**** BEGIN LOGGING AT Mon Jan 25 02:59:57 2010 Jan 25 07:38:19 good morning Jan 25 07:44:38 khem: ping? Jan 25 07:57:42 hm, is there any official documentation for opkg.conf? I wonder if I could force opkg to fetch package files to a persistent directory Jan 25 08:14:33 filip: did you try google? http://wiki.openwrt.org/oldwiki/openwrtdocs/packages Jan 25 08:15:44 mckoan: yup, I found "option cache" :) Jan 25 08:18:37 oldwiki times out on me, but nvm Jan 25 08:57:20 khem: ping again? Jan 25 09:30:54 is there a known problem with mknod and fakeroot in recipes ? I'm trying to create some static devices nodes in /dev but it seems it's not working ( I followed the tutorial at: http://docs.openembedded.org/usermanual/html/fakeroot.html ) Jan 25 09:32:00 I also tried a recipe par of openembedded (ezxd) and it's not working, devices aren't block devices in data.tar.gz Jan 25 09:33:03 pix_: no idea, for the image creation you should use the device table to create nodes Jan 25 09:36:28 zecke: thanks for pointing me to IMAGE_DEVICE_TABLE Jan 25 09:51:32 zecke: ah, right, that aclocal patch. yeah, I should check that in. Jan 25 09:52:47 pb__: thanks. Jan 25 09:52:56 pb__: we are now down to < 100 patches in the queue Jan 25 09:53:22 pb__: but I'm trying to form a opinion on patch flow. Stuff like bugzilla queues vs. mail/patchwork vs. gitorious merge requests Jan 25 09:54:29 zecke: yeah, I have been reading (some of) the mails on that subject over the past few days Jan 25 09:54:48 personally I would be happiest with bugzilla, and second happiest with patches on the mailing list (approximately as now) Jan 25 09:55:23 pb__: but you have not seen gitorious (neither did I from the upstream side) Jan 25 09:55:24 pb__: I think the best option is an integration in git itself Jan 25 09:55:30 I'm working on that Jan 25 09:56:10 zecke: indeed, I haven't seen gitorious. but, that seems a bit of a high hurdle to make every casual contributor jump over. Jan 25 09:56:36 I can see that gitorious pull request would make sense for large patch series, or people who are submitting a lot, but then those folks tend to have direct checkin access to the oe master tree anyway. Jan 25 09:57:30 I am not very convinced, though, that asking the guy who just wants to change one line in a recipe to set up a gitorious branch so that it can be merged is a very good use of time. Jan 25 09:57:50 pb__: well, he would have a local commit anyway? Jan 25 09:58:14 pb__: and git push might be easier than git format-patch and get the mailer do the right thing with that? Jan 25 10:03:00 zecke: he probably will have a local commit, yeah (assuming he is using git locally, though I guess most folks will be) Jan 25 10:03:33 but he might well not have a gitorious account, and he might resent needing to set one up just to send a patch to oe. I know I would. Jan 25 10:04:34 pb__: I'm thinking about a solution where setting up an account is not necessary Jan 25 10:04:44 pb__: what if making a merge request in gitorious is just like git push gitorious-server-oe please-merge? Jan 25 10:04:54 morning all Jan 25 10:05:07 pb__: if you are in the position of a small one line patch, would you be interested in cleaning it up to be mergable Jan 25 10:05:19 ah, fair enough. if that can be made to work then I guess it would be fine. Jan 25 10:05:30 maximum efficiency for both committer and contributor (but some possibility for lower quality contributions) Jan 25 10:05:30 zecke: well, one would hope that a one-line patch would not require very much cleaning up. Jan 25 10:05:51 pb__: I know it was talked about, I don't know if it is already implemented. Jan 25 10:06:56 righto Jan 25 