, wifistix builds, I am going to push so you guys can test Oct 02 17:03:12 great! Oct 02 17:03:16 EXTRA_OEMAKE works for me Oct 02 17:03:19 compilewise Oct 02 17:03:55 I'll have a look at what you are doing in the file... not sure why it did't work for me, but I'm glad its working now Oct 02 17:04:21 build or run time problem? Oct 02 17:04:32 build-time Oct 02 17:04:35 hmm Oct 02 17:04:54 the build kept trying to pull userspace headers instead of kernel ones best I could tell Oct 02 17:05:01 hmm Oct 02 17:05:21 mine built, and it is pushed Oct 02 17:05:42 Crofton: weren't you going to be working on that thesis today ;) Oct 02 17:05:48 yeah Oct 02 17:05:55 I forgot the print out at home :) Oct 02 17:06:07 I am going home after my 2 PM meeting Oct 02 17:06:09 hobbies are WAY more important than a thesis Oct 02 17:06:16 at this point yes Oct 02 17:06:18 priorities people! Oct 02 17:06:19 I agree ith scruggs! Oct 02 17:06:29 It is pretty much ready for my advisor to look through Oct 02 17:06:58 crofton sopunds good Oct 02 17:07:12 what's the point of packages/dbus/dbus-1.0.2/cross.patch ??? Oct 02 17:07:23 it adds another [ac_cv_have_abstract_sockets=no] Oct 02 17:07:34 it looks voodoo to me Oct 02 17:08:19 bedboi: for me it fixed a build error I was getting on dbus-glib I was getting yesterday on angstrom/arm/uclibc Oct 02 17:08:32 damn Oct 02 17:08:37 on mipsel it doesn't work Oct 02 17:08:57 for some reason i get: checking abstract socket namespace... configure: error: cannot run test program while cross compiling Oct 02 17:08:57 hmmm Oct 02 17:09:10 thats the exact error I got before the patch Oct 02 17:09:17 lol Oct 02 17:09:48 yep but check the patch, i can't understand Oct 02 17:09:55 [ac_cv_have_abstract_sockets=yes], Oct 02 17:09:55 + [ac_cv_have_abstract_sockets=no], Oct 02 17:09:55 [ac_cv_have_abstract_sockets=no] Oct 02 17:10:01 it's nonsense to me Oct 02 17:10:20 woglinde, bug #3105 Oct 02 17:10:20 you just add another line... What about adding 100 of them? Oct 02 17:10:21 bedboi no Oct 02 17:11:07 some configure/autotool macros needs 3 options for crosscompiling Oct 02 17:12:23 03crofton 07org.oe.dev * ra0d6cd0b... 10/ (3 files in 3 dirs): wifistix-module : Clean up bb file, add configuration file. Oct 02 17:12:39 damn, i hate autotools Oct 02 17:12:48 bedboi hm Oct 02 17:12:59 they doing a lot of good stuff Oct 02 17:13:04 woglinde: do you have any good reason not to hate them? Oct 02 17:13:06 but the people dont read the manual Oct 02 17:13:18 woglinde: the manual is bigger than the bible Oct 02 17:13:24 nobody can read it Oct 02 17:13:28 nobody does that Oct 02 17:13:29 :) Oct 02 17:13:33 bedboi, they are nice as long as someone provides you with a set of working auto scripts.. as soon as you have to start working on them for yourself its ARGH ;] Oct 02 17:13:33 ???? Oct 02 17:13:40 bedboi I read it Oct 02 17:13:45 woglinde: the bible? Oct 02 17:14:00 no the manuals to autoconf automake and libtool Oct 02 17:14:44 well there is no doubt that autotools are quite messy... just start drawing down the dependencies between the different files/programs/steps for bootstrapping/configuring.. you will get dizzy Oct 02 17:15:01 yep exactly Oct 02 17:15:35 and as soon as a project gets big it's horrible to maintain Oct 02 17:15:58 word up Oct 02 17:16:11 bedboi: check siteinfo.bbclass to see which site files mips(el) is using Oct 02 17:16:12 gilligan__: it's voodoo Oct 02 17:16:24 also I wish that just every kind of program/application out there was using pkg-config.. but they aren't ;/ Oct 02 17:16:57 yep and that mess the things up Oct 02 17:16:58 badly Oct 02 17:18:08 a bit of advice, please someone.... After a recent mtn pull;mtn update, I get an issue with portmap package. - fakeroot: command not found. Post a bug report, or look for the culprit in these sacred halls?? Oct 02 17:18:29 apt-get install fakeroot Oct 02 17:18:47 gilligan__: k thx Oct 02 17:19:05 oe should build fakeroot-native on it's own Oct 02 17:19:16 so it seems portmap is missing it in DEPENDS Oct 02 17:19:24 that's what I should have thought. Oct 02 17:19:47 this is on task 998. Wouldn't fakeroot have been used many times already? Oct 02 17:20:01 not many packages need it Oct 02 17:20:13 only ones that need to install stuff with root permissions Oct 02 17:20:14 k Oct 02 17:20:23 e.g. tinylogin, sysvinit, makedevs Oct 02 17:36:01 I've ordered a chumby Oct 02 17:36:36 Crofton: :) Oct 02 17:37:28 Crofton: why mwlan0 and not