**** BEGIN LOGGING AT Thu Oct 04 02:59:56 2007 Oct 04 06:05:02 Hi. I hope someone can help clear up my understanding of the meaning of gcc-cross vs. gcc-cross-sdk. Oct 04 06:06:36 gcc-cross is the cross-compiler used to build all the target recipes. Oct 04 06:06:49 but, what is gcc-cross-sdk then? Oct 04 06:07:37 is it a recipe to cross-compile/build a native compiler for target? Oct 04 06:25:46 morning Oct 04 06:28:26 morning Oct 04 06:28:51 EsbenH: I think gcc-cross-sdk if for making cross compiling SDK Oct 04 06:31:20 as in a stand-alone toolchain (as in meta-toolchain)? Oct 04 06:31:56 what is the difference between a "cross compiling SDK" and gcc-cross? Oct 04 06:37:12 gcc-cross is for internal OE use, no use outside of OE Oct 04 06:52:35 unfortunately, gcc-cross-sdk cannot build for powerpc Oct 04 06:53:09 and the problems seems to be rather fundamental, so the archs that can be built are just being lucky. Oct 04 06:53:40 target header files are used by native compiler :-( Oct 04 07:11:01 good morning all Oct 04 07:13:25 good morning koen. Oct 04 07:13:26 yo koen Oct 04 07:19:44 morning, is there an option to remove the /boot directory from rootfs when using the redboot bootloader? Oct 04 07:25:36 Morning. Oct 04 07:25:54 hey Henryk Oct 04 07:26:04 nik0n: multiple ways Oct 04 07:26:20 nik0n: RDEPENDS_kernel-base = "" in your kernel recipe is one way Oct 04 07:27:34 ROOTFS_POSTPROCESS_COMMAND += " rm ${IMAGE_ROOTFS}/boot/uImage* ;" is another Oct 04 07:27:55 hi koen Oct 04 07:28:44 Henryk: mono built for me Oct 04 07:29:01 koen: where to put the ROOTFS_POSTPROCESS_COMMAND? into my distro.conf? Oct 04 07:29:09 koen: ah, excellent Oct 04 07:29:12 MACHINE.conf Oct 04 07:31:05 Henryk: the armv5te version is in the angstrom feeds, the armv4t one will follow shortly after Oct 04 07:31:07 Koen: I'm struggling with my uImage not being compressed in the mkimage step. this is after an update with no other changes. Is there a likely place that a setting may have been changed? Oct 04 07:31:22 koen: I followed the HOWTO on http://scratchbox.org/wiki/OpenEmbedded based on your suggestion in blog. It worked nicely, thanks! Oct 04 07:31:24 russf: packages/linux/linux.inc Oct 04 07:31:32 koen: thanks! Oct 04 07:32:12 EsbenH: it slightly changes, florian (when he's back online) knows more details, since he's actually using it :) Oct 04 07:32:51 thx koen Oct 04 07:33:17 koen: It would be much nicer if gcc-cross-sdk and meta-toolchain worked instead. Oct 04 07:35:41 I spent all day fighting with first crosstool and then oe meta-toolchain/gcc-cross-sdk Oct 04 07:38:30 Have anybody had the obscure idea to support cygwin as target for OE? Oct 04 08:07:55 hello, yesterday i was directed to the mplayer package to enable the overlay2 ycbcr Oct 04 08:08:01 and to dump camera output to it Oct 04 08:08:18 what is the link of the package repository for open embedded Oct 04 08:08:30 i am using ptxdist at the moment Oct 04 08:09:41 on the pxa270 Oct 04 08:09:43 http://www.openembedded.org/filebrowser Oct 04 08:14:03 thanks Oct 04 08:14:27 i found pxa-video_out.patch, but only a function declaration Oct 04 08:14:38 where is the rest Oct 04 08:14:48 video_out_pxa Oct 04 08:14:57 i am looking in http://www.openembedded.org/filebrowser/org.openembedded.dev/packages/mplayer/files Oct 04 08:18:39 vo_pxa.c Oct 04 08:25:06 Henryk, cbrake_away: http://www.angstrom-distribution.org/repo/?name=libmono&arch=&action=search Oct 04 08:26:08 morning all Oct 04 08:27:45 hey RP & zecke Oct 04 08:28:14 moin Oct 04 08:42:11 hi all Oct 04 08:44:34 koen: your very restrained this morning Oct 04 08:47:04 Cool! I just got the UPS tracking ID for the two Neo1973 devices heading for me :-) Oct 04 08:50:53 how big is the mono runtime? Oct 04 08:58:43 depend on what you need to install, it's pretty fine-granular. basic install is somewhat like 3 MB Oct 04 08:58:54 hm not bad Oct 04 09:06:58 when will mips big-endian be supported by OE? Oct 04 09:07:52 Henryk : Interesting. How about the compatibility with windows .net programs ? I understand than COM objects (or how windows calls them) should not work on linux Oct 04 09:08:16 nikon hm provide the patches and it will be fast integrated Oct 04 09:08:21 XorA: I am ? Oct 04 09:08:45 koen: argtable has lasted all night without a dissaprove Oct 04 09:08:58 hehe Oct 04 09:09:08 org.openembedded.dev/argtable/argtable.bb Oct 04 09:09:58 * steliosk thinks that XorA wants it to be disapproved :) Oct 04 09:10:24 XorA: and argtable_2.7.bb actually builds 2.6 Oct 04 09:10:44 steliosk: I just like reading koen's larting Oct 04 09:11:06 steliosk: as far as I understand it mono implements .NET 1.x fully, so any 100% .NET program should run. However, the 1.x API missed some features that's why some developers used P/Invoke to directly access the underlying operating system's libraries which then will be non-portable. .NET 2.x seems to be more versatile but is not yet fully supported under mono. Oct 04 09:11:38 I'm still not convinced that OE should support the notion of "compatible kernel" Oct 04 09:12:04 maybe if we make OE depend on dbus we'll be freed of haralds whining... Oct 04 09:13:20 Henryk: Hmmm makes sense. Would be interested to see if i could get mono going on powerpc. Oct 04 09:13:28 are opie related bugs allowed in the bugtracker? Oct 04 09:13:57 dcordes: opie has its own bugtracker Oct 04 09:14:06 koen: where can I find that? Oct 04 09:26:23 good morning Oct 04 09:31:01 koen: do you mean mantis? Oct 04 09:33:44 dunno Oct 04 09:33:50 opie is dead and I don't use it Oct 04 09:35:11 my akita is horribly broken so I take the chance and exerpiment a bit Oct 04 09:38:54 BTW: who is Wolfgang Spraul? Oct 04 09:39:47 zecke: works for OM Oct 04 09:40:12 I have never heard about him Oct 04 10:01:08 * koen adds http://www.hol.cityvox-de.com/restaurants_berlijn/mai-thai-imbiss_1268/ProfilLieu as POI to maemo-mapper Oct 04 10:01:38 * koen adds http://maps.google.nl/maps?sourceid=mozilla2&ie=UTF-8&oe=utf-8&um=1&q=vuong&near=Berlijn,+Deutschland&fb=1&view=text&latlng=52526660,13407772,7243383145051888058&dtab=2&reviews=1&sa=X&oi=local_result&resnum=1&ct=result as POI as well Oct 04 10:02:31 zecke: is it wise to bump webkit SRCREV now, or do apps need adjusting to alp's changes? Oct 04 10:02:43 koen: they still need it Oct 04 10:02:51 koen: give me 30 minutes to do so Oct 04 10:05:28 zecke: I just discovered that http://fotoimpex.de/ is real close to vuong :) Oct 04 10:08:54 hmm, anyone here ever used or needed path names longer than 100 bytes in ipkg and/or have a suggestion on how i might debug the problem? Oct 04 10:09:30 I suspect we'll need to patch ipkg for that Oct 04 10:09:33 *again* Oct 04 10:10:03 well it at least appears to have code for a longname gnu extension Oct 04 10:15:47 XorA: you've mentioned autorev patch for bitbake on OM mailing list, where can I find that patch? Oct 04 10:16:01 ynezz: OE mailing list Oct 04 10:20:33 hi Oct 04 10:20:43 pls, can somebody apply it? Oct 04 10:22:15 koen: ugh, though apparently there is a different format for including long link names in tar files, which then is actually used in the ipk Oct 04 10:23:43 XorA: how much do you want to pay me/richard to add you to berlios? Oct 04 10:23:50 XorA: do you have an account yet? Oct 04 10:24:04 zecke: no Oct 04 10:24:33 ynezz: apply it yourself Oct 04 10:24:51 it's fine with me, i've patched it myself Oct 04 10:25:09 XorA: when did you send the patch? create an account at berlios.de and I can add your as comitter Oct 04 10:25:33 but some OM users hit that bug... Oct 04 10:25:57 zecke: its RP's patch and its not suitable for mainline Oct 04 10:25:57 XorA: obviously "communicate and think before comitting" is applied to the BitBake svn as well but I trust you... Oct 04 10:26:22 [oe] SRCREV change bitbake bug Oct 04 10:26:24 is the mail Oct 04 10:26:30 http://lists.linuxtogo.org/pipermail/openembedded-devel/2007-September/003035.html Oct 04 10:27:15 when are files copied to tmp/staging ? Oct 04 10:27:35 hm.. does subscribing to the OE mailing list require some human interaction ? subscribed via the web interface some minutes ago and still haven't received a reply/verification mail Oct 04 10:29:12 I'll retract my last statement Oct 04 10:29:30 http://bitbake.berlios.de/manual <-- that's all there is for bitbake documentation? Oct 04 10:30:44 gilligan__: yes Oct 04 10:30:58 ;[ Oct 04 10:31:00 gilligan__: there is some more at http://www.openembedded.org/user-manual (kind of hard to find) Oct 04 10:31:07 gilligan__: and we have a usermanual on OE Oct 04 10:32:39 right.. well i'll just read trough everything I can find now while my current build is running. I can *use* bitbake, but still lack understanding of its internals Oct 04 10:35:42 gilligan__: are you a talented doc writer? Oct 04 10:36:05 zecke, I should be very careful with how I answer this ;) Oct 04 10:36:46 zecke, but seriously.. I wouldn't mind writing some stuff for the wiki and helping to clean it up a bit Oct 04 10:36:47 gilligan__: Good documentations needs a good writer and enough time. We are handicapped with both topics Oct 04 10:37:19 ah, ok, I found the ipkg bug. the logic for long link names is broken in at least two ways Oct 04 10:37:42 zecke, well at this point I lack the knowledge to write about any internals, but I think I could start of by doing some, lets say.. "house keeping" / "janitor editing" Oct 04 10:38:06 gilligan__: that would be a start, we are happy to answer any questions Oct 04 10:43:27 zecke, some small things come to mind immediately.. "OE and your distro" and "getting started" should for example be merged I think. Especially because the first has some vital information that occurs nowhere in the latter .. for example that there are problems with using dash with OE. Oct 04 10:44:14 plus of course things like empty/non-existent pages Oct 04 10:44:36 fixing argtable -- sorry about that ... Oct 04 10:46:36 zecke, some other suggestions from my side would be to simplify/change the local.conf sample somewhat. Use angstrom as default and add ANGSTROM_MODE for example. Also the sanity checker should probably check if /bin/sh is linked to dash or bash Oct 04 10:47:03 gilligan__: local.conf.sample Oct 04 10:47:28 zecke, well exactly that should be changed yes :) Oct 04 10:47:46 eh..WTF Oct 04 10:47:49 http://www.openembedded.org/wiki/MonotonePhraseBook Oct 04 10:47:59 filled