10:07:39 zecke: I think you will likely want to have more than one branch. Jan 25 10:07:59 Laibsch: the "please-merge" would create a merge request branch and all the stuff Jan 25 10:08:51 zecke: I'd be interested to have a play with that stuff if you can get it to work. Jan 25 10:09:29 pb__: would you create a bugzilla account to post a patch? Jan 25 10:09:41 all that said, though, I would have thought that "git format-patch" and then pressing the "attach" button in your email program would be very nearly as easy for most submitters, and has the advantage that it is a process they are probably already familiar with (since 99% of open source projects work that way) Jan 25 10:10:04 zecke: no, for the same reason, but there are bugzilla tools to allow you to not have to do that. Jan 25 10:10:24 pb__: 20% of the 300 patches had formatting issues (line wrapping) Jan 25 10:11:18 yeah, I guess people must be accidentally inlining the patches into their emails rather than attaching them. I am not sure why that keeps happening. Jan 25 10:11:33 I guess it is possible that there is some wiki page out there which is giving bad advice. Jan 25 10:11:49 pb__: With bugzilla the workflow would be. Upload patches to a specific component, set the review flag, discuss it in bugzilla. But to work for one time contributions it must be login free? Jan 25 10:11:55 heck, I'd do C&P Jan 25 10:12:12 zecke: yes, I think a login-free interface is desirable for casual contributors. Jan 25 10:12:32 pb__: I think we insist on the bad instructions (asking people to use inline mails to ease review) Jan 25 10:12:42 so, if you send email to oe-patches@, it automatically opens a bug and then attaches the patch to it. Jan 25 10:13:00 zecke: ah, in that case we should stop doing that. all decent MUAs allow you to display text attachments inline anyway. Jan 25 10:13:28 pb__: ah you still would do a mail? I would have a script (maybe stolen from webkit) in the contrib folder to create a bug and upload commits? Jan 25 10:13:31 it is admittedly slightly irritating when people attach a file as application/binary and you have to save it before you can view it, but that doesn't happen often and it is less annoying than getting the lines wrapped by mistake. Jan 25 10:14:18 zecke: I guess a script would work as well, but again I kind of feel that submitting by email is an interface that everybody will already be familiar with, and the fewer "weird" tools we expect them to use the better. Jan 25 10:15:08 pb__: hehe, I wonder of how much of our habbit play with that? E.g. modern people like web forums too? Jan 25 10:15:36 So far I think going back to bugzilla for the review and using the bugzilla queue is the workflow we can have Jan 25 10:15:53 (witness, for example, the fact that people seem to prefer "git format-patch" and then sending an email, rather than just using git send-email) Jan 25 10:15:55 specially for recipes that need work and would just be forgotten about in patchwork. Jan 25 10:16:33 zecke: heh, true, these kids Jan 25 10:17:24 pb__: hmm, I feel a bit like hacking patchwork to automatically create bug reports and add comments Jan 25 10:18:00 pb__: so our review queue would be something like this: http://webkit.org/pending-review? Jan 25 10:18:39 zecke: why not the other way round? Jan 25 10:19:18 set up a mailing list oe-patches@ or oe-contributions@, have that gated to bugzilla which sends out mails to the list on ticket creation? Jan 25 10:19:56 Laibsch: Right. For now I want to compare patchwork vs. bugzilla workflow Jan 25 10:19:58 but I feel some people may object to bugzilla spam Jan 25 10:20:21 Laibsch: so having access to both seemed like a good idea Jan 25 10:20:30 ~seen kergoth Jan 25 10:20:32 kergoth was last seen on IRC in channel #edev, 2d 7h 59m 7s ago, saying: 'i presume he wants it ported to his board/device, which has an arm processor.'