wlan0 ? Oct 02 17:37:57 we'll have to ask the gumstix followers :) Oct 02 17:38:14 right, they want to be 'special' ;) Oct 02 17:38:23 I suppose Oct 02 17:38:31 I'll see what I can find out Oct 02 17:38:43 if it's wlan0 all OE magic works Oct 02 17:39:17 since wlan0 is in the default /etc/network/interfaces Oct 02 17:39:26 yeah Oct 02 17:39:28 and in almost all scripts Oct 02 17:39:38 Let's see what we can find out Oct 02 17:40:53 +++ packages/wifistix/wifistix-modules/wifistix.conf 4f54c69eb79183559844c1a626d6423719ac141f Oct 02 17:40:54 @@ -0,0 +1,4 @@ Oct 02 17:40:54 +install pcmcia /sbin/modprobe --ignore-install pcmcia && modprobe pxa2xx-cs Oct 02 17:40:54 + Oct 02 17:40:55 +alias mwlan0 mcf25 Oct 02 17:41:10 it looks like that creates mwlan0 Oct 02 17:41:11 I know Oct 02 17:41:18 but I'm a modprobe n00b Oct 02 17:41:22 heh Oct 02 17:41:24 me too Oct 02 17:41:29 inmods Oct 02 17:41:33 er insmod Oct 02 17:41:42 we could just alter the module source if needed to get wlan0 :) Oct 02 17:42:14 IMNSHO we don't need bugcompatibility with buildroot Oct 02 17:42:15 urg Oct 02 17:42:34 I want to maintain patch compatibility Oct 02 17:43:50 do_compile_prepend() { for i in *.c ; do sed -i -e s:mwlan0:wlan0:g $i ; done} Oct 02 17:44:16 can't you find something productive to do :) Oct 02 17:44:45 * koen is scanning the roll of film he developed today Oct 02 17:44:58 ah Oct 02 17:45:00 against all odds, I didn't screw it up Oct 02 17:45:12 you developed personally? Oct 02 17:45:17 yes Oct 02 17:45:20 cool Oct 02 17:45:24 the first time ever(tm) Oct 02 17:45:33 in high school I played with that Oct 02 17:46:28 almost a full roll of guinnea pigs Oct 02 17:46:43 * koen needs to get something more exciting to photgraph close to home Oct 02 17:52:25 Crofton: I would have no issue with wlan0, I have no idea why Craig used mwlan0 Oct 02 17:52:38 ok Oct 02 17:52:54 I guess I'll look in the source, but have a meeting in a few minutes Oct 02 17:52:58 I pondered just making it wlan0 with the move to OE Oct 02 17:53:21 I'll try changing it and see if anything blows up Oct 02 17:53:24 I'll have to see if I can use my laptop on the plane Oct 02 17:53:37 Why wouldn't that work? Oct 02 17:53:45 mechanical reasons Oct 02 17:53:54 I fly peasant class Oct 02 17:54:06 Ah, that's my issue. 6'4" Oct 02 17:54:22 I have found I cannot get the laptop far enough away from me tpo use the kb Oct 02 17:54:31 Same here Oct 02 17:54:34 I play games on the 770 Oct 02 17:54:46 Gives me a good excuse to read a book Oct 02 17:54:48 * koen will read Hunters Of Dune - Brian Herbert in the train to berlin Oct 02 17:55:04 good idea, go to the bookstore Oct 02 17:55:04 hopefully it won't aggrevate me as all the other dune novel he did Oct 02 17:55:32 how long is the train ride? Oct 02 17:55:34 would be fun to come to Berlin, but can't spare the time & cash right now Oct 02 17:56:09 Crofton: slightly over 4 hours Oct 02 17:56:11 you shoudl try FOSDEM Oct 02 17:56:15 it is a nut house Oct 02 17:56:37 well crazy packed hallways Oct 02 17:57:12 oh well meeting time Oct 02 17:59:49 * koen wonders why fuji neopan 1600 isn't available in 120 roll film Oct 02 18:00:16 ah well, onto ilford delta 3200 for OEDEM Oct 02 18:01:28 this is an interesting time to get interested in analog photography... :) Oct 02 18:01:45 koen: so are you doing your own c41 processing now? Oct 02 18:01:54 sakoman: no, only b/w Oct 02 18:02:10 I can get c41 done at the store real quick Oct 02 18:02:18 b/w needs to get shipped to a lab Oct 02 18:02:27 ~4 day turnaround Oct 02 18:02:31 stock up on chemicals while they still make thm :-) Oct 02 18:02:49 yeah Oct 02 18:08:24 Crofton: how much are the chumbies nowadays? Oct 02 18:09:04 I vote for wlan0 too Oct 02 18:10:37 sakoman: the man at the store said "buy a pack of paper and get an enlarger for free" Oct 02 18:11:10 koen: that is *really* funny! Oct 02 18:13:11 scruggs: did the wifistix stuff work for you? Oct 02 18:13:29 built fine - about to flash the image Oct 02 18:16:52 looks like the devname is set to mwlan in marvell-devicename.patch Oct 02 18:17:33 I'll change it to wlan and flash that and see if it works Oct 02 18:22:08 koen: where is the proper location(s?) to put module_autoload_xxxx ? Oct 02 18:23:17 koen: I know it works in machine.conf, would it also be OK in a task? Oct 02 18:25:05 sakoman: IMO the proper place is the kernel recipe, since you need to rebuild that to make it work Oct 02 18:25:24 the machine.conf is the