with PORN spam Oct 04 10:50:15 EsbenH : During the do_stage Oct 04 10:51:45 yay, we are a successful project, our wiki is getting spammed Oct 04 10:52:09 steliosk: thought so, but I just tried rebuilding glibc in several ways, and stdio.h was not overwritten in my staging dir. Oct 04 10:52:47 again? Oct 04 10:53:11 gilligan__: not here Oct 04 10:53:34 HenryK : *warning Mono related question* Any idea why i get errors about missing mscorlib.dll during linking of some windows C# apps ? Oct 04 10:53:47 zecke: you're too late, someone already reverted it Oct 04 10:53:56 no spam here too Oct 04 10:54:03 zecke, someone just changed it.. or it was before and was only in my firefox cache, I don't know.. well now it is gone Oct 04 10:54:04 http://www.openembedded.org/node/22/revisions Oct 04 10:54:13 it was there when gilligan__ wrote the comment, I saw it too Oct 04 10:54:35 okay..so i am not making this up ;] Oct 04 10:54:47 was starting tow wonder myself hehe Oct 04 10:55:06 * steliosk thinks that in many cases it will difficult to distinguish between PRON span and the actual monotono phrasebook ;) Oct 04 10:55:11 steliosk: you mean not on OE? No idea, I've only packaged mono, not yet actually used it Oct 04 10:55:30 maybe require login and/or add one of those wotcha thingies Oct 04 10:56:18 HenryK : Yes in OE. I am not an expert in Mono either Oct 04 10:56:31 Re-Captcha would be cool Oct 04 10:57:09 EsbenH : Weird... Did you try to clear (-c clean glibc) and rebuild or just made changes and re-run it ? Oct 04 10:57:34 virtual/kernel is designated by the currently used machine right? Oct 04 10:57:48 steliosk: corlib is not automatically pulled in as a dependency. you'll have to install either libmono-corlib1.0-cil or libmono-corlib2.0-cil Oct 04 10:57:57 steliosk: tried both -crebuld and -cclean and then just bitbake Oct 04 11:00:48 hi guys Oct 04 11:01:04 HenryK : looks possible. will try it later Oct 04 11:01:52 hi bernardo Oct 04 11:02:34 I'm having ssl certificates problems with svn fetching fontconfig-2.4.1-r1, and I've already tried doing a "svn ls" and storing the certificate permanently. Oct 04 11:02:36 EsbenH : Not sure why. There were some changes made recently by RP in the way the toolchain puts things in staging Oct 04 11:02:44 hi woglinde Oct 04 11:03:29 EsbenH : what glibc/distro/arch you are building for ? Oct 04 11:03:51 glibc for ppc603e Oct 04 11:03:52 one of the errors is that the certificate name for 'www.angstrom-distribution.org' is 'www.linuxtogo.org' Oct 04 11:04:00 distro is custom Oct 04 11:04:45 the other error is that 'stage.maemo.org' has no certificate Oct 04 11:05:01 any ideas on how can I circumvent this? Oct 04 11:06:59 Do we have a wiki page on SRCREV yet? Oct 04 11:07:11 there is an email on the list about it .. Oct 04 11:07:17 EsbenH : I am using glibc 2.5 gcc 4.1.1 on our distro and builds ok. Angstrom builds also (or was a couple of weeks ago) on efika Oct 04 11:08:20 can someone tell me where oe_runmake and oenote and the like are defined/implemented ? Oct 04 11:09:22 steliosk: building for arm, right? Oct 04 11:09:25 new item for OEDEM agenda, remove DISTRO=generic :-) Oct 04 11:10:50 yeah Oct 04 11:10:53 XorA, that is what I suggested earlier on ;] Oct 04 11:10:58 or fix, mickeyl seems to prefer that Oct 04 11:11:07 EsbenH : nop ppc. Efika is ppc603e Oct 04 11:11:22 but basically, we can't let people use it when it is broken Oct 04 11:11:22 Crofton: fix is cp angstrom*.conf to generic *.conf :-) Oct 04 11:11:34 * Crofton|home checks the usd to euro rate and cries Oct 04 11:11:37 * steliosk will be at OEDEM after the code sprint for the "fun stuff" Oct 04 11:11:53 steliosk, when will you arrive? Oct 04 11:12:13 Crofton: how many hours until you fly Oct 04 11:12:30 Probably Sunday or if i manage to postpone a meeting on Saturday Oct 04 11:12:30 well, I start flying in 6 hours Oct 04 11:12:38 cool Oct 04 11:12:49 but here is an hour of driving Oct 04 11:13:06 * XorA avoids that by not being allowed to drive Oct 04 11:13:07 and I need to be there at least an hour early as it is an international departure Oct 04 11:13:18 steliosk: could you try and see if you can build gcc-cross-sdk ? Oct 04 11:13:29 * Crofton|home lives in the US where you are hosed without a car Oct 04 11:15:20 cbrake, http://www.openembedded.org/wiki/OeFaq#development number 11 should probably changed / link changed to your updated article on not using MACHINE_EXTRA_RDEPENDS right? Oct 04 11:15:27 Crofton: and don't you dare wear anything that lights up or has any wires anywhere at the airport. Oh, and no play-doh Oct 04 11:15:37 Crofton|home: I was not hosed, people just made fun out of me by driving a bike ;) Oct 04 11:16:02 * Bernardo is away: Ausente por agora. Oct 04 11:16:54 ah looks like the UK is going to end up with no postal service Oct 04 11:17:33 zecke, where is oe_runmake defined ? Oct 04 11:17:40 gilligan__: base.bbclass Oct 04 11:18:18 zecke, ah..thanks Oct 04 11:18:20 zecke, my friend used to ride is bike to work, until he retired Oct 04 11:19:07 gm Oct 04 11:19:18 * XorA suddenly pictures a new horror movie where a American gets lost in Amsterdam Oct 04 11:19:23 Crofton|home: I was in high school, everybody got a car, I only had a mountain bike Oct 04 11:19:23 hi Oct 04 11:19:35 zecke, in the US? Oct 04 11:19:48 * Crofton|home thinks it sounds like the US :) Oct 04 11:20:18 Crofton|home: yeah :) Oct 04 11:20:27 where? Oct 04 11:20:42 Crofton|home: Redlands, California Oct 04 11:20:51 Crofton|home: I was there as an exchange student Oct 04 11:20:59 Sorry :), they are all crazy in CA Oct 04 11:21:16 Hmm, here in Berlin (at least in the central areas) it's the other way around: You're hosed when you got a car. Oct 04 11:21:35 * Crofton|home does not live in a large city Oct 04 11:26:12 03cbrake 07org.oe.dev * r3c08e9dd... 10/ (4 files in 3 dirs): argtable 2.7: mv to packages, and fix 2.7 version info in recipe Oct 04 11:27:40 hrm.. is there any special reasons for using variables like ${S} and ${D} .. that's about as bad as C programs with global variables called j,k and tmpKKd.. I figured that ${S} means the source directory in given context.. no idea what $D is.. but ${SRC_DIR} or something like that would work, no? :) Oct 04 11:28:01 D = DESTINATION Oct 04 11:28:09 http://people.openmoko.org/henryk/ipkg-long-link-name.patch fixes the ipkg bug with regards to long link names Oct 04 11:28:56 gilligan__: those are used very often and in almost any recipe, so they are short (less typing) and everyone knows them anyways Oct 04 11:29:05 S, D, PN, PV, PR Oct 04 11:30:06 gilligan__: I think that is portage inheritance, but I'm not sure Oct 04 11:30:27 the less-typing is what I also get from ppl using absurd global variables in C .... I think readability > "fast typing" at any point of time and in any sort of situation Oct 04 11:32:42 Henryk, well i'm not everyone then..and I have no idea what they are.. how do i find out ? Oct 04 11:34:38 gilligan__: http://www.openembedded.org/user-manual&dpage=recipes_directories Oct 04 11:34:53 gilligan__: searching the OE manual is helpful: http://www.openembedded.org/%7Ekoen/usermanual.pdf Oct 04 11:35:10 gilligan__: I'll try to update the pdf version today as I find that one a lot easier to use Oct 04 11:36:04 cbrake, yeah well.. question is what should people focus on.. the pdf or the wiki, because the information they both deliver seems to differ a lot ? Oct 04 11:36:24 gilligan__: they are both generated from docbook (or whatever) source Oct 04 11:36:51 gilligan__: I think the html version is updated daily. The OE manual is not a wiki. Oct 04 11:37:12 obviously ;] Oct 04 11:40:54 so I suppose PR = release, PV = version Oct 04 11:41:41 PR = Package Revision Oct 04 11:42:22 aka Recipe Version Oct 04 11:42:44 gilligan__: http://www.openembedded.org/wiki/Bbfile Oct 04 11:43:02 gilligan__: you could add ${D} to this page Oct 04 11:43:37 * koen thinks basic info like that doesn't belong in the wiki, but in the manual Oct 04 11:44:54 XorA: should be apply Erin's gsmd patches for phonebook and sms in OE or keep waiting till upstream finally catches up? Oct 04 11:48:52 Henryk: 13:47 < torpor> NOTE: Task failed: Error: /home/j/hak/openmoko/openembedded/packages/ipkg/ipkg-0.99.163/./lonk-link-name.patch not found. Oct 04 11:49:00 Henryk: is that related to your todays question? Oct 04 11:49:23 morning Oct 04 11:49:47 hey hrw Oct 04 11:50:02 zecke: your 30 minutes are up Oct 04 11:50:33 koen: I need to upgrade my bios first Oct 04 11:50:54 your EFI? Oct 04 11:51:14 koen: workstation one Oct 04 11:51:29 zecke: hmm, you're missing packages/ipkg/files/lonk-link-name.patch (damn, I got a typo in the name, at least it's the same typo in all places) Oct 04 11:51:34 cbrake, added Oct 04 11:51:43 This is truly a third world country, last night (12:30AM) my wife received a tech support call from India :) Oct 04 11:51:53 Henryk: this was torpor in #openmoko Oct 04 11:52:04 cbrake, there seems to be some junk after the D paragraph I added Oct 04 11:52:22 cbrake, not sure what to make of that :) Oct 04 11:52:25 koen: and my slave crashed... Oct 04 11:53:05 i Oct 04 11:53:08 zecke: ah, yes Oct 04 11:53:13 me Oct 04 11:53:18 i just took the line out of the .bb file .. Oct 04 11:53:56 gilligan__: faq updated thx -- I'll be moving this to the manual eventually -- feel free to checkout org.openembedded.documentation and provide patches :-) Oct 04 11:54:09 cbrake, alright Oct 04 11:54:16 do we any list of subjects for OEDEM? in previous year we had such one Oct 04 11:54:41 hrw: mickeyl had one, but feel free to start with some topics Oct 04 11:54:54 hrw: at least I will volunteer to take minutes and publish them as we speak :} Oct 04 11:55:43 ;) Oct 04 11:55:52 Henryk: so is the fix to change lonk to long? Oct 04 11:56:26 torpor: no, the fix is to add packages/ipkg/files/lonk-link-name.patch Oct 04 11:57:09 03henryk 07org.oe.dev * r1b098bf1... 10/ (1 packages/ipkg/ipkg_0.99.163.bb): ipkg: fix long filenames Oct 04 11:57:14 03koen 07org.oe.dev * rf4240373... 