. Jan 25 10:36:13 morning Jan 25 10:38:54 hi hrw Jan 25 10:39:26 hrw did you by any chance see my email on my packaged staging experiment ? Jan 25 10:39:38 eFfeM_work: on ML? Jan 25 10:39:50 eFfeM_work: I have ~70 emails from oe ml to view Jan 25 10:40:40 hrw on OE ML, mailed last sat, title = packaged staging fails in do_deploy for kernel Jan 25 10:41:46 hrw: we still need to find a conclusion for your svn problem with the fetcher? Jan 25 10:43:15 zecke: I forgot about it Jan 25 10:43:28 zecke: I did not got hit by that problem in last months Jan 25 10:43:47 manz: ping Jan 25 10:44:32 mckoan: how goes xinput_calibrator? Jan 25 10:45:10 hrw: it's a difficult work, harder than I thought Jan 25 10:45:43 hrw: looks like in xorg community a few people are interested to touchsreen issues Jan 25 10:46:41 hrw: it would be nice to have a solution. So putting in -r and @ will fix it? Jan 25 10:46:56 zecke: I have to dig out mail and read my patch Jan 25 10:47:28 is it compulsory to register the IDENTIFY command (/msg nickserv help) to speak in oe channel? manz is not able to chat Jan 25 10:47:47 hrw: http://patchwork.openembedded.org/patch/1334/ Jan 25 10:48:11 mckoan: apparently there was some flooding on freenode.. I would be happy if this would be turned off.. again Jan 25 10:48:28 pb__: do you happen to have channel admin rights? Jan 25 10:49:53 I believe it's a network wide change. Jan 25 10:50:51 broonie: #beagle do not have +R Jan 25 10:52:12 we don't have it on #webkit, it was turned on during the flood.. and is off again Jan 25 10:52:34 e.g. the beloved CIA-* is not able to post commits Jan 25 10:53:39 zecke: sorry, no, I think you need kergoth for that Jan 25 10:53:50 mckoan: what's your problem with xinput_calibrator? Jan 25 10:54:43 ynezz: I was talking with ias about that Jan 25 10:54:54 s/ias/Tias Jan 25 10:55:50 ynezz: xinput_calibrator is working but settings are ignored Jan 25 10:56:04 ynezz: also xinput settings are ignored Jan 25 10:56:20 hrw: did you test xinput_calibrator there ? Jan 25 10:57:59 mckoan: ok Jan 25 10:58:56 mckoan: and I don't think, that xorg community don't care about touchscreens, they do Jan 25 11:01:05 ynezz: I didn't say that, I said "a few people" in fact Tias was very kind and supportive (like you was), and I also sent him a patch for xinput_calibrator Jan 25 11:05:18 re Jan 25 11:05:30 mckoan: my Xserver is old a bit Jan 25 11:07:06 hrw: I thought you was using the same version with your Angstrom+AT91sam9263EK Jan 25 11:07:46 mckoan: at91sam9263ek do not require calibration Jan 25 11:09:29 hrw: if you change the display it does ;-) Jan 25 11:09:46 mckoan: but I do not change it Jan 25 11:09:55 hrw: :-D Jan 25 11:13:40 mckoan: just curious, what kind of touchscreen it's and what driver for X do you use? Jan 25 11:14:12 mckoan: nm, I found the thread on xorg-dev Jan 25 11:16:17 mckoan: what's the outcome? you switch to evdev/evtouc driver and use xinput_calibrator? Jan 25 11:16:30 s/evtouc/evtouch/ Jan 25 11:17:59 mckoan, ynezz: which XInputExtension version you have on boards? Jan 25 11:20:20 hrw: I don't know offhand, but if it's something with X server version, I use latest .dev Jan 25 11:22:58 I use stable/2009 Jan 25 11:24:14 eh, chromium is slooow on a sa1110. that's weird Jan 25 11:24:32 ah, and it doesn't work Jan 25 11:25:08 Slow on sa1110... that doesn't seem like a surprise. Jan 25 11:25:09 chromium? that shooter? Jan 25 11:25:27 broonie: it seems slower than