next best thing Oct 02 18:25:43 I'm not sure if it works in other recipes, since as I said, the kernel recipe needs to know about it Oct 02 18:25:52 koen: OK, works for me! Oct 02 18:26:05 I'll keep it in my machine for now Oct 02 18:26:44 Once we reach some sort of stability with gumstix we can move it to the kernel Oct 02 18:31:10 sakoman: http://www.flickr.com/photos/koenkooi/1473117176/ Oct 02 18:31:16 * koen heads to the tv Oct 02 18:32:58 koen: lot of fun isn't it? you'll never forget the smell of fixer :-) Oct 02 18:34:22 scruggs: let me know if your experiment works. I'm trying it too but it will be a bit before the build completes Oct 02 18:35:46 scruggs: looks like the only change is marvell-devicename.patch? Oct 02 18:36:03 sakoman: will do - right now the build completes but I'm not seeing any .ko files generated... looking into it now Oct 02 18:42:47 sakoman: I fgrep -r for mwlan and that was the only file that matched (other than the file it patched) Oct 02 18:45:33 sakoman: are you building against what Crofton pushed an hour or 2 ago, or the files you sent me? Oct 02 18:46:00 what I sent you plus the name change Oct 02 18:46:09 try and get Craig to use the same patch :) Oct 02 18:46:46 in eet package (efl libraries) I noticed that the .so link is included in the libeet0 package, instead of the libeet-dev package. I don't know if this happens in all libraries or just in eet. Any clues? Oct 02 18:47:04 Crofton: if you still need a working u-boot.bin for netmicrowifi-vx a user just posted one on the gumstix list Oct 02 18:47:32 ah Oct 02 18:47:33 cool Oct 02 18:47:49 Crofton: with your latest changes, the build completes fine but I don't see any .ko files generated - is it working for you? Oct 02 18:47:57 http://www.davehylands.com/gumstix-wiki/verdex-1533-wifi/ Oct 02 18:48:00 is the link Oct 02 18:48:13 Dave is pretty reliable, so I'm sure it'll work Oct 02 18:49:13 scruggs: my build just completed. A quick glance at the rootfs indicates that all the files are where they should be. I'll try a reflash now. Oct 02 18:49:58 sakoman: awesome! it would really be nice to have a good ol wlan0 like everyone else ;) Oct 02 18:50:17 Gonna grab some lunch while it downloads :-) Oct 02 18:50:29 using serial link Oct 02 18:51:10 when the ipkgs are created do the files get removed from install? Oct 02 18:51:34 dunno.. Oct 02 18:51:51 but they'd still be in the src folder where they were compiled, right? Oct 02 18:52:21 Henryk: mono patch pushed -- thanks! Oct 02 18:54:49 03cbrake 07org.oe.dev * rdf94b58a... 10/ (1 packages/linux/linux.inc): linux.inc: replace == with = to make posix shell compliant Oct 02 18:54:55 03henryk 07org.oe.dev * r7604e2e8... 10/ (8 files in 3 dirs): (log message trimmed) Oct 02 18:54:55 mono 1.2.5.1: added mono.bbclass, many changes required for packaging Oct 02 18:54:55 New file: packages/mono/mono-mcs-intermediate_1.2.5.1.bb Oct 02 18:54:55 Compiles mono in native mode with standard prefix, then tars up the Oct 02 18:54:56 resulting tree and puts the tarfile into staging Oct 02 18:54:58 New file: packages/mono/mono_files.py Oct 02 18:55:00 Automatically generated using collect-path.py (attached to this mail) Oct 02 18:56:12 Crofton: the generated ipks only have the wifistix.conf in the data archive - no modules Oct 02 18:56:24 bugger Oct 02 18:56:30 is yours the same? Oct 02 18:58:15 is ${base_libdir} correct? Oct 02 18:59:37 hmm... not sure - I'll check some other bb files Oct 02 19:00:01 I wonder if "oe_runmake install" is bad Oct 02 19:00:57 I don't think its that - the compile step completes too fast Oct 02 19:01:06 yeah Oct 02 19:01:11 no .ko anywhere Oct 02 19:01:22 madwifi uses: FILES_${PN} = "/lib/modules/" in its bb file Oct 02 19:01:46 I think thats what I went from when I made the first shot at the recipe Oct 02 19:01:48 heh Oct 02 19:02:05 that should also be fixed :) Oct 02 19:02:16 :) Oct 02 19:17:23 sakoman: odorless fix :) Oct 02 19:24:34 the compile log has make, but nothing happened Oct 02 19:28:22 yep - I've changed back to what sakoman sent earlier now and its building again, but I don't see why it is failing (or not failing and completing without a failure but still doing nada) Oct 02 19:28:31 hmmm Oct 02 19:29:23 quoting issues Oct 02 19:29:53 sounds likely Oct 02 19:31:35 I dropped the quotes, stupidly Oct 02 19:31:43 now I need to figure out how to get them back in Oct 02 19:33:44 'FLAGS="${CFLAGS}"' Oct 02 19:35:04 ok Oct 02 19:35:09 commiting