10/ (1 packages/ipkg/files/lonk-link-name.patch): ipkg: check in missing patch Oct 04 11:57:31 great! Oct 04 11:57:39 so i should fix the .bb file back to its old values then. Oct 04 11:58:24 hehe, thanks koen Oct 04 12:01:47 grrr, fog i Philadelphia Oct 04 12:01:53 suppose to clear though Oct 04 12:07:10 In Germany, do they have two or three pronged outlets Oct 04 12:10:51 zecke: do we have proper amount of power outlets at oedem? Oct 04 12:10:54 Crofton: 2 pronged and ground on the side Oct 04 12:11:14 Crofton|home: I'll bring an adaptor for you (no voltage conversion!) Oct 04 12:11:18 excellent, I can't find my three pronger Oct 04 12:11:41 koen, I will need more than one :) Oct 04 12:11:54 hrw: that will be solvable Oct 04 12:11:54 bring a us extension block Oct 04 12:12:01 heh Oct 04 12:12:06 that does not work ... Oct 04 12:12:23 we tried it in Italy, the grounding gets messed p and it trips outlets Oct 04 12:13:51 my GB->EuroPlug adapter should be safe for use in .de Oct 04 12:19:50 hm.. aren't the kernel boot params stored in proc somewhere? Oct 04 12:20:14 /proc/cmdline ? Oct 04 12:20:48 right..thanks.. Oct 04 12:20:57 i was looking in /proc/sys/kernel.. ;] Oct 04 12:24:48 zecke: so no need for power extenders - only cat5 (optional) and powercord for laptop/toys needed? Oct 04 12:25:30 hrw: well, extenders are always good, but we have plenty here Oct 04 12:27:52 anyone else here with a compulab besides cbrake and me? :) Oct 04 12:29:06 gilligan__: cm-x270? Oct 04 12:29:26 hrw, yep Oct 04 12:30:02 gilligan__: iirc someone in OH has them but I do not remember who Oct 04 12:30:20 hrw: There was a brief agenda in the wiki for OEDEM Oct 04 12:30:31 hrw: I think we should perhaps mention it on the mailing list Oct 04 12:31:07 gilligan__: I have one, its just a bit dusty atm :-( Oct 04 12:31:10 we should update it also Oct 04 12:31:25 * Crofton|home charges his nerd devices Oct 04 12:32:12 ReaperOfSouls, actually nevermind.. I just figured it out.. I was just being silly and swapping the order of some arguments of nandwrite.. nevermind :) Oct 04 12:32:39 RP: http://www.openembedded.org/oedem2007 you mean? Oct 04 12:35:03 RP, hm..then again.. seems like I do have some probs :) Oct 04 12:36:26 RP, I suppose u've used eraseall / nandwrite to write a rootfs to the nandflash on your compulab at some point? Oct 04 12:39:26 RP, I wonder why the erase-size of the nand flash(128M) is set to only 16KB ? Oct 04 12:40:12 gilligan__: it depends on NAND chip Oct 04 12:40:19 ah, ok Oct 04 12:43:09 jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00018024: 0x5f3b instead Oct 04 12:43:09 Further such events for this erase block will not be printed Oct 04 12:43:09 Node at 0x0001bcb0 with length 0x00000a1e would run over the end of the erase block Oct 04 12:43:09 Perhaps the file system was created with the wrong erase size? Oct 04 12:43:09 mount: /dev/mtdblock3: can't read superblock Oct 04 12:43:11 hm Oct 04 12:56:48 can't get a direct flight to berlin.... Oct 04 12:56:51 cbrake, good timing.. ;] Oct 04 12:57:01 ~lart airlines Oct 04 12:57:02 * ibot gives airlines a "free" copy of Windows and then charges double for "Upgrades" Oct 04 12:57:33 koen: llvm is on for tomorrow? Oct 04 12:57:49 what's the best alternative option for reaching berlin by train ? Munich or Frankfurt ? Oct 04 12:58:26 steliosk: Frankfurt is closer and there is a "sprinter" (express) Oct 04 12:58:55 zecke : Any idea where i can get see the train timetables ? Oct 04 12:59:08 steliosk: bahn.de Oct 04 12:59:44 to those who might not know : "Frankfurt Hahn" is NOT frankfurt.. it takes hours to get from there to Frankfurt :) Oct 04 13:02:03 huhu Oct 04 13:05:56 Jin^eLD: Hi! Oct 04 13:06:01 hey likewise ! Oct 04 13:06:03 Jin^eLD: Will you attend OEDEM?? Oct 04 13:06:11 OEDEM is where again? Oct 04 13:06:15 (Me, not sure yet) Oct 04 13:06:20 I will be in Linz at the CELF conference Oct 04 13:06:28 Jin^eLD: ok, I try to be there Oct 04 13:06:30 Jin^eLD: check the wiki, it is about 10m from here Oct 04 13:06:31 CELF Oct 04 13:06:36 :-) Oct 04 13:06:45 zecke: here is where? :) Oct 04 13:06:54 Jin^eLD: somewhere in Berlin Oct 04 13:06:55 berlin Oct 04 13:07:04 ah, I guess I won't make it then Oct 04 13:07:10 zecke knows *exactly* where :-) Oct 04 13:07:13 :) Oct 04 13:07:20 zecke: any objections if i join you at OEDEM? Oct 04 13:07:37 henryk why no`t? Oct 04 13:08:00 ok, then count me in Oct 04 13:08:11 Henryk: sure, in the worst case we will search a bigger room :) Oct 04 13:08:43 * Bernardo is back. Oct 04 13:10:10 any open access points in Frankfurt airport? Oct 04 13:10:36 gilligan__: look carefully at the eraseall and nanwrite options -- I use "flash_eraseall -j" Oct 04 13:11:04 Crofton|home : You got by train from Frankfurt to Berlin ? Oct 04 13:11:10 s/got/go Oct 04 13:11:29 fly Oct 04 13:11:30 it is holiday season in the office and I have counted 21 seats Oct 04 13:11:45 cbrake, any specific reason why you split up the nand in two partitions? Oct 04 13:11:56 cbrake: I use flash_eraseblock without any options and it works fine Oct 04 13:12:20 nud: do you have jffs2 summary enabled? Oct 04 13:12:27 gilligan__: mount time Oct 04 13:12:32 mmh probably not Oct 04 13:12:47 koen: done Oct 04 13:12:49 cbrake, ah.. i see Oct 04 13:13:39 gilligan__: nud uses the cm-x270 :-) Oct 04 13:14:13 cbrake, I still get tons of errors about bad magic block Oct 04 13:14:16 nud: CONFIG_JFFS2_SUMMARY -- speeds up your mount time a little Oct 04 13:14:44 jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x0023001c: 0xdb3b instead Oct 04 13:14:47 tons of those Oct 04 13:15:03 I had something like that at the beginning Oct 04 13:15:10 gilligan__: you are running flash_eraseall -- emphasis on the *all* part ? Oct 04 13:15:13 the issue was a missing eraseall before the overwrite Oct 04 13:15:18 cbrake, yes Oct 04 13:15:56 gilligan__: I don't know then. you are using the summary jffs2 image? Oct 04 13:16:13 `flash_eraseall -j /dev/mtd3` Oct 04 13:16:26 another reason you can get those is if you haven't set up your erase block in the jffs2 command (with -e) . Oct 04 13:16:46 I have a nor and a nand and the eraseblocks are different. if I forget to change it, I'll get those to spit out like crazy Oct 04 13:16:47 cbrake, that one I can't even mount because it can't read the superblock Oct 04 13:17:15 oh, and the other thing is to make sure you select the "no cleanmarkers" since you are doing the nand eraseall and nand write.jffs2 Oct 04 13:18:57 well the erase size seems to be okay.. 16kB - that's what /proc/mtd lists as well Oct 04 13:19:16 what cm-x270 do you use, gilligan__ ? Oct 04 13:19:21 w or l ? Oct 04 13:19:26 nud, w Oct 04 13:19:33 ah, me too Oct 04 13:20:16 gilligan__: out of curiosity, do you plan to use the video input device ? Oct 04 13:20:48 nud, nope Oct 04 13:21:37 christopher, no clearmarkers ? where? Oct 04 13:26:20 I don't need to specify any of the options for nandwrite do I ? the jffs2 oob legacy support or something like that (which doesn't work anyway) Oct 04 13:26:21 cbrake: with "CONFIG_JFFS2_SUMMARY" you have to put the .jffs2.summary file somewhere ? Oct 04 13:26:44 gilligan__: nandwrite /dev/mtd3 theimage.jffs2 Oct 04 13:26:55 nud, yes that is what i am doing Oct 04 13:27:06 nud: the cm-x270 OE build both jffs2 and jffs-summary images Oct 04 13:27:32 cbrake: yes I know, was just asking what I had to do with the summary image Oct 04 13:27:33 nud: so, if you have SUMMARY enabled, then flash the summary image instead of the jffs2 image Oct 04 13:27:39 oh Oct 04 13:27:40 ok Oct 04 13:27:59 gilligan__: turn off CONFIG_JFFS2_SUMMARY in the kernel and give that a try. Oct 04 13:28:16 again the same ...`mount -t jffs2 /dev/mtdblock3 /mnt/nand` --> tons of errors Oct 04 13:28:52 gilligan__: actually, mount -t jffs mtd3 /mnt/nand is more correct, but probably does not matter Oct 04 13:29:11 cbrake, i see.. Oct 04 13:29:30 cbrake, and you are right, doesn't matter Oct 04 13:29:59 zecke : Any idea what's the room rates on Honigmond ? Oct 04 13:30:41 steliosk: no idea, I assume it is fully booked out Oct 04 13:31:06 steliosk: http://www.honigmond-berlin.de/assets/images/autogen/a_preise1.jpg Oct 04 13:31:20 hrm.. how can this work fine for nud&cbrake but not for me... grrr Oct 04 13:32:30 zecke : yeah seen that but 95 to 145 is quite a distance..... Do they have an online reservation page ? Oct 04 13:32:34 koen: could you send me your .tex files of your celf presentation? Oct 04 13:32:44 steliosk: I think they don't Oct 04 13:33:36 cbrake, is the info in /proc/mtd from what you've specified in the boot cmdline or is that gathered from hardware? Oct 04 13:33:44 zecke : gaestehaus-berlin-mitte seems to be full also... Oct 04 13:34:02 any other hotel near by ? Oct 04 13:34:42 gilligan__: both -- partitions will not show up if phyiscal devices are not detected Oct 04 13:34:50 steliosk: I can only say maps.google.com, enter novalisstrasse 10 and search for hotels nearby Oct 04 13:35:10 * steliosk gets a higl res google map photo, and searches for a bench to sleep Oct 04 13:35:27 steliosk: I'm sorry I need to get some food now, in the worst case bring your sleeping bag and we will find a place for you Oct 04 13:36:31 cbrake, but the erase size of 16kB should be fine ? http://www.linux-mtd.infradead.org/doc/jffs2.html suggests to use 64kB as default Oct 04 13:37:16 gilligan__: I would need to dig into this -- don't know offhand Oct 04 13:40:08 zecke : Thanks, i appreciate it. I am past sleeping bag age unfortunately :) (and its better to save the space for someone that else that might need it more). I'll try and see if i can get something in the vicinity Oct 04 13:40:32 steliosk: hehe, bbl Oct 04 13:40:33 *food* Oct 04 13:53:57 cbrake, hm.. just lost overview.. where is the call to create a jffs2 image for angstrom/compulab ? Oct 04 13:54:15 gilligan__: no cleanmarkers is a parameter for mkfs.jffs2. Oct 04 13:54:17 EXTRA_IMAGECMD_jffs2 = "--little-endian --eraseblock=0x20000 --pad=0x20000 --no-cleanmarkers;\ Oct 04 13:54:17 sumtool -i ${DEPLOY_DIR_IMAGE}/${IMAGE_NAME}.rootfs.jffs2 \ Oct 04 13:54:17 -o ${DEPLOY_DIR_IMAGE}/${IMAGE_NAME}.rootfs-summary.jffs2 \ Oct 04 13:54:17 -e 0x20000 -p; \ Oct 04 13:54:17 cd ${DEPLOY_DIR_IMAGE}; \ Oct 04 13:54:18 rm -f ${DEPLOY_DIR_IMAGE}/${IMAGE_LINK_NAME}.rootfs-summary.jffs2; \ Oct 04 13:54:20 ln -s ${IMAGE_NAME}.rootfs-summary.jffs2 ${IMAGE_LINK_NAME}.rootfs-summary.jffs2 \ Oct 04 13:54:22 " Oct 04 13:54:38 someone sent that to me a while back. it puts the sumarry in but also uses the no clean markers. Oct 04 13:54:50 which you need when using nand flash from what I've ready Oct 04 13:54:51 read Oct 04 13:55:49 christopher, that's exactly the stuff I was looking for.. in which file