midori. opposite than x86 Jan 25 11:25:33 hrw: nope, chromium browser Jan 25 11:25:43 hrw, ynezz: Package: xinput Version: 1.5.0-r0.5 Jan 25 11:25:53 Package: xinput-calibrator Version: 0.5.0-r1.5 Jan 25 11:26:11 X.Org X Server 1.7.3 Jan 25 11:26:46 RP: are you still around? Jan 25 11:29:04 pb__: http://patchwork.openembedded.org/patch/959/. what do you think about that? Jan 25 11:34:12 -> bbl Jan 25 11:34:12 parted suxx Jan 25 11:34:31 zecke: yes Jan 25 11:36:40 RP: do you know if this is still current? http://patchwork.openembedded.org/patch/932/? Jan 25 11:36:48 the patch still applies Jan 25 11:37:18 zecke: yes, looks sane Jan 25 11:40:55 okay, I will push it then Jan 25 11:42:57 zecke: sounds good, thanks Jan 25 11:43:22 zecke: You seem to be doing well on your current mission, nice work! Jan 25 12:04:38 RP: I just hate when my patches don't get applied... at django I had to wait ~2 years Jan 25 12:07:42 it's same in OE :) Jan 25 12:07:46 or was Jan 25 12:08:02 not two years Jan 25 12:08:09 yes Jan 25 12:08:26 ynezz: the oldest patch I saw this time was from march Jan 25 12:08:42 yes, on patchwork Jan 25 12:21:45 good morning Jan 25 12:23:10 gm Jan 25 12:23:17 hi Jan 25 12:32:50 I'm trying to set the root password and looked at the zap_root_password code. I more or less copied (appended) the commands into ROOTFS_POSTPROCESS_COMMAND (in my image.bb) but it doesn't update the /etc/passwd. It looks like my commands are never being executed or they fail to execute. I don't get any errors. Is this the preferred way of doing it? Jan 25 12:35:28 zecke: I'm doing my best with bitbake Jan 25 12:35:53 zecke: Having one target should help a lot. I have half your parser patches applied to master now Jan 25 12:36:03 Just finding time for the other half Jan 25 12:36:10 and I fixed a few bugs :) Jan 25 12:53:51 Hi. Just a quick questiong regarding python within bb-file. If I define a new python function, then "bb.data.expand('${D}/usr/lib', d)" apparently is not a valid function. Why is that, when it works perfectly in "python do_install"? Jan 25 12:57:26 Hasse_: Hi Hasse Jan 25 13:05:18 -> re Jan 25 13:23:08 RP: oh, which bugs? I like to hear about errors Jan 25 13:46:10 re Jan 25 13:53:26 wb hrw Jan 25 13:53:40 rkirti: how goes maemo stuff? Jan 25 13:55:17 hrw: hildon-desktop and stuff are still the old ones Jan 25 13:55:21 havent looked at them yet Jan 25 13:55:45 my local OE copy has newer .bb files for the rest of the libs Jan 25 13:56:12 * rkirti should put them on github Jan 25 14:00:54 hrw: You have been developing on maemo recently, anything specific (ie. paritcular app) that you are working on ? Jan 25 14:03:54 rkirti: only my player Jan 25 14:07:54 :-) Jan 25 14:08:14 so your n900 has the same fate as your n800 - music player :-P Jan 25 14:17:39 rkirti: not quite - n900 is my primary phone Jan 25 14:17:57 I rarely use it as music player Jan 25 14:24:40 ok Jan 25 14:24:40 * jovox just did the best modification of his computer ever: Removed the Caps Lock key from the keyboard! Jan 25 14:25:00 jovox: ? Jan 25 14:25:21 I highly recommend :-) Jan 25 14:25:28 I hate that key and never uses it Jan 25 14:25:37 * rkirti thinks the CapsLock-Esc switch is the best modcification a ViMmer can get Jan 25 14:25:48 jovox: I swap Esc and Caps - very useful in ViM Jan 25 14:29:36 rkirti: I would rather map Control to Caps - just like it was in many machines before Jan 25 14:31:44 hrw: use of that being ? though I agree, given the number of times I press in Ctrl- in general apps. but if Esc isnt swapped with caps, it gets very cumbersome while coding, to switch modes Jan 25 14:34:37 rkirti: my first computer (atari 65xe) had control