and psuhing Oct 02 19:35:13 sorry about that Oct 02 19:37:14 scruggs, fixed Oct 02 19:37:42 well, there is a .ko in install :) Oct 02 19:37:43 I'll sync up and give it a go Oct 02 19:38:10 did you happen to change the devname too? ;) Oct 02 19:42:44 koen: aw, that's no fun! I like that smell. Brings back good memories Oct 02 19:43:01 scruggs: flashing right now Oct 02 19:43:30 took a nice long lunch, it is just so beautiful out today Oct 02 19:45:03 80F out, but we just had our first rain of the season so the air is clear and Mt. Shasta, Mt. Lassen, and the Trinity Alps are dusted with first snow Oct 02 19:47:48 we got our first snows here in CO recently too - gotta love this time of year Oct 02 19:49:03 I live at 800 feet, so fortunately my only snow encounters are of the visual kind -- unless I choose to drive 20-30 miles up the road Oct 02 19:49:22 I live at 7300 ft :) Oct 02 19:50:02 no shortage of snow here Oct 02 19:50:06 I like 800 ft cause I can grow bananas & palm trees :-) Oct 02 19:52:27 scruggs: woo-hoo! build was successful: wifi starts up at boot and ntpd picks up the time! Oct 02 19:52:41 sweet! Oct 02 19:52:52 I'm about 1 minute away from a boot myself Oct 02 19:53:20 Think there still might be some module ordering issues because I get: Error for wireless request "Set ESSID" (8B1A) : Oct 02 19:53:20 SET failed on device wlan0 ; No such device. Oct 02 19:53:56 I think pcmcia hasn't fully started the first time the driver tries to init the wlan0 device Oct 02 19:54:25 In fact looking at the boot messages I'm pretty sure that is the case Oct 02 19:54:33 hmm.. the module didn't autoload for me Oct 02 19:54:45 But the code is smart enough to get its act together later Oct 02 19:55:05 Did you remember to put the autoload in your device file? Oct 02 19:55:09 but when I modprobe it it shows up as wlan0 and gets an ip lease Oct 02 19:55:39 yep - I added module_autoload_mcf25 = "mcf25" to gumstix-connex.conf Oct 02 19:57:29 03crofton 07org.oe.dev * r6d07e119... 10/ (1 packages/wifistix/wifistix-modules_5.0.16.p0.bb): wifistix-modules : Fix quoting in EXTRA_OEMAKE. Oct 02 19:57:30 and you did the /etc/network/interfaces change? Oct 02 19:58:26 I changed it to use wlan instead of mwlan, so I didn't make any change to interfaces Oct 02 19:59:03 I am pretty sure it is mwlan for marvell wlan Oct 02 19:59:22 Check to see if you have mcf25 in /etc/modules Oct 02 19:59:45 root Oct 02 19:59:48 :-) Oct 02 19:59:53 Damn windows! Oct 02 20:00:04 Never pick the right one :-) Oct 02 20:00:05 what is windows? Oct 02 20:00:16 not the os! Oct 02 20:00:20 heh Oct 02 20:00:30 RP: you there? Oct 02 20:00:30 :) Oct 02 20:00:40 I can't decide if things are better when the annoying grad student is not working and not talking to people Oct 02 20:00:53 or trying to work that may have to be un done later Oct 02 20:00:55 XorA|gone: yes Oct 02 20:01:14 RP: any idea where /sys/bus/platform/devices/w100fb/extmem went? Oct 02 20:01:32 sakoman: what should the entry in /etc/modules look like? Oct 02 20:01:40 XorA: not really. Is the patch still applied? Oct 02 20:02:05 I have: ### update-modules: start processing /etc/modutils/mcf25 ### update-modules: end processing /etc/modutils/mcf25 Oct 02 20:02:10 RP: Ill have to check Oct 02 20:02:18 scruggs: should be a list of modules to be autoloaded Oct 02 20:02:39 hi rp Oct 02 20:02:45 hi woglinde Oct 02 20:02:48 generated by update modules. Mine has the list of stuff I specified Oct 02 20:03:18 yep - mine just doesn't have mcf25 Oct 02 20:04:48 scruggs: add it by hand & reboot. At least you be able to see if everything else is ok then Oct 02 20:05:03 is this my fault? Oct 02 20:05:34 I see it in install Oct 02 20:06:16 Crofton: no clue! I really should do a clean build to make sure that everything works right. I haven't done one in quite a while Oct 02 20:06:52 if it is in your install perhaps scruggs needs to do the clean build! Oct 02 20:07:00 sakoman: when I did that my boot freezes just after the: 'After register driver' message Oct 02 20:07:16 RP: any idea which patch? Oct 02 20:07:22 I'm not checking images atm Oct 02 20:07:57 scruggs: well that sucks! is your image the result of a clean build? Oct 02 20:07:59 XorA: w100_extaccel looks likely Oct 02 20:08:05 Oct 02 20:08:08 seems to be applied Oct 02 20:08:09 oops Oct 02 20:08:12 I build cfio.ko and mcf25.ko Oct 02 20:08:19 no - its about 3 days old Oct 02 20:08:51 I think mine is a bit fresher than that Oct 02 20:09:00 * Crofton wonders if "oe_runmake install" does what I am expecting Oct 02 20:09:25 others use it Oct 02 20:09:32 Crofton: what isn't it doing properly? Oct 02 20:09:59 * Crofton is thinking aloud Oct 02 20:10:07 they get into ./install Oct 02 20:10:13 so that should be fine Oct 02 20:10:56 Crofton: seems to be doing the right thing for me Oct 02 20:11:39 yeah Oct 02 20:12:02 removing smc91x and rfcomm from modules got it working and autoloading mcf25 Oct 02 20:12:27 guess I'll blow away tmp and start a fresh build Oct 02 20:12:35 Crofton: I'll do a pull/update and try a clean build with what you checked in (plus my collection changes for everything except wifistix) Oct 02 20:13:18 sakoman: did you change mwlan to wlan for what you just tested? Oct 02 20:13:24 I have three machines with OE, I try to do clean rebuilds regularly to catch problems before you guys :) Oct 02 20:13:27 yes, works just fine Oct 02 20:13:39 should I go ahead and push that change? Oct 02 20:13:41 worked for me too Oct 02 20:13:51 crofton: my vote is yes Oct 02 20:13:53 w100-extmem.patch ahah Oct 02 20:14:00 my vote is yes too Oct 02 20:14:16 and I know what koen's vote is :) Oct 02 20:14:19 I'll wait for you to do that before I pull Oct 02 20:14:32 delete the m from wmlan? Oct 02 20:14:36 RP: ahah, patch has been missing since 2.6.20, shows how long since I booted my c7x0 Oct 02 20:14:37 er mwlan Oct 02 20:14:50 crofton: yes~!1 Oct 02 20:15:01 yep - in the patch & the conf file Oct 02 20:15:52 and in /etc/network/interfaces in netbase if you changed it to mwlan0 Oct 02 20:16:07 ok, I am fixing the patch Oct 02 20:16:33 ok, I needed to remove the alias also Oct 02 20:16:36 let me know when it's done & I'll pull/update/clean build Oct 02 20:16:39 no, make the alias say wlan? Oct 02 20:16:50 I made it say wlan Oct 02 20:16:57 alias wlan0 mcf25 , correct? Oct 02 20:17:21 I used: alias wlan0 mcf25 Oct 02 20:17:21 think so Oct 02 20:17:27 yeah Oct 02 20:17:32 works fine Oct 02 20:17:35 this is what I am goign to commit Oct 02 20:20:08 pushed, no checking if it builds Oct 02 20:20:15 * Crofton knows this is backwards ... Oct 02 20:20:37 ok Oct 02 20:29:01 crofton: ok, I'll do the testing :-) Oct 02 20:31:14 XorA: Its in poky's 2.6.22 Oct 02 20:31:26 RP: damnation I just cleaned it up to apply as well Oct 02 20:31:47 XorA: Sorry :/ Oct 02 20:32:01 RP: I should have thought to steal from poky like normal Oct 02 20:32:49 RP: its not in my poky Oct 02 20:33:04 w100_extmem-r0 Oct 02 20:33:18 I've asked this before and I'll try again. I've been trying to build gtkpod for an arm9 and have built libgpod with no problem. then, when I build gtkpod, with autotools inherited and pkg-config, it never includes libxml2 include directory when it compiles and doesn't find libxml/xmlversion.h. I have reached the end of knowing what else to try. Is there a different way to specify include directories? Oct 02 20:35:36 we really need to figure out how to do stuff like work on wifistix in a branch and merge when done :) Oct 02 20:36:46 Crofton: have you looked at how Craig is setting up the machine files in his branch? Oct 02 20:36:54 not lately Oct 02 20:37:09 I pulled once, need to update and diff Oct 02 20:37:33 RP: for your review and inclusing in RPSRC http://www.xora.org.uk/~dp/w100_extmem-r1.patch Oct 02 20:38:44 I'm was just pondering what the best way would be to organise a gumstix build... there are lots of options Oct 02 20:38:51 any thought? Oct 02 20:39:00 I got to graduate from here, fees keep going up .... Oct 02 20:41:08 03crofton 07org.oe.dev * r7f2e567f... 10/ (2 files in 2 dirs): wifistix : Change wifistix wireless device name from mwlan to wlan. Oct 02 20:41:15 03crofton 07org.oe.dev * r86c3bb40... 10/ (1 packages/wifistix/wifistix-modules/wifistix.conf): wifistix : Update conf file with new device name. Oct 02 20:41:54 crofton: without a custom /etc/network/interfaces the wifistx won't come up automatically. need to add an auto wlan0. also the stock interfaces file assumes the use of supplicant Oct 02 20:42:38 the needs of the dedicated differ from the needs of the hobbyist .... Oct 02 20:42:45 :) Oct 02 20:43:08 it Oct 02 20:43:27 it's like we need to build a custom interfaces file based on the hardware ... Oct 02 20:44:07 we could, or we could just auto both eth0 and wlan0 and live with the error message at boot Oct 02 20:44:27 that's what I'm doing at the moment Oct 02 20:44:41 the supplicant issue is larger though :-) Oct 02 20:44:50 