is that? Oct 04 13:56:01 gilligan__: conf/machine/compulab-pxa270.conf Oct 04 13:56:08 ah.. *doh* .. ok Oct 04 13:56:09 ;] Oct 04 13:56:44 gilligan__: did not we figure out once that --pad=0x20000 does not work, you want just --pad, or was that nud ... Oct 04 13:58:04 cbrake: I had the issue Oct 04 13:58:31 uh..yes.. wasn't that pad patched out recently ? Oct 04 13:58:56 christopher: I think you want --pad (drop 0x20000) Oct 04 14:00:14 -e, --eraseblock=SIZE Use erase block size SIZE (default: 64KiB) Oct 04 14:00:14 (usually 16KiB on NAND) Oct 04 14:00:16 ah, okay Oct 04 14:00:25 ^ from `sumtool --help` Oct 04 14:02:22 uhm... but 16KiB = 16384 = 0x4000 Oct 04 14:02:38 and not 0x20000 Oct 04 14:03:01 or is my brain on suspend right now? :) Oct 04 14:03:44 because sumtool has `-e 0x20000 ` Oct 04 14:05:41 cbrake? Oct 04 14:09:48 :) Oct 04 14:10:53 silence Oct 04 14:11:58 gilligan__: most nand flash has an erase block size of 128KiB Oct 04 14:12:51 gilligan__: but you could check the datasheet for the part on your module to make sure Oct 04 14:14:18 bye Oct 04 14:17:11 when I pad a jffs2 image I use -p with no space between the -p and padsize Oct 04 14:19:08 cbrake: yeah, whoops. that will pad too much. Oct 04 14:19:19 that's in the machine file Oct 04 14:19:22 cbrake, well I was just wondering because sumtool says 16kIB Oct 04 14:19:35 you can use -p or --pad= as they are the same Oct 04 14:20:14 gilligan__: what chip are you using ? Oct 04 14:20:21 scruggs: the issue is we were trying to pad out to the erase block size, so you want to use --pad with no options Oct 04 14:20:31 christopher, whatever is standard on the compulab.. i have no idea actually :) Oct 04 14:21:01 cbrake: thats what I figured, I've just never gotton it to work that way Oct 04 14:21:04 gilligan__: ok. hmmm. you may want to find that out so you can look and see on a data sheet. I am not familiar with that one, but maybe someone else is building for that Oct 04 14:21:10 scruggs: otherwise, mkfs.jffs2 things you are specifying the "image" size, and if you specify a the erase block size, the image is much larger and it does not do any padding Oct 04 14:21:39 scruggs: --pad with no size pads to the eraseblock size Oct 04 14:21:50 #EXTRA_IMAGECMD_jffs2="--eraseblock=0x20000 --no-cleanmarkers" Oct 04 14:21:57 that's what I used before putting the summary in Oct 04 14:22:01 ahh... got it Oct 04 14:22:18 christopher: that is fine as long as you don't use the .jffs2 image Oct 04 14:22:45 sorry, now I am lost..should I drop the 0x20000 from the EXTRA_IMAGECMD in compulab-pxa270_2.6.22.bb now or what? :) Oct 04 14:23:03 gilligan__: what matters is what the chip is, not what someone who wrote docs for sumtool thinks it might be :-) Oct 04 14:23:26 cbrake, right Oct 04 14:23:32 gilligan__: drop it for --pad, leave it for --eraseblock Oct 04 14:23:39 ok Oct 04 14:23:40 gilligan__: the version in OE is correct Oct 04 14:24:42 * cbrake thinks there is a tool that will print out mtd detailed info, but I can't get flash_info to do anything ... Oct 04 14:25:31 mtd_debug Oct 04 14:25:39 ? Oct 04 14:25:43 zecke: I was weak and did the celf one with apple keynote Oct 04 14:27:16 gilligan__: yes, that is it. Use the numbers that gives you: mtd_debug info /dev/mtd3 Oct 04 14:27:59 cbrake, 64KiB Oct 04 14:28:07 cbrake, 64KiB (erase size) Oct 04 14:28:18 gilligan__: interesting, my board is 128KiB Oct 04 14:28:26 gilligan__: must have different flash chips Oct 04 14:29:07 gilligan__: dmesg| grep NAND -- what mfg and size does it report? Oct 04 14:29:52 cbrake, ah sorry it is actually 16KiB for mtd3 (still not 128KiB though) Oct 04 14:30:26 zecke: http://scap.linuxtogo.org/files/288a84cc642629c94a5f153bafa35334.png Oct 04 14:30:42 zecke: I'll try to have a look later tonight at updating the webkit recipe Oct 04 14:30:43 cbrake, nothing at all for NAND Oct 04 14:31:25 cbrake, actually dmesg only prints lots of alignment trap messages and nothing else ;] Oct 04 14:32:15 gilligan__: well, look through your kernel boot log, and find where it is detecting the NAND flash -- should be something like: http://pastebin.ca/725539 Oct 04 14:32:26 gilligan__: echo 2 > /cpu/proc/alignment Oct 04 14:33:22 cbrake, there is nothing in /var/log/messages besides the trap messages Oct 04 14:33:51 but i can just reboot and check the output Oct 04 14:34:40 ah, there we go Oct 04 14:34:43 cbrake, NAND device: Manufacturer ID: 0xec, Chip ID: 0x79 (Samsung NAND 128MiB 3,3V 8-bit) Oct 04 14:35:18 gilligan__: well, try the different block size and let me know if it works -- mtd_debug should be telling you what is there Oct 04 14:35:41 cbrake, yep will do Oct 04 14:35:52 gilligan__: I thought for sure we are running 128MiB flash on some of our units, and have not had any problems ... Oct 04 14:36:10 hey all Oct 04 14:36:34 is there a way to change the hashing method bitbake uses to determined what .bb files have changed sinc it last ran? Oct 04 14:39:07 steliosk: ping Oct 04 14:39:23 koen : pong Oct 04 14:40:46 steliosk: did you say that you'll bring the stk1000 boards to oedem? Oct 04 14:42:15 koen : hmmm i am supposed to get them tomorrow, as dhl need it my signature, due to the high value of the shipment Oct 04 14:42:43 * steliosk totally forgot about them Oct 04 14:45:33 cbrake, now it works fine Oct 04 14:46:25 cbrake, so i guess maybe it would make sense to add a variable to the compulab conf ? which is either L or W .. which designates erase size of the nand and also designates the serial to either ttyS0 or ttyS1 ? Oct 04 14:47:23 gilligan__: good to know. flash size is not tied to L or W, so I think we need another var like CMX270_NAND_ERASEBLOCK_SIZE Oct 04 14:49:52 cbrake, ah right Oct 04 14:52:19 cbrake, what are the chances of getting write access to the oe repo? I'd gladly add that variable and the needed extra logic.. should be within my scope of understanding.. ;] Oct 04 14:52:45 scruggs: ping Oct 04 14:53:03 yet i am a newbie and I don't know how strict your guys policies are Oct 04 14:53:57 gilligan__: you probably need to talk to the OE core team Oct 04 14:54:49 sakoman: pong Oct 04 14:56:17 * gilligan__ waves at OE core team Oct 04 14:57:28 cbrake, whom can I bug about this ? no names given in the oe FAQ on that matter Oct 04 14:57:39 http://www.openembedded.org/people Oct 04 14:58:28 ah alright Oct 04 14:58:29 scruggs: looking at the ssh issue again. did you set up a root password? Oct 04 14:58:42 Serial can be set from the machine conf file Oct 04 14:59:07 sakoman: no, never did set it Oct 04 14:59:08 gilligan__: http://www.openembedded.org/wiki/MonotonePhraseBook Oct 04 14:59:15 scruggs: in my log file I see: Oct 4 06:00:47 sakoman-gumstix-custom authpriv.warn dropbear[544]: user 'root' has blank password, rejected Oct 04 14:59:38 funny that doesn't happen for you too! Oct 04 14:59:53 sakoman: if you have DISTRO_TYPE = "debug" dropbear allows empty passwords Oct 04 15:00:13 koen: you are a genius :-) Oct 04 15:00:19 sakoman: http://www.openembedded.org/repo/org.openembedded.dev/packages/dropbear/dropbear.inc Oct 04 15:00:39 now we need to understand why there is a minute delay prior to the password prompt! Oct 04 15:00:51 you pick up all sort of good tidbits of info hanging around here Oct 04 15:01:12 the patch isn't applied by default for obvious reasons :) Oct 04 15:01:31 scruggs: yeah -- I had turned off debug to see what effect it had on image size Oct 04 15:02:28 angstrom*image* (expect minimal) inject gdb, strace, etc into the rootfs in debug builds Oct 04 15:02:33 koen, mail sent Oct 04 15:03:04 sakoman: mine has dropped to a 3sec delay Oct 04 15:03:06 koen: yes, I noticed that. hadn't yet looked for other effects though Oct 04 15:03:25 scruggs: what did you do to reduce the delay? Oct 04 15:03:59 sakoman: which is fine if you have like 16megs of flash, but below 8 it gets in the way Oct 04 15:04:09 I think it was a firewall issue on my host Oct 04 15:04:21 that's why I like people like you coming up with stuff for smallish systems Oct 04 15:05:01 koen: I'm trying to squeeze my image down to about 6MB so I have 10MB free for a years worth of history Oct 04 15:05:40 * koen still wants to do a zigbee mesh around the house to collect temp and humidity data Oct 04 15:05:51 scruggs: hmmm . . . did you have the same issue with buildroot images? Oct 04 15:06:18 I did a tcpdump on my host as I was trying to ssh to the stix and noticed it doing lots of dns queries and it obviously wasn't getting a response Oct 04 15:06:34 cbrake, the current kernel command line setting should mount /dev/mtd3 as root if it can be mounted and only otherwise fallback to the rootfs in RAM, right ? Oct 04 15:06:46 koen: If the solar market ends up being too small I may branch out into other monitoring type apps Oct 04 15:07:01 no, not with br - I think it was just a fluke for me - I had my firewall down to connect to my work vpn Oct 04 15:07:36 scruggs: nothing like that on my end. got to put some thought into how to debug this delay Oct 04 15:07:43 I think if you tell dropbear not to do reverse lookups on connections it will go away? Oct 04 15:08:05 scruggs: any suggestions on how to do that? Oct 04 15:08:22 option in the build recipe? Oct 04 15:08:49 I had the same issue with my gentoo box having a large ssh delay... Oct 04 15:09:11 digging back to what I did then, but I think it was the lookup issue... need to check Oct 04 15:11:30 ok, on my gentoo box I added a 'UseDNS' line in sshd_config... but I'm not seeing a config file for dropbear immediately Oct 04 15:12:06 sakoman: is your dhcp server pushing any nameserver entries? Oct 04 15:12:38 yes Oct 04 15:14:31 hmm... google isn't being much help either Oct 04 15:14:43 cbrake, will only mount mtd3 as root when I remove the rdinit argument - is there some convenient way to choose booting with mtd3 as root or just the ramdisk ? (other than specifying the whole boot cmdline that is) Oct 04 15:15:40 scruggs: you are a genius too! The first name server is down right now! Oct 04 15:15:43 gilligan__: boot through the initramfs, and then to a switch_root to the main rootfs Oct 04 15:15:55 :) Oct 04 15:16:01 cbrake, ah.. ;] Oct 04 15:16:12 was about to ask if you could ping them Oct 04 15:16:39 gilligan__: this is how you implement sw update functionality :-) Oct 04 15:17:17 cbrake, yeah.. Oct 04 15:17:40 scruggs: I think my dsl modem picks the dns servers up via