**** BEGIN LOGGING AT Thu May 10 03:00:03 2012 May 10 05:16:25 gm May 10 07:10:50 good morning May 10 07:42:55 morning all May 10 07:46:27 hi all May 10 09:19:07 good morning May 10 09:20:36 Good morning. May 10 09:25:28 I'm just scratch my head for a couple of days now we should/could use OE/yocto to build regular debian packages for different target (Distro versions) May 10 09:25:54 I read something going into that direction here: http://lists.debian.org/debian-embedded/2010/11/msg00074.html May 10 09:26:45 hrw: thanks for you comment there May 10 09:28:56 hwinkel: Hey, what about RPM-based distros then? May 10 09:28:59 ;) May 10 09:48:48 hwinkel: why you need to do that? May 10 09:48:59