gnome-vfs uses xml2 but doesn't do anything more than include it as part of DEPENDS Oct 02 20:45:05 it seems like people are happy letting stuff error at boot, as long as it does not have a hug timeout Oct 02 20:46:05 seems that way. I plan to do a custom interfaces per product though Oct 02 20:46:19 sakoman: I haven't edited interfaces and my wlan0 starts at boot ok - what is happening for you? Oct 02 20:46:26 yeah, I suspect that in the end, fine tuning occurs Oct 02 20:47:00 scruggs: cool! maybe it isn't needed anymore Oct 02 20:47:14 koen sounded like it would work Oct 02 20:47:21 seems to work fine Oct 02 20:47:29 checking Oct 02 20:47:47 * koen has auto wlan0 in his interface file Oct 02 20:47:50 in fact, not having the pre-up iwconfig line gets rid of the kernel oops at boot as an added bonus Oct 02 20:47:56 never took th etime to RFC it Oct 02 20:48:14 koen: just get NetworkManager working Oct 02 20:48:30 on my TODO Oct 02 20:49:11 koen, You were right about portman needing a DEPENDS for fakeroot-native. Putting in tracker. Oct 02 20:49:43 scruggs: removing the auto wlan0 results in no networking at boot for me! Oct 02 20:50:34 sakoman: I think you are misundering me Oct 02 20:51:21 hang on a sec Oct 02 20:51:53 scruggs: very well could be misunderstanding! Oct 02 20:52:21 scruggs: did the experiment just now. works with auto, doesn't work without Oct 02 20:52:40 sorry - I thought you meant my line about iwconfig Oct 02 20:52:52 you are correct - doesn't start without auto wlan0 Oct 02 20:53:38 I thought it was starting at boot - my mistake Oct 02 20:54:15 scruggs: hence my comment about the need for a custom interfaces for gumstix in netbase :-) Oct 02 20:54:41 or about adapting the standard OE one Oct 02 20:55:15 That's precisely what I did: added the auto to the std OE one! Oct 02 20:55:49 ok - to make the situation more complicated Oct 02 20:55:59 if I remove mcf25 from /etc/modules Oct 02 20:56:15 * Crofton heads for home Oct 02 20:56:25 and then load it after I login it auto starts wlan0 and grabs an ip from dhcp Oct 02 20:57:06 scruggs: isn't that what you would expect? Oct 02 20:58:08 what changes in that context to have it automatically bring up the interface? Oct 02 21:00:09 I would expect that loading the module would do it Oct 02 21:00:22 isn't that what you did post boot? Oct 02 21:00:33 mabye... guess I should read a bit more on netbase Oct 02 21:01:10 so I have not edited /etc/network/interfaces at all, so there is no auto wlan0 Oct 02 21:01:54 I guess my question is what is different about the module autoload that it doesn't also bring up that interface as it does if you modprobe it manually Oct 02 21:02:56 I don't know the answer to that question :-) Oct 02 21:03:24 guess I don't really need to know, as long as the behavior is consistent :) Oct 02 21:03:31 Good thing to research when I get a little time Oct 02 21:03:48 but with that in mind, I agree with you that we need to auto wlan0 somehow Oct 02 21:04:46 I just created a machine directory in a netbase collection, copied the standard interfaces file into it and added the auto wlan0 Oct 02 21:05:28 will it work if you add the machine dir to the regular netbase package, or does it only work with a collection? Oct 02 21:05:43 I think I've already asked you that Oct 02 21:05:52 and you said no Oct 02 21:06:10 either will work. I like the collections approach cause I like to keep my mods separate and easily visible Oct 02 21:06:20 ahh. ok Oct 02 21:06:59 well, I have a clean build started so I'm gonna go out and enjoy the day for a bit Oct 02 21:07:00 do you autoload the rfcomm and smc91x modules in your build? Oct 02 21:07:19 I'm just about to do the same Oct 02 21:07:32 checking Oct 02 21:08:49 I autoload pcmcia, smc91x, mcf25, gumstix_gadget, and g_file_storage Oct 02 21:09:24 ok - thx Oct 02 21:09:26 probably should remove the smc91x since I'm not using it Oct 02 21:09:57 scruggs: ttyl! Oct 02 21:10:05 good to know its not causing you problems though Oct 02 21:10:11 take it easy! Oct 02 21:16:29 03xora 07org.oe.dev * rd3bdc1b5... 