dhcp from the isp Oct 04 15:18:26 cbrake, some additional scripts for convenience, maybe auto-login and a script that executes switch_root after a 10sec delay or so Oct 04 15:18:48 cbrake, guess my boss should be happy with that Oct 04 15:18:51 cbrake ;] Oct 04 15:26:18 cbrake, that worked nicely Oct 04 15:27:44 gilligan__: another happy OE/cm-x270 user! -- make sure you send compulab a note :-) Oct 04 15:28:27 cbrake, sure thing Oct 04 15:28:42 cbrake, will do when i'm back at my office on monday Oct 04 15:40:24 totally unrelated, but.. *really* awesome... http://picozilla.com/files/135761l1_13.jpeg :) Oct 04 16:21:45 past security Oct 04 16:23:19 Crofton770: +31619260390 if you want to call me on friday to find us Oct 04 16:25:10 roa>phl soon Oct 04 16:25:18 thanks koen Oct 04 16:25:48 i will try and be at train station Oct 04 16:27:20 koen, i am 5404495437 Oct 04 16:27:44 iapologize for notknowing the countr code Oct 04 16:28:12 * Crofton770 needs to figure that out Oct 04 16:34:51 the monotone handbook has to be by far the best documentation of any SCM i've seen so far Oct 04 16:36:18 Thats not a tough award to win Oct 04 16:37:46 hehe Oct 04 16:38:48 Anyone know how to change how bitbake hashes files to determine if they haven't changed between runs? Oct 04 16:41:06 i have no idea about bitbake internals at all ;/ Oct 04 16:44:28 hmm,howlong to oedem from brazil Oct 04 16:45:35 Anybody know how to avoid the append of "lib" prefix in package names? Oct 04 16:47:45 Cause i have a hybrid package with libs and bins files... the lib package can have this prefix... but I don't want this prefix to -utils package Oct 04 16:47:48 Crofton770: long, since it's probably tap, which means 4h+ transfer in lisboa Oct 04 16:48:28 aloisiojr: DEBIAN_NOAUTONAME_ = "1" Oct 04 16:48:28 tap? Oct 04 16:48:47 the operator Oct 04 16:48:49 koen, many tks Oct 04 16:49:08 ah Oct 04 16:50:40 once they had free wireless in phl Oct 04 16:51:21 Crofton770: your country code should be +1 Oct 04 16:52:21 so +15404495437 Oct 04 16:52:56 Crofton770: yes Oct 04 16:53:30 thanks Oct 04 16:53:55 message received Oct 04 16:54:00 * koen concludes pycurses is underdocumented Oct 04 16:54:06 Crofton770: :) Oct 04 16:54:31 * Crofton770 wonders how much that will cost Oct 04 16:54:45 gota run boarding Oct 04 16:54:47 in dollars or real money? Oct 04 16:55:08 heh bye Oct 04 16:55:20 bye, enjoy that coach service! Oct 04 16:56:06 with my 1.82m I routinely run into problems Oct 04 16:56:15 why is all travel designed by midgets? Oct 04 17:00:09 reading the openembedded manual i find it a bit odd that the chapter describing how to build a package recipe precedes the chapters on getting and setting up OE/bitbake ;] Oct 04 17:03:02 * koen removes some coffee from his keyboard Oct 04 17:05:16 what's the idea behind bitbake collections ? the section in the oe handbook is only a stub .. Oct 04 17:05:46 "overlay" in gentoo terms Oct 04 17:06:09 sorry, never used gentoo and hence not portage :] Oct 04 17:06:37 koen, well the stub is sort of self explanatory.. Oct 04 17:07:02 koen, just don't know about those priority values.. ? Oct 04 17:07:25 koen, if files overlap the one from the set with higher priority is picked i guess? Oct 04 17:07:36 yes Oct 04 17:07:41 ok Oct 04 17:07:41 ;] Oct 04 17:09:11 Anybody know why the .so symlink is included in lib package instead of the -dev package? you don't need the .so link in runtime... Oct 04 17:10:47 XorA|gone: ping Oct 04 17:24:37 Tartarus: Why would you want to do that? What problem do you want to solve? Oct 04 17:39:26 I'm reading the bitbake manual. It says that a .bb file sets its PN, but I'm looking at the gnash .bb file which doesn't. Oct 04 17:39:38 Does PN (and PV) get set from the .bb filename by default? Oct 04 17:41:07 (this is the perfect question to ask, because you can't tell me to RTFM ... since you already know I'm doing that. :) Oct 04 17:43:12 I would suppose that information is given implicitly through the .bb basename.. but i'm anything but an expert (yet...:) Oct 04 17:45:28 ahhh, it says that in Example 4.7. Ahhh, the dangers of writing dox which serve as both reference and tutorial. Oct 04 17:48:10 Hmmmm..... that makes me wonder about writing combined reference and tutorial documentation from the same docbook file. #if reference and #if tutorial sections, so to speak. Oct 04 18:00:26 bahh, LTIB requires sudo to run :-\ Oct 04 18:05:53 Someone in the channel had told me that they ordered an MPC8223E-RDB but i forgot who it was :( Oct 04 18:17:58 gilligan_: hey Oct 04 18:18:05 steliosk: the other name of the hotel was mercure Oct 04 18:18:13 hey zecke Oct 04 18:18:19 steliosk: did you find a hotel yet? Oct 04 18:32:03 hmm.. nothing like a corrputed ext3 disk Oct 04 18:33:04 ljp: hey Oct 04 18:33:05 * ljp grumbles and starts over Oct 04 18:33:24 het zecke: you going to dev days? Oct 04 18:34:41 ljp: yupp Oct 04 18:35:01 hopefully I will not get "I'm a Troll, ask me" T-Shirt Oct 04 18:38:31 you dont have to wear thatif you dont want to.. Oct 04 18:39:30 zecke : i found 2. But need to see what date/time i will arrive, as there is a high possibility i have to fly to Paris on the 9th... Oct 04 18:39:50 steliosk: don't emulate that you are busy ;) Oct 04 18:40:49 zecke : heh the offered me a hotel and 2 French girls.... Oct 04 18:41:41 steliosk: iirc it was likewise that ordered the mpc8223 Oct 04 18:42:51 what was the nick of Michael Krelin again? Oct 04 18:43:09 zecke: polyonymous|note Oct 04 18:43:16 polyonymous|note: hey :) Oct 04 18:43:24 :) Oct 04 18:43:25 polyonymous|note: I forgot "hash" in the parenthesis Oct 04 18:43:25 hey Oct 04 18:43:44 uhm... lemme take another look at the expression :) Oct 04 18:43:52 polyonymous|note: the usage of sha1 in git is only a checksum Oct 04 18:44:04 koen|away : Do you know if he will be at OEDEM ? Oct 04 18:44:16 zecke, and? Why would you want to remember it? Oct 04 18:44:23 polyonymous|note: so when linus leaves home, he remembers the hash Oct 04 18:44:34 what for? Oct 04 18:44:42 polyonymous|note: when he clones the tree from somewhere or get backs home he can compare it Oct 04 18:44:56 polyonymous|note: if it is the same hash, his tree is right, otherwise it was corrupted Oct 04 18:45:05 steliosk: not according to http://www.openembedded.org/wiki/OEDEM2007 Oct 04 18:45:20 polyonymous|note: so git's usage of sha1 is for checksumming, security comes from gpg signing a hash Oct 04 18:45:34 zecke, got it. Oct 04 18:46:09 well, there are signable tags Oct 04 18:46:10 polyonymous|note: security is built-in into monotone (and a reason why it is slow...) Oct 04 18:46:19 polyonymous|note: right, but we tag seldomly in OE :) Oct 04 18:46:50 right ;-) Do you really think this level of security is needed? Oct 04 18:47:21 polyonymous|note: to some degree I'm paranoid :) Oct 04 18:47:32 polyonymous|note: and now we have it, I appreciate it Oct 04 18:47:51 zecke, if only monotone slowness was the only problem... Oct 04 18:48:09 polyonymous|note: let us msg :) Oct 04 18:48:36 zecke, I'll have to change nick and identify myself with server :) Oct 04 18:48:41 polyonymous|note: you don't Oct 04 18:48:51 polyonymous-note: I have set unfiltered on at NickServ Oct 04 18:48:57 zecke : found 2 hotels 1) mitArt Linienstr. 139 - 140, 10115 Berlin Oct 04 18:49:05 too late :) Oct 04 18:49:43 BTW: Every one coming by train, check your schedule! Bahn is on strike Oct 04 18:50:59 oh man..... There are strikes in Germany ? You spend too much time in Greece guys... Oct 04 18:58:38 zecke: drat, the ICE I'm on is on strike (amsterdam -> berlin) Oct 04 18:59:00 koen: 2/3 of all trains should travel Oct 04 18:59:18 yes, I'm on the 1/3 Oct 04 18:59:29 you suck Oct 04 19:00:16 :) Oct 04 19:00:31 oh Oct 04 19:00:58 btw. how much is ticket for ice amsterdam-berlin? Oct 04 19:01:28 first classs hengelo -> berlin is €60 Oct 04 19:01:54 nice rice Oct 04 19:01:59 price even Oct 04 19:02:54 koen: yeah, some how the fares in .NL are lower than in germany :} Oct 04 19:03:04 koen: or they forget to pay the "Extra Fee" for the ICE Oct 04 19:03:10 * steliosk wonders if he will be also on the 1/3 Oct 04 19:03:23 or it's strike price :p Oct 04 19:03:33 koen: let us create our company in Amsterdam Oct 04 19:03:40 gerwinin: hey! Oct 04 19:03:47 hey zeck Oct 04 19:03:47 gerwinin: everything set? Oct 04 19:03:52 Yes Oct 04 19:03:55 amsterdam sucks Oct 04 19:04:09 Koen: As a real eindhovenaar I can only confirm this Oct 04 19:04:25 Zecke: do you need a place to sleep over ? Oct 04 19:04:59 Koen: How is the budget from trolltech spent ? Oct 04 19:05:08 gerwinin: not yet spent :) Oct 04 19:05:23 gerwinin: I have no idea, I haven't received *any* answer on my questions Oct 04 19:05:31 gerwinin: that would rock, do you have a big house? you already host ~5 people? Oct 04 19:05:52 Zecke: Last year I had more than 14 people in my house Oct 04 19:06:04 hehe Oct 04 19:06:25 Koen/Zecke: We will be present on some other fair in holland as well Oct 04 19:06:25 hrw|gone: koen: I was at my mother and picked up covers and such for you ;) Oct 04 19:06:50 gerwinin: I'm going to bitch Mr. Engel to finish his fs... Oct 04 19:07:15 zecke: okay :) Oct 04 19:07:19 haha Oct 04 19:07:32 zecke: I started working on a gui for oe Oct 04 19:07:43 zecke: yeah, pull joerns goatee if he resists Oct 04 19:08:43 gerwinin: I saw we have a slot on Sunday morning Oct 04 19:08:49 zecke: yes Oct 04 19:10:16 zecke: I originally asked for a slot a bit later because sunday morning is not the most strong morning for most people living in Eindhoven :) Oct 04 19:11:01 gerwinin: yeah, you seem to go out on Saturday and not on Friday... Oct 04 19:11:51 zecke: I mostly go on friday because of fc-eindhoven Oct 04 19:12:44 gerwinin: hehe, I'm with PSV Eindhoven because my uncle was a striker there.. (IIRC) Oct 04 19:13:22 zecke: hehehe was he called freyther as well ? Oct 04 19:13:51 zecke: we are already playing bottom of the first division now for three years Oct 04 19:14:05 hi all Oct 04 19:14:09 long time no see gerwinin Oct 04 19:14:15 gerwinin: Johannes Roehrig (but somehow he is called Fred) Oct 04 19:14:20 Hey likewise Oct 04 19:15:49 ah Oct 04 19:16:24 hengelo -> zutphen -> arnhem -> duisburg -> hanover -> berlin Oct 04 19:16:34 only 4 hours more traveling time Oct 04 19:16:41 zecke: in the time your uncle was playing for psv we were better than psv but that is