there, my first internet terminal had [control][caps][a] order Jan 25 14:37:52 zecke: Small stuff, you broke require vs include (as far as I could tell) and there was a __body__ that needed to become "body" Jan 25 14:40:52 RP: broke in a way that include became required or breaking required becoming include? Jan 25 14:41:07 RP: did you ever came across a LRU cache impl in python? Jan 25 14:41:13 come.. Jan 25 14:42:51 hi, what should I do about pngcrush which uses plain Makefiles? Jan 25 14:42:55 it says: Jan 25 14:43:25 | install: cannot stat `pngcrush-native': No such file or directory Jan 25 14:43:39 ah ok maybe I get it Jan 25 14:43:41 typo Jan 25 14:44:07 install PN Jan 25 14:44:12 in pngcrush Jan 25 14:44:19 and pn in -native is -native Jan 25 14:44:30 use BPN Jan 25 14:45:21 zecke: http://git.openembedded.org/cgit.cgi/bitbake/commit/?id=824a032672c8673cae28e492aa916a4750b94efa Jan 25 14:45:34 zecke: no, I've not seen one Jan 25 14:45:37 ok I'll look thanks Jan 25 14:45:47 I often have the problem when I make -static recipes Jan 25 14:48:24 ok thanks a lot Jan 25 15:25:54 morning Jan 25 15:32:28 morning Jan 25 15:36:07 does OpenEmbedded not support shadow passwords? Jan 25 15:38:18 there is a 'shadow' recipe, does that not count? Jan 25 15:45:15 kergoth: hey! there are some of your patches in the queue. It would be cool if we could land them... I need to run now though Jan 25 15:45:34 kergoth, I guess there's more to it than that Jan 25 15:45:39 i think most were just pending acks, i don't even recall what was out there, i'll take a look Jan 25 15:46:14 cia seems stuck again Jan 25 15:50:27 morning Jan 25 15:57:45 why is udev recipe so messy? Jan 25 15:58:06 heh, may as well ask that about half the recipes in the tree :) Jan 25 15:58:44 well udev is particularly scary Jan 25 16:03:43 * hrw -> off Jan 25 17:00:41 hrmph Jan 25 17:01:59 we need to find and kill cases where any task but do_stage/do_populate_staging modify staging contents. Jan 25 17:02:13 kergoth: oh yes Jan 25 17:02:26 kergoth: none in poky (as far as I know) Jan 25 17:02:26 the python build puts a config/Makefile in staging in do_compile_append, but not in do_stage Jan 25 17:02:36 a build of python-django against a pstage package of python goes boom :) Jan 25 17:02:40 kergoth: ick Jan 25 17:02:42 yes Jan 25 17:02:45 indeed Jan 25 17:03:18 * kergoth thinks its pretty disgusting to shove things into staging in do_compile, whatever the reasons Jan 25 17:06:03 kergoth: The sooner we make any recipe doing the break, the better... Jan 25 17:06:11 s/the/that/ Jan 25 17:06:42 agreed Jan 25 17:07:14 * kergoth works around it and adds the do_compile badness to the todo Jan 25 17:13:21 Is there a way to get a package that includes kernel headers / source (so I can build a kernel module outside of OE environment)? The -dev package of the kernel only seems to include the symbol table.... Jan 25 17:13:58 i don't think that exists yet.. it probably should. for now you can point the kernel module build to staging Jan 25 17:14:08 unless you really need it insid ethe target filesystem? Jan 25 17:15:05 No. Have a SOM that comes with a pre-build kernel for customers to play with. They'll likely wan to write modules, so I need a way to redistribute. Would like to avoid them having to learn all of OE.... Jan 25 17:16:25 guess a hand-rolled tar-ball will have to do.... Jan 25 17:19:08 ah Jan 25 17:19:33 well, you could always add it to the recipe yourself :) Jan 25 17:20:32 I'm trying to upgrade libpcre to 8.01 and I'm getting the following error message: ERROR: QA Issue with libpcre: non -dev package contains symlink .so: Jan 25 17:20:39 does anyone know how to