10/ (1 packages/xorg-xserver/xserver-kdrive-imageon_1.2.0.bb): Oct 02 21:16:29 xorg-xserver/xserver-kdrive-imageon_1.2.0.bb : enable built in fonts Oct 02 21:16:29 so it will start Oct 02 21:43:30 anyone able to build packages for wrt54g? Oct 02 21:43:35 XorA|gone: Its on the rpsys site now, sorry for the delay Oct 02 21:43:40 openwrt is a _pain_ Oct 02 21:43:43 really a pain Oct 02 21:46:22 I'm getting a parse error from bitbake: http://pastebin.com/m1caf179b on the mono bb file.. any ideas Oct 02 21:48:28 cbrake, ping Oct 02 21:48:37 Crofton|home: hello Oct 02 21:49:00 cbrake, you see the comment about parse error in mono? Oct 02 21:49:09 Crofton|home: just looked at it Oct 02 21:49:14 thanks Oct 02 21:50:32 scruggs: is the rest of the build still working? Oct 02 21:51:08 * cbrake is puzzled why he is not seeing that here Oct 02 21:51:25 it exits when it hits the parsing errors... I've not tried to bypass the parse step yet Oct 02 21:52:00 scruggs: what version of python are you running? Oct 02 21:52:51 Python 2.4.4 - gentoo build host Oct 02 21:53:16 scruggs: ok, 2.5.1 here -- that has to be it Oct 02 21:53:46 so we need to make that python 2.4 friendly ... Oct 02 21:54:15 or I can just upgrade if I'm a bit outdated Oct 02 21:55:05 I just parsed Oct 02 21:55:20 Crofton|home: what python ver? Oct 02 21:56:04 2.4 Oct 02 21:56:08 hmm Oct 02 21:56:13 scruggs: I have no idea then Oct 02 21:56:28 let me nuke tmp and rebuild Oct 02 21:56:46 I'll co a fresh copy of .dev and see what that does Oct 02 21:57:09 I need to leave for a couple hours, but I'll look at this when I get back tonight -- just leave me messages on IRC with what you find. Oct 02 21:57:15 ok Oct 02 21:57:19 will do Oct 02 22:00:35 mono? I'm here Oct 02 22:01:28 it is python 2.4 friendly, I don't have any other version Oct 02 22:01:38 reparsing with clean tmp Oct 02 22:04:38 I just parsed ok Oct 02 22:05:08 anyone ever tried compiling packages for wrt54g? Oct 02 22:05:12 hmm, but nothing is working Oct 02 22:05:15 not me Oct 02 22:05:27 wtf, need to run for a bit Oct 02 22:05:28 scruggs: what's multimachine? Oct 02 22:06:03 If I deliberately create a parse error at that location I don't see multi_machine_after_parse(d) in the printed environment Oct 02 22:06:42 Henryk: not sure what you are referring to? Oct 02 22:07:29 scruggs: the trace that you posted shows multi_machine_after_parse(d) in line 028 which is not in the mono code and so must be included from somewhere else Oct 02 22:07:49 openwrt is too bad Oct 02 22:07:49 koen: do you have any experience with that? Oct 02 22:08:06 I'm not sure where it is coming from.. Oct 02 22:08:58 I just confirmed with mtn st that I have no changes in my working copy, nukeed tmp/, and bitbake angstrom-minimal-image still throws the same parse error for me Oct 02 22:12:27 scruggs: hmm, ok. I can't see anything that would be wrong with the line though Oct 02 22:14:17 Henryk: I'm still digging... I have collections setup (nothing to do with mono, but you never know) so I'm going to try to get rid of any external vars and see what I get Oct 02 22:16:55 ah, I got it, I think Oct 02 22:17:06 there's a space in front of multi_machine_after_parse(d) Oct 02 22:17:42 yeah - hope that it cause I'm still getting the error after what *should* be a clean run Oct 02 22:18:25 multi_machine_after_parse(d) is indented with 4 four spaces, while I'm using 1 tab Oct 02 22:18:48 is there a fix I can try locally for you? Oct 02 22:19:31 scruggs: edit ./classes/multimachine.bbclass and remove the spaces in front of multi_machine_after_parse(d) and add one tab instead Oct 02 22:19:39 will do... brb Oct 02 22:20:27 made the change... parsing now Oct 02 22:21:23 I think that did it... made it past that file in the parse... Oct 02 22:22:23 yep - parsed 100% - thx Henryk! Oct 02 22:22:27 oh that's just great, because the spaces/tab-inconsistency is all over the place :) tabs seem to be in the majority, though Oct 02 22:22:54 I'll keep that in mind in case I run into any more parse errors Oct 02 22:36:51 Henryk: still having issues.. the parse complete but the 1st package to compile is failing... I'll pastebin the error Oct 02 22:39:08 ok Oct 02 22:40:18 http://pastebin.com/m157b7122 Oct 02 22:41:35 hmm ${MULTIMACH_ARCH} is not being set ... Oct 02 22:43:54 scruggs: ah, maybe try to clean your cache and the package first Oct 02 22:44:18 though, hmm, you did that ... Oct 02 22:44:33 I did rm -rf tmp/ Oct 02 22:46:49 I've not yet done a clean checkout of .dev (although I don't see why it should matter given my mtn diff), so I guess I can try that next Oct 02 22:52:16 fwiw I am doing a rebuild Oct 02 22:52:44 I have an offchance theory but it's ridiculous. Can you try to deliberately break the multimachine.bbclass (Insert a ":" after the multi_machine_after_parse(d) in the python anonymous block) and thus get bitbake to