long long time ago Oct 04 19:16:43 * koen packs an extra book Oct 04 19:16:59 Koen: Are you in the train now ? Oct 04 19:17:04 no Oct 04 19:17:08 * koen is still at home Oct 04 19:22:22 koen: how much of the nokia 800 kernel/system is open? Oct 04 19:22:35 * gerwinin is fighting with qt to get something decent Oct 04 19:23:02 hi ljp Oct 04 19:23:06 hi Oct 04 19:23:08 ljp: well, the schematics are on the web :-) Oct 04 19:23:23 ljp: Where can I go with my questions regarding qt-eclipse plugin seems that it is hanging on autocompletion Oct 04 19:23:45 just wondering what parts Trolltech can put pressure on Nokia to open, so OE can build Oct 04 19:24:05 gerwinin: either #qt, or qt-interest mailing list Oct 04 19:24:34 gerwinin: do you have a place to sleep in berlin already? Oct 04 19:24:41 hi Oct 04 19:24:47 ljp: wifi module is closed Oct 04 19:24:54 ljp: and Nokia cannot open it Oct 04 19:25:14 zecke: I am not sure I am comming doing some project with imx27 now which is progressing really slowly and needs to be finished next week Oct 04 19:25:37 anyone know where to find lennert buytenhek on irc these days? Oct 04 19:25:52 zecke: Besides that I need to prepare our comming presence at t-dose and nlgg conference Oct 04 19:25:56 hi hrw Oct 04 19:27:27 zecke: I am going to opensource my hardware designs + driver as well and need to meet up with my lawyer for which project this is all possible Oct 04 19:27:56 gerwinin: you have to show/explain me your current day job Oct 04 19:28:15 hrw: is that all? besides hildon stuff Oct 04 19:29:16 zecke: I am a hardware/software designer , I developed some simple usb devices , Next to that I am renting myself out to other companies Oct 04 19:29:21 ljp: there are few more things but I do not know do I am allowed to talk about it Oct 04 19:29:37 ljp: ask RP - he know more Oct 04 19:32:59 zecke: hehehe but in software I am not as good as you or likewise :) Oct 04 19:34:30 steliosk: what was the street name again? Oct 04 19:34:47 zecke: I sadly can't bring any vla now because the stores are still closed at my new departing time Oct 04 19:35:44 koen oh no Oct 04 19:35:44 koen: you could get a frikandel at the trainstation and use a USB lamp to keep it warm? Oct 04 19:36:07 steliosk: oh there is a hotel? Oct 04 19:36:18 steliosk: no idea, it could be loud Oct 04 19:36:45 vla? Oct 04 19:36:55 zecke, are you talking to yourself, or have I managed to put steliosk on ignore? :) Oct 04 19:37:01 hrw: some sort of pudding, but don't name it like this... Oct 04 19:37:21 polyonymous-note: you might call it that way, I say sith force (and count on the backlog...) Oct 04 19:37:32 vla is comparable to the english 'custard' Oct 04 19:38:14 ah uh um. ok. Oct 04 19:38:36 hrw -> http://en.wikipedia.org/wiki/Vla Oct 04 19:39:21 woglinde, is it opensource? :) Oct 04 19:39:25 ~curse git for crap merge Oct 04 19:39:26 May you be reincarnated as a Windows XP administrator, git for crap merge ! Oct 04 19:39:38 zecke : found 2 hotels 1) mitArt hotel Linienstr. 139 - 140, 10115 Berlin Oct 04 19:39:40 polyomous lol Oct 04 19:39:59 night everyone Oct 04 19:40:09 nite steliosk Oct 04 19:41:11 steliosk: ahh, 1) is a nice place, it should be fine Oct 04 19:45:11 I hate git Oct 04 19:45:19 fatal: Untracked working tree file 'Documentation/ko_KR/HOWTO' would be overwritten by merge. Oct 04 19:45:22 Merge with strategy recursive failed. Oct 04 19:45:49 even "git reset --hard" does not help Oct 04 19:45:55 hrw, perhaps, because untracked working tree file would be overwriten by merge? Oct 04 19:46:00 hrw, git clean ? Oct 04 19:47:26 maybe one day git will start merge util.... Oct 04 19:47:47 hrw: no, you will need to use git-mergetool file/conflicted yourself Oct 04 19:47:50 instead of adding cvs conflict markers... Oct 04 19:48:09 right. Oct 04 19:49:29 git pull;git-mergetool onefile;git-mergetool 2ndfile;....; git add onefile 2ndfile ...; argh Oct 04 19:49:33 hrw: I just sent a mail to oe-dev, it has my cell number :) Oct 04 19:49:46 argh == rm -rf tree;git clone Oct 04 19:50:00 hrw: git-rebase --abort? Oct 04 19:50:20 polyonymous-note: one --abort too many killed all my commits :} Oct 04 19:51:04 zecke, I don't think it's possible at all in git. All commits are preserved until you --prune. Oct 04 19:52:17 Warning: No merge candidate found because value of config option "branch.libertas.merge" does not match any remote branch fetched. Oct 04 19:52:21 No changes. Oct 04 19:52:28 this is my favorite git message Oct 04 19:52:37 crappy as hell Oct 04 19:53:06 hrw, but what's crappy about it? Oct 04 19:53:13 hehe Oct 04 19:53:39 hrw, you don't have a [branch] with remote/head set to merge, why do you pull? Oct 04 19:54:13 polyonymous_: I like Andrew Mortons introduction to git in the google techtalk Oct 04 19:54:47 1154 modified files... time to drop tree and reclone Oct 04 19:54:52 * chouimat hates git Oct 04 19:54:57 zecke, was it in that famous video? If yes, I don't remember it, if not, I don't know it :) Oct 04 19:55:10 why hate something you just don't know? Oct 04 19:55:48 polyonymous_ because the humane thing todo .. hate and fear what you don't know ;) Oct 04 19:56:05 chouimat, can't argue with that :) Oct 04 19:56:09 that doesn't sound so humane to me Oct 04 19:56:14 polyonymous_: something about Linus creating a tool that makes you feel dumb Oct 04 19:56:46 polyonymous_: I used rcs, cvs, svn, bk, monotone with many projects. getting git working atleast as good as cvs looks like beyond my possibilities Oct 04 19:56:48 zecke, ah :) Well, I have to admit Linus failed miserably on this account in my opinion :) Oct 04 19:57:18 lots of people use GIT and are very very happy with it Oct 04 19:57:22 hrw, I'm sure it's not. It's just that you have a prejudice. Oct 04 19:57:27 git isn't that hard to use. Oct 04 19:57:31 polyonymous_: and every time I tried it I had bad experience and lost a few days of work so ... I prefer to use Hg, It's not perfect but it's easy to use and work for my needs Oct 04 19:57:34 the failure isn't on Linus's part Oct 04 19:57:49 kergoth, so, am I that badly behind that apache learnt to start hooks? Oct 04 19:58:16 CosmicPenguin, the failure to make me feel dumb? :) Oct 04 19:58:37 polyonymous_: i don't know how it does it, but it does. i push over webdav and it updates the metadata on the remote end, i do it all the time. Oct 04 19:58:41 kergoth: hehe Oct 04 19:58:43 chouimat, in git it's practically *impossible* to lose work, it doesn't delete anything on it's own accord. Oct 04 19:59:09 polyonymous_: as long as simple "scm fetch/merge/update" and "scm diff" works without problems I can use it. But I still did not passed to get this nice working with git Oct 04 19:59:13 kergoth, I don't believe you, maybe it's just that it's not always necessary to update metadata. Oct 04 19:59:27 then why does it say "Updating remote server info Oct 04 19:59:27 " on every push to the remote repo? Oct 04 19:59:33 is it just trying to fool me? Oct 04 19:59:36 somehow i doubt it Oct 04 20:00:08 kergoth, no idea. But surely no hook is running on remote side, maybe it's doing it over webdav from client, I don't know. I don't really use dumb protocols. Oct 04 20:00:15 on angstrom and gpe, is there a way to start X and matchbox without gpe-dm so that it will do it automatically? Oct 04 20:00:19 kergoth, but surely can't start hooks on remote. Oct 04 20:00:31 christopher: xserver-nodm is what you need Oct 04 20:00:35 maybe it is, but maybe you should actually test before you spout nonsense about my point about webdav being useless Oct 04 20:00:42 the fact is, it works. Oct 04 20:00:43 hrw, it works fine, but I think you at least confusing pull and fetch :) Oct 04 20:00:46 hrw: ok, thanks, I will try that. will that load up matchbox too? Oct 04 20:00:55 christopher: yes Oct 04 20:01:08 polyonymous_: fetch == fetch; pull == fetch+merge Oct 04 20:01:35 hrw: sweet! that will be exactly what I need. Thanks. oh, and does it export DISPLAY too? I know some programs need DISPLAY set to :0 in order to work Oct 04 20:01:36 hrw, I'm not saying you don't know the difference, it's just that your complaint made me thing you're confusing them when you're working. Oct 04 20:02:45 polyonymous_: I use olpc kernel tree on my laptop to test 8385cf card. from time to time I 'git pull' to get updates. each time I do 'quilt pop -a' before pull to get rid of possible conflicts Oct 04 20:02:48 kergoth, Okay, we both say "nonsense": you - about hooks running on remote side via webdav and me (maybe) - about inability to update metadata via webdav. The difference is that you still prefer smart protocol :) Oct 04 20:03:09 polyonymous_: and I do not use git to track my changes - tats what for I have quilt Oct 04 20:03:27 hrw, meaning you haven't learnt to resolve conflicts :) Oct 04 20:03:47 ERROR: No providers of build target xserver-nodm (for []) Oct 04 20:03:47 christopher@linuxserver:/oe/build$ Oct 04 20:04:05 christopher: I do not remember how recipe is called Oct 04 20:04:30 polyonymous_: git cheatshit tells to use "git add file1 file2; git rebase --continue" Oct 04 20:04:39 hrw: ok, I'll look around then Oct 04 20:05:00 grrrrr stupid computer ... /me wonders if problem is kernel or X related Oct 04 20:05:02 hrw: git is starting to work pretty well as a quilt replacement for me -- at least for kernel work Oct 04 20:05:07 polyonymous_: I think that I will keep two trees instead and limit git usage to pull Oct 04 20:05:21 hrw, `git add` after you've resolved conflict :) Oct 04 20:05:26 polyonymous_: ofcourse Oct 04 20:05:35 hrw, and? it doesn't work? Oct 04 20:06:02 hrw: what version of git are you running? Oct 04 20:06:20 polyonymous_: I ended with >1000 modified files (according to 'git status') which I did not touched. so I prefer to start from scratch Oct 04 20:06:23 cbrake: 1.5.3.3 Oct 04 20:06:37 hrw: should be new enough Oct 04 20:07:07 hrw, what does it have to do with files *you* touched? You're doing merge, remember? Oct 04 20:08:00 polyonymous_: I do not plan to go back to that tree to check what else was wrong Oct 04 20:08:37 hrw, having files touched by others in your tree when doing merge is not "wrong". But I'm not trying to make you go back to that tree :) Oct 04 20:10:46 polyonymous_: found the commit in question. http://www.ohloh.net/projects/278/contributors/13268/commits/13136174 Oct 04 20:11:08 so as you say, it doesnt call