fix this? Jan 25 17:20:43 Yes. Plan is to extend our meta-sdk-toolchain to add kernel source as well as crosstools. Just checkin if there was an easier way.... Thanks. Jan 25 17:21:23 as far as i know not, though i could certainly be unaware of it. i haven't seen anything of the sort in the kernel classes, anyway :) Jan 25 17:21:46 * kergoth thinks we should package those files anyway to do native development on the target device Jan 25 17:21:53 * kergoth thinks we don't do enough testing of that Jan 25 17:22:55 Crofton|work, looks like i got nixed for fosdem this year. sad... so sad. I thought about funding it myself, but w/ the little one too hard to justify the time and cost. Jan 25 17:23:16 kgilmer: ah, that's too bad Jan 25 17:24:13 * kergoth was hoping to go to elc this year at least, but has to pay a bunch of back taxes from his ohand contract from 06, cause he's a moron, instead :) Jan 25 17:24:51 yeah pb__ i'm bummed Jan 25 17:24:56 :( Jan 25 17:25:07 kgilmer, I keep checking fares Jan 25 17:25:14 but I will have serious work then also Jan 25 17:25:16 re Jan 25 17:25:48 kergoth: how to make in local.conf PREF_PROV_virtual/kernel for machine? Jan 25 17:26:13 i don't understand the question :) Jan 25 17:26:35 kergoth: in local.conf I want to set other virtual/kernel then it is set in machine config Jan 25 17:26:40 ah Jan 25 17:26:51 PREFERRED_PROVIDER_virtual/kernel_local = should do Jan 25 17:26:58 or any other override, really Jan 25 17:27:11 * kergoth thinks that machine should probably be using ?= there though Jan 25 17:27:21 sure Jan 25 17:27:49 kergoth: doh, how do you come to still have arrears from '06? Jan 25 17:28:19 your IRS must be more forgiving than the UK revenue service :-} Jan 25 17:28:21 I was an idiot and didn't send in the estimated tax payments for the contract, since it wasn't automatically taken out (my first contract), and then kept putting off dealing with it Jan 25 17:28:30 i'm sure the penalties and interest will be painful Jan 25 17:29:21 the interest+penalties to the state of arizona was about half of the original amount i owed for the state taxes.. i imagine the federal tax will be as bad or wrose Jan 25 17:29:26 heh Jan 25 17:29:32 * kergoth learned his lesson :( Jan 25 17:30:12 doh Jan 25 17:31:27 anyone building gstreamer? oe-dev pull from friday seems to not build complaining that gstreamer-plugins-base-0.10.22 is out of date Jan 25 17:32:35 kergoth: other then that you are fine? Jan 25 18:02:56 what's the "proper" way to supply a different fstab, without modifying base-files_xx.bb? can it be done in the image recipe? Jan 25 18:04:28 pb__: any comments for the checksums discussion? :) Jan 25 18:13:32 hi pb Jan 25 18:14:02 .quit Jan 25 18:36:03 RP: any thoughts on making the cooker accessible to the metadata, as a means of accessing the preference information in the current task data, or, alternatively, just making the taskdata object available? Jan 25 18:53:51 pb__: hi Jan 25 18:54:19 pb__: are you ok with disabling target-optspace for ppc ? Jan 25 18:54:54 I dont want to hack gcc configury because it will PITA to have it for every version of gcc moving forward Jan 25 19:42:21 morning Jan 25 19:42:37 yo rhw Jan 25 19:42:39 err Jan 25 19:42:40 morning hrw Jan 25 19:43:17 hi mickeyl Jan 25 19:43:20 mickeyl: fosdem? Jan 25 19:43:26 hey hrw, mickeyl Jan 25 19:44:20 hrw: still didn't book, still planning to come though Jan 25 19:44:22 yo kergoth Jan 25 19:44:57 usb wifi suxx Jan 25 19:45:55 yep Jan 25 19:46:40 I used cf, minipci, sdio, spi based wifi - all worked fine Jan 25 19:46:50 beagleboard accepts usb ;( Jan 25 20:06:48 in I need to import string in python Jan 25 20:06:56 