print the surrounding python code? Oct 02 22:52:45 ok - fresh checkout of .dev throws the same parse error... I'll edit the class file and try again.. Oct 02 22:52:57 urg Oct 02 22:53:03 (or is there a better way to see what code get's actually executed?) Oct 02 22:53:06 what distro are you building on? Oct 02 22:53:11 gentoo Oct 02 22:53:31 scruggs, you may want to wait until tomorrow AM Oct 02 22:54:18 ok, forget that deliberately breaking experiment, doesn't work Oct 02 22:54:51 ok... just edited and its parsing, so we'll see what happens for this go-around Oct 02 22:56:07 same failure on do_unpack of 1st package :( Oct 02 22:56:36 ok, I think I kind of found it, again related to whitespace issues: patch.bbclass is included before multimachine.bbclass and has an python anonymous block that ends with an if condition and likely the call to multi_machine_after_parse is interpreteted to be inside the if block Oct 02 22:56:44 64 or 32 bit distro? Oct 02 22:57:10 64 Oct 02 22:57:25 I am doing a clean build on a 64 bit machine Oct 02 22:57:48 scruggs: just a moment, I'm working something out Oct 02 22:58:14 Henryk: still here - if there is anything I can test lmk Oct 02 23:01:41 scruggs: ok, apparently spaces are more common in the python anonymous blocks than tabs. undo your changes and then apply the patch I'll post in a minute Oct 02 23:02:00 will do Oct 02 23:03:39 parsing works on fedora 7 /64' Oct 02 23:04:15 * scruggs wonders if his verson of python was compiled with --be-really-obnoxious-about-whitespace Oct 02 23:04:20 heh Oct 02 23:04:54 I guess the best long term strategy would be to have bitbake normalize the amount and type of whitespace Oct 02 23:10:24 scruggs: try http://people.openmoko.org/henryk/mono-after-parse.patch Oct 02 23:11:10 grabbing it now... Oct 02 23:14:08 patch is applied,,, starting a build Oct 02 23:15:49 Henryk: as long as you are around :) - if I want to get openmoko running on my stix is there anything special I should start looking at as examples? Oct 02 23:16:20 and that seems to fix it - its on package 13 now on the build Oct 02 23:16:39 I don't know, I've neve looked into anything like that Oct 02 23:17:14 ok - I'm sure I can get it going if I dig in there Oct 02 23:17:34 thx for the fix - the build is still going without error Oct 02 23:19:42 scruggs: ok, great. So this was a case of "Ein Geisterfahrer? Hunderte!" Oct 02 23:25:48 (There's another one of those in ./packages/uclibc/uclibc.inc) Oct 02 23:45:39 http://bugs.openembedded.org/show_bug.cgi?id=1738#attach_1172 Oct 02 23:45:44 is this bug closed? Oct 02 23:45:49 i mean there's patch available Oct 03 00:42:33 scruggs, I'll talk to some openmoko guys about getting it going on some stix this weekend Oct 03 00:44:38 Crofton: that would be great! I've had a lot of success just building various openmoko targets, but I've not found any 'porting openmoko to other platforms' instructions Oct 03 00:59:34 Hey all Oct 03 00:59:54 Anyone seen bbimage fail to work, complaining about python-pysqlite when bb isn't installed system wide? Oct 03 01:06:27 Henryk: change pushed -- strange python was responding differently on different machines ... Oct 03 01:10:59 ubuntu 7.04 Oct 03 01:11:01 ga Oct 03 01:11:24 03henryk 07org.oe.dev * r9b11d89e... 10/ (1 classes/mono.bbclass): Oct 03 01:11:24 mono.bbclass: fix whitespace problem Oct 03 01:11:24 - fix for the problem experienced by scruggs on IRC today. There Oct 03 01:11:24 was an inconsistency in the amount and type of whitespace used in the Oct 03 01:11:24 python anonymous blocks between mono.bbclass and multimachine.bbclass. Oct 03 01:11:24 - fixed a small problem in the error output of mono_do_clilibs Oct 03 01:49:07 is there any special reason why modules like hidp and rfcomm and autoloaded by kernel.bbclass? Oct 03 01:50:42 scruggs, most of us have gone to bed :) Oct 03 01:50:53 not too many OE guys west of you :) Oct 03 01:51:05 gn Oct 03 01:53:47 tend to forget that sometimes :) Oct 03 01:58:35 scruggs: I'm still here, but I'm not much use :-) Oct 03 02:00:10 scruggs: btw, my clean build worked just fine. no issues with crofton's push Oct 03 02:00:38 on to my gnuplot libstdc++ issue next Oct 03 02:04:35 sakoman: mine is working better :) Oct 03 02:05:23 I think I know what the problem is (issue with bluetooth + wifistix-cf card) but I need to test it a bit more to be sure Oct 03 02:07:20 Crofton's push worked fine for me too **** ENDING LOGGING AT Wed Oct 03 02:59:56 2007