the hook, but it does update the dumb protocol metadata Oct 04 20:11:14 so webdav is still a viable option Oct 04 20:11:52 kergoth`bbl, agreed on that. But personally, I'm afraid of updating repository via webdav :) Oct 04 20:12:25 kergoth`bbl, so, it turns out we were both right and both wrong :) Oct 04 20:13:49 indeed. my description of why it works was wrong, but the original point was valid. i like using webdav in some ways, apache2 gives a lot of flexibility on authentication Oct 04 20:14:04 the only downside is the odd behavior of the locking i ran into a couple times over the past few weeks Oct 04 20:14:20 seems to only happen on an interrupted push, should see if i can reproduce it reliably Oct 04 20:15:22 kergoth`bbl, and I agreed with your original point that theoretically push over webdav is possible ;-) But honestly, would you want it in OE? (not that git is an option, hrw would kill me if I will even *think* it is:)) Oct 04 20:16:12 polyonymous_: well, we have an apache running, I'm paranoid so apache2+ldap would be nice for user management (I don't want to give logins....) Oct 04 20:16:16 i wouldn't mind it being used for OE, but what i would mind is yet another scm switch. it's just too much trouble to be worthwhile for minimal gain Oct 04 20:17:15 zecke, that's what I mentioned as the first drawback of git, but webdav doesn't sound fun to me even if it works to some certain extent. Oct 04 20:17:16 we can postpone the switch, but decide on what we need now Oct 04 20:17:28 polyonymous_: *if* OE will change SCM then it will be git. there is no sense in changing to something else I think. Oct 04 20:17:41 kergoth`bbl, I don't think it's a minimal gain, I don't think it's too much trouble and I don't think we should switch either :) Oct 04 20:18:10 hrw, I'm glad to hear it. And I agree with likewise - it's something to think of, not rush into. Oct 04 20:18:11 * hrw finally builds 2.6.23-rc9-hrt0-cfs22 Oct 04 20:18:40 hrw: it was xserver-nodm-init and it does work and pulls up x...so, if I want it to pull up matchbox at the same time, should I just modify the Xsession script or is there something else I would need for that? Oct 04 20:19:12 hrw: cool, but why not -rt instead of -hrt? Oct 04 20:19:25 polyonymous_: cvs, svn, svk are not good. bk is not free anymore. mercurial/bazaar/darcs are not common Oct 04 20:19:39 likewise: hrt enable hpet on my ich4 laptop Oct 04 20:20:14 hrw, neither is monotone, but we use it ;-) I think being common is not of the highest importance (but not unimportant either, of course). Oct 04 20:20:23 likewise: iirc -rt changes more then I want Oct 04 20:20:39 hrw: true :-) Oct 04 20:20:54 I don't know much about mercurial/bazaar/darcs, but last time I looked at darcs I didn't like it. mercurial and bazaar left me indifferent, but my knowledge here is rather limited. Oct 04 20:20:58 polyonymous_: we are one of biggest projects using monotone and some stuff was made due to our requests Oct 04 20:21:11 woglinde: BTW: I finally have a DEC keyboard! Oct 04 20:21:19 zecke NO! Oct 04 20:21:24 that cant be Oct 04 20:21:24 mtn has definately come a long way Oct 04 20:21:36 woglinde: yeah, it is yours, your wife sold it to me Oct 04 20:21:42 no Oct 04 20:21:46 hrw, I know. That's exactly an example of why being common isn't *that* important. Oct 04 20:21:50 I am typing at it right now Oct 04 20:21:53 woglinde: let me make a picture Oct 04 20:21:57 woglinde: it is just a sticker ;) Oct 04 20:21:59 woglinde, it's an illusion. Oct 04 20:22:09 yeah, mtn works well enough. interesting though we don't see much branching in OE, but perhaps it is more the nature of the project Oct 04 20:22:18 zecke: what kind of DEC-keyboard? Oct 04 20:22:34 oxo: from digital, plain ps2 keyboard Oct 04 20:22:39 cbrake, we don't see much branching, but we do have heckofalot of merges, does it make sense? :) Oct 04 20:22:40 oh, a new one Oct 04 20:22:52 * oxo is having a HUGE DEC/apollo yardsale soon Oct 04 20:23:13 zecke german? Oct 04 20:23:24 polyonymous_: good point Oct 04 20:24:18 woglinde: http://www.openembedded.org/~zecke/digital.jpg Oct 04 20:24:27 cbrake, that's one of the things I don't like in monotone, it provokes such a repository history. but it's surely not a reason to rush in for another SCM Oct 04 20:25:49 polyonymous_: how would it differ if I would use git for OE but do not use own branch at all? Oct 04 20:26:16 polyonymous_: my stuff would appear as merge anyway just like it is today? Oct 04 20:26:18 polyonymous_: its interesting that in git, I branch stuff all the time -- list like I'm supposed to, but I've never made a branch in mtn to do some private work. I need to learn how to use mtn better -- surely it can do many of the same things ... Oct 04 20:26:22 zecke: interested in some 'older' DEC stuff? Oct 04 20:26:37 * mwester perks up his ears at the mention of "branching" and "OE" in the same sentence, without the word "no" in the middle. Oct 04 20:26:40 :P Oct 04 20:26:41 hrw, with git you can (a) rebase instead of merge your local changes and (b) not push if it would create two-head. Oct 04 20:26:48 oxo: well, I wonder what my boss would say if we get a PDP11 Oct 04 20:27:00 How does a rebase differ from a change-set-based merge? Oct 04 20:27:16 mwester: linearisation of history Oct 04 20:27:33 * oxo is getting rid of a lot of PDP11/VAX/apollo systems Oct 04 20:27:34 polyonymous_: rebase move my commits to top of 'HEAD' - rigt? Oct 04 20:27:36 zecke U see Oct 04 20:27:38 Ok, but it still amounts to the same thing in terms of code motion. Oct 04 20:27:45 ups I see I meant Oct 04 20:27:51 cbrake, branches in monotone aren't made *convenient*. That is it doesn't mandate such a history (almost), but provokes it. Oct 04 20:27:54 hrw, yup Oct 04 20:28:03 hrw, on top of whatever upstream you push into Oct 04 20:28:03 zecke so which one did you murder Oct 04 20:28:20 polyonymous_: push or pull? Oct 04 20:28:21 mwester, are you saying that you will end up with the same code? :) Oct 04 20:28:21 woglinde: suddenly it was in the office, and placed it under my desk ;) Oct 04 20:28:31 hrw, what push and pull? Oct 04 20:28:33 woglinde: now one month has passed and I claim it Oct 04 20:28:49 polyonymous_: It would be desirable if you ended up with the same code, but I somehow doubt it! Oct 04 20:28:56 hrw, I'd say `git fetch oe && git rebase oe && git push oe` - something along these lines. Oct 04 20:28:58 polyonymous_: I do commit; then pull; then rebase - so my commit is on top of what I pulled. Oct 04 20:29:11 hrw, right. Oct 04 20:29:15 hrw, oh no Oct 04 20:29:28 hrw, if you commit and *pull* - then you have merge on top of your commit. Oct 04 20:29:43 hrw: branch first? Oct 04 20:29:43 hrw, but if you fetch and rebase you have linear history Oct 04 20:29:55 The _theory_ of change sets is a nice idea, but it never quite works out that way, which is why I intensely dislike SCM tools that (wrongly) encourage developers to think in terms of patches and change sets. Oct 04 20:29:59 so commit, fetch, rebase (which will also merge) and push? Oct 04 20:30:14 cbrake, of course you do branch at some extent - like after fetch you have refs/remotes/oe/master and you checkout -b master oe/master Oct 04 20:30:26 haha, sorry that you are talking about git ;) Oct 04 20:30:28 hrw, what do you mean "rebase which will also merge" ? Oct 04 20:30:52 polyonymous_: well, you could git-reset HEAD^1 and then rebase on the remote? Oct 04 20:31:15 zecke, and? Oct 04 20:31:26 zecke, I mean, I'm not sure what are you talking about :) Oct 04 20:31:27 polyonymous_: I have week old tree. I do commit in it. then I fetch (which will give me 500 changesets from upstream). when I do merging of that 500 csets? Oct 04 20:31:32 A rebase sounds to me like a reverse-merge --- one merges the parent's deltas back to the child branch so as to "resync" the child. It's still a merge! Oct 04 20:32:07 polyonymous_: commit/fetch/rebase/push looks like 'lets ignore those 500 csets' Oct 04 20:32:08 hrw, rebase is not "merge" - you will have your commits on top of what you've pulled, therefore there's no single purely merge revision. Oct 04 20:32:25 * mwester thinks this discussion needs a large whiteboard and set of colored markers... :) Oct 04 20:32:33 hrw, no, it's not. you will have all your local commits in history *after* your 500 revisions Oct 04 20:32:36 not your Oct 04 20:32:45 mwester, yeah, right :) Oct 04 20:32:56 mwester, I think git rebase --help actually has it in ascii-art :) Oct 04 20:33:14 hrw: I branch before committing anything, that way pulls happen on the master branch without messing with my branch, and then I rebase the branch Oct 04 20:33:25 mwester, why is it still a merge if you don't have a single merge revision? Oct 04 20:33:35 polyonymous_: so it will extract my commits to temporary something, remove them, merge upstream 500 csets and then apply my stuff on top - right? Oct 04 20:34:06 If I have modifications on my branch, and I rebase, then my changes must be merged with the changes on the parent. Oct 04 20:34:20 hrw, technically it will not either extract or remove them, it will apply them on top of upstream 500 (right) and leave them unreachable in your objects directory Oct 04 20:34:37 mwester, in terms of resulting code - yes, but not in terms of repository history Oct 04 20:34:55 polyonymous_: ok. but when it will happen? during 'git rebase'? Oct 04 20:35:01 hrw, yes Oct 04 20:35:18 Oh, well yes -- but is it the code or the history that is important? Oct 04 20:35:20 ok, tx Oct 04 20:35:23 hrw, actually, due to the lack of whiteboard I'd really recommend `git rebase --help` Oct 04 20:35:28 hrw: hi Oct 04 20:35:37 hrw: small issue Oct 04 20:35:48 mwester, readable and usable history is important if you want to make real branches and throw code back and forth. Oct 04 20:36:04 I am new, following the wiki I tried to chkout mtn and got this: mtn: misuse: layout of database /home/niv/devel/OE.mtn doesn't match this version of monotone Oct 04 20:36:12 I use monotone 0.35 and when I try to mtn --db=$PWD/OE.mtn db migrate I got mtn: error: database schema 48fd5d84f1e5a949ca093e87e5ac558da6e5956d is unknown Oct 04 20:36:22 Agreed. But if the code is merged (rebased) incorrectly, then the history is at best meaningless, and at worst, false. Oct 04 20:36:27 upgrade monotone Oct 04 20:36:53 niv_one_three: or download a snapshot for