which ipk do I need to install? Jan 25 20:07:33 for what do you need the old string module? Jan 25 20:07:43 gnuradio needs it to build Jan 25 20:07:47 urghs Jan 25 20:07:49 stringold then :) Jan 25 20:07:52 yep Jan 25 20:07:54 thanks Jan 25 20:07:59 almost all functions these days are living in the string object these days Jan 25 20:08:01 tell them to update Jan 25 20:08:02 :) Jan 25 20:08:10 I have list my mind and am going to see if gnuradio will build native on the beagle Jan 25 20:08:12 will do Jan 25 20:08:16 sounds cool Jan 25 20:08:22 sounds nutty Jan 25 21:02:21 khem: sounds ok to me Jan 25 21:06:24 pb__: cool. I dont think many ppc's are so space concerned like some of the arms Jan 25 21:06:39 and plus it will only be libgcc and libstdc++ Jan 25 21:06:58 few extra kbs Jan 25 21:37:57 RP: can you remind me, is your release plan now to drop 1.10 and merge it and master with the actual client/server split being conditional? am i remembering that correctly? :) Jan 25 21:53:46 woglinde, the boost-dev package is looking for a boost library and won't install Jan 25 21:54:04 I am goign to try using the boost package and see what happens Jan 25 21:54:07 aeh? Jan 25 21:57:33 woglinde, try builing an image that includes boost-devel Jan 25 22:26:09 How is the easiest way to get something done at first boot? Jan 25 22:27:43 otavio - put a line in /etc/inittab Jan 25 22:28:01 robtow: at _first_ boot ;-) Jan 25 22:28:14 Does anyone know/remember what version of python bitbake 1.8.9 works with ? i am getting some weird [Errno 39] Directory not empty errors Jan 25 22:28:33 otavio hm opie packages has it Jan 25 22:29:01 woglinde: opie. Let me look Jan 25 22:29:06 woglinde: remember which? Jan 25 22:29:26 hm no Jan 25 22:29:35 but other packages has it too Jan 25 22:29:44 like dropbear Jan 25 22:29:50 woglinde: ahh; indeed Jan 25 22:29:51 maybee look at it Jan 25 22:32:00 otavio - have that one liner modify itself :-p Jan 25 22:34:04 otavio: well, if its both first boot and at install, obviously a postinst would do, otherwise i expect an init script w/ a file touched to indicate an operation already complete would be the way to go, don't think there's a generic mechanism, but i could be wrong Jan 25 22:34:33 kergoth: yes; I was looking for a generic way of doing it Jan 25 22:34:38 * kergoth nods Jan 25 22:34:43 kergoth: it would be nice to be supported Jan 25 22:35:33 could do something like a /etc/firstboot.d/ .. for each script there, run it, then remove it upon successful execution.. but of course that would require rw / :) Jan 25 22:47:18 kergoth: yes; that would be a nice thing to have Jan 25 22:47:58 kergoth: looking more carefully it wouldn't solve my use case since I need it to run just before networkmanager starts and after hal has been started but would work for most use cases; Jan 25 22:48:04 ah Jan 25 22:48:09 hmm Jan 25 23:03:24 Is there some hidden definitive documentation for writing bitbake files? Jan 25 23:06:46 The manual is...uhm...lacking. Jan 25 23:34:44 my console-image is reduced to 380K where I would like it to be that small but it does not boot. It was ok yesterday Jan 25 23:34:49 what changed ? Jan 25 23:41:11 it just have opkg, sysvinit and udev Jan 25 23:41:17 left in there Jan 25 23:43:21 ccffb0d58e3b36b3be440b9ceadb05ebabe0d847 has UTF character in history log Jan 25 23:44:11 khem try with debug init Jan 25 23:44:28 woglinde: the image is not populated Jan 25 23:44:39 ??? Jan 25 23:44:46 console-image Jan 25 23:44:53 it does not have all the ingredients Jan 25 23:45:01 and bitbake happily baked it Jan 26 00:25:33 otavio - look at /etc/init.d/rcS **** ENDING LOGGING AT Tue Jan 26 02:59:56 2010