your mtn version Oct 04 20:36:55 mwester, incorrect code is always a problem, but why would it be rebased/merged incorrectly? :) Oct 04 20:36:59 thanks, I use monotone 0.35 aint it the latest? Oct 04 20:37:00 niv_one_three: upgrade monotone, or download a older snapshot Oct 04 20:37:13 niv_one_three: 0.36 is out, but it should have the same schema Oct 04 20:37:36 hi zecke Oct 04 20:37:50 koen|away: could you send me your .tex files of your CELF talk? Oct 04 20:38:01 any idea what I could present at FOSS.IN? Oct 04 20:38:38 polyonymous_: Example: on the parent you clean up a function by removing unused variables. On the branch, I have implemented a loop using the previously-unused variable. The merge/rebase will result in "Undefined variable" on my branch. Oct 04 20:38:44 ok, I got a simpad , and managed to flash the angstron bin fine Oct 04 20:39:09 I want to install the wlan-ng to use my prism 2 wifi card Oct 04 20:39:18 mwester, this problem would common to both rebase and merge approaches. Oct 04 20:39:27 That's exactly my point. Oct 04 20:39:32 mwester, I'm not saying scm will write code for you :) Oct 04 20:39:41 to do this I need to compile the bin for it on my pc, so I want to create the angstrom env Oct 04 20:39:43 There is only a terminology difference between rebase and merge. Oct 04 20:39:49 mwester, well, we're talking about repository history, provided it's meaningful at all :) Oct 04 20:40:14 how to download a OE.mtn for monoton 0.35 ? Oct 04 20:40:44 mwester, no, not terminology. Suppose you want to cherrypick commit from the other branch which comes from italy and bears unnecessary resemblance with spaghetti? Would it be fun to cherrypick changes from such a branch? Oct 04 20:41:26 oh ok - I see it in http://www.openembedded.org/snapshots/ Oct 04 20:41:27 It would be no different than merging back from the mainline to the branch. The commands used would be different, but the results would be the same. Oct 04 20:41:37 niv_one_three: http://www.openembedded.org/snapshots/ Oct 04 20:41:45 niv_one_three: hehe Oct 04 20:42:21 (And, no, it's never fun to merge someone elses mess -- that's why the developers who create the branch must own the task of merging it back, IMO) Oct 04 20:42:23 mwester, it would take ages to figure out which changes are relevant for you. Oct 04 20:42:36 zecke : pls tell me what should I do to build wlan-ng for my wifi card to run on the angstrom ? Oct 04 20:42:37 mwester I agree Oct 04 20:43:06 niv_one_three: no idea :) Oct 04 20:43:21 mwester, in fact developer must be responsible for making his branch mergeable. Oct 04 20:43:33 niv_one_three: bitbake wlan-ng? Oct 04 20:43:35 polyonymous_: that's a process problem, not a tool issue. Even the simplest tool can organize relevant changes, if the developers were disciplined and put major features on feature branches to begin with. Oct 04 20:44:30 niv_one_three: how you connected that card? Oct 04 20:44:35 mwester, well, tool choice does have implication. Of course no tool can solve workflow problems, only assist. Oct 04 20:45:04 In any case, I'll support any SCM tool in use (although I reserve the right to complain about horrible mis-implementations, such as MTN's performance) -- I just think that we (OE devs) need to adopt a branching and merging set of rules. Oct 04 20:45:50 I think we can do better in terms of OE stability if we would use SCCS, as long as we would branch!!! Oct 04 20:45:53 * mwester hides Oct 04 20:46:02 mwester: There is no issue with creating branches. Its the maintenance of then that worries people. That is true regardless of the SCM Oct 04 20:46:14 RP: You are exactly correct! Oct 04 20:46:35 hrw: the card is connected via pcmcia Oct 04 20:46:48 of course, SCM is no panacea. But it surely affects performance. Oct 04 20:46:59 IMO, OE has matured to the point that the devs need to commit themselves to maintenance (and closing out) of branches. Oct 04 20:47:36 mwester: I already maintain one and I personally can't spare any more time for that Oct 04 20:47:39 niv_one_three: then you need hostap or orinoco rather then wlan-ng Oct 04 20:47:44 zecke: I assume you mean I need to run "bitbake wlan-ng" just get the source.. right? how to compile? Oct 04 20:47:49 Of course everyone WANTS to work on the mainline, but (so to speak) the kitchen here has gotten very crowded, and the cooks are getting in each others' way too often. Oct 04 20:47:50 zecke: do you have a bio about yourself ? Oct 04 20:48:36 mwester: care to take care of a branch? A stable one? Oct 04 20:48:37 gerwinin: "Junky, on crack, looking worse than wookie" is probably not okay? Oct 04 20:48:37 hrw: thanks , so I need to run bitbake hostap - then how to get the .ipk? Oct 04 20:48:51 zecke: for the intro it is :-) Oct 04 20:49:37 zecke: :) Oct 04 20:49:48 gerwinin: grab the one from aKademy2006? Oct 04 20:50:06 Zecke: okay Oct 04 20:50:08 mwester: The real problem is that a stable branch will only ever be stable for the aims the people mantaining it have. Poky is rock solid for arm. I'd bet powerpc builds explode badly... Oct 04 20:50:27 Zecke: Can you bring an openmoko telephone ? Oct 04 20:50:43 mwester: Its safe to assume nobody deliberately makes .dev unstable, the shear volume of changes just make it so Oct 04 20:51:03 niv_one_three: under 2.6 both are in kernel Oct 04 20:51:05 RP: I agree. Oct 04 20:51:43 RP: No cook in the kitchen is to blame, but regardless, the soup spills often... Oct 04 20:51:56 gerwinin: yes, it will either run Qtopia or OpenMoko (or probably both) Oct 04 20:52:01 mwester: So we need better cleaners? :) Oct 04 20:52:09 :D That's a solution! Oct 04 20:52:10 hrw: so why can't I find it in /lib/`uname -r` ? Oct 04 20:52:35 niv_one_three: no one enabled them in kernel config? Oct 04 20:53:12 RP: speaking of kernel config -- did you had a look at zaurus configs which I sent to you? Oct 04 20:53:33 hrw: Only breifly :-( Oct 04 20:53:48 We can issue the honorary title "Chief Sanitary Engineer" to a different person every week, and put them in charge of fixing build problems! :D Oct 04 20:54:23 hrw: After OEDEM I should have slightly more free time hopefully... Oct 04 20:54:42 not in the bin provided in http://www.angstrom-distribution.org/unstable/images/simpad/20070831/ Oct 04 20:55:58 hrw: I guess I need to build my own kernel for the simpad , right Oct 04 20:56:24 yes Oct 04 20:56:44 can I use bitbake to get a .jffs2 file? Oct 04 20:57:00 yes Oct 04 20:57:12 ynezz: please tell me how? Oct 04 20:57:19 hrw: When do you arrive at OEDEM btw? Oct 04 20:57:50 RP: saturday Oct 04 20:58:23 hrw: ok, same as me effetively... Oct 04 20:58:26 RP: will be ~13-14 at oedem Oct 04 20:58:36 niv_one_three: read the docs, IMAGE_FSTYPES = "jffs2" Oct 04 20:59:31 thanks ynezz Oct 04 21:00:10 17 min to download OE.mtn.bz2 .. Oct 04 21:00:50 I just hope I can understand the devel env, and manage to add the bluetooth hack to me simpad Oct 04 21:00:58 any simpad user here? Oct 04 21:01:41 niv yes Oct 04 21:04:05 woglinde: do you use angstrom? Oct 04 21:08:31 niv yes Irun successfull angstrom with opie images with uclibc Oct 04 21:08:44 but it can be broken now Oct 04 21:09:01 but it worked 2 mothns ago Oct 04 21:11:26 bye all Oct 04 21:11:30 see you at oedem Oct 04 21:11:41 bye hrw Oct 04 21:12:15 see you hrw Oct 04 21:24:38 dam it I d/l the 0.35 OE.mtn and still can't chkout Oct 04 21:26:25 ok, got my mistake seems I got 0.31 installed as well Oct 04 21:29:05 mtn --version Oct 04 21:29:37 you really want the later ones as they are speedier. Oct 04 21:57:49 hi khem Oct 04 21:58:27 hello likewise Oct 04 21:58:35 I have been away on vacation Oct 04 21:58:53 khem: did you enjoy it? Oct 04 21:59:03 yes Oct 04 21:59:10 I took no comps with me :) Oct 04 21:59:17 khem: good Oct 04 21:59:38 yeah I felt it was a good one Oct 04 21:59:49 khem: where did you go? Oct 04 21:59:51 otherwise I ended up working more on vacation Oct 04 22:00:04 I was in India :) my homeland Oct 04 22:00:15 I try to take it more easy as well lately -- otherwise works eats me Oct 04 22:00:20 in the Himalayan ranges :) Oct 04 22:00:37 khem: wow. would love to go there once. Oct 04 22:01:18 yeah it's a different world altogether Oct 04 22:01:22 khem: have you been there before? Oct 04 22:01:41 its you and the mountain goats (ibex) Oct 04 22:01:45 yes Oct 04 22:02:16 I live in the Netherlands which is flat as hell. Oct 04 22:02:42 its you, flatness and cows, over here :-) Oct 04 22:02:53 and a whole lot of crowd Oct 04 22:03:55 yeah I have been to Amsterdam in 2004 Oct 04 22:04:07 and also to Den Haag Oct 04 22:46:49 hey Oct 05 00:01:12 scruggs: I'm looking at the scp issue. Are you libc or glibc? Oct 05 00:01:42 hi gang. how do you adjust the patch -p N argument in a bb file? i see what looks like "strippath" in patch.bbclass, but don't find this in any recipe. Oct 05 00:02:45 sakoman: uclibc Oct 05 00:03:51 OK, same here. Just wanted to see whether the issue was the same for libc and uclibc. guess we don't have that bit of data :-) Oct 05 00:04:50 I did notice the latest dropbear version in oe is 49 and in the br its 48 Oct 05 00:05:14 I was going to change the 'preferred' dropbear to see if that made any difference Oct 05 00:15:28 sakoman: did you ever figure out what was pulling in glibc and causing the QA sanity failure in your build? Oct 05 00:16:51 after digging - it's pnum=patch_level. Oct 05 00:22:19 scruggs: no, though I have determined that I get the sanity error for libstdc++ even when it *isn't* in my install! Oct 05 00:22:28 go figure Oct 05 00:23:00 need to get a good chunk of time to debug. past few days have just had a few minutes here & there Oct 05 00:23:40 off to dinner now. ttyl Oct 05 00:25:04 ttyl Oct 05 02:03:42 Henryk: so do you get up early, or stay up realy late :-) Oct 05 02:03:57 Henryk: I'll push your recipe in the morning ... Oct 05 02:05:40 both, kind of. I was staying up late for the last week or so (e.g. go to bed at 8am, wake up at 3pm) but am now trying to get that into 'normal' times for OEDEM at the weekend. Oct 05 02:05:45 cbrake_away: thanks Oct 05 02:40:14 I'm confused. Oct 05 02:41:24 automake seems to generate code that makes an empty directory simply if include files *might* be distributed, however bitbake looks for the existance of a subdirectory and if found, attempts to copy files even if none are found. **** ENDING LOGGING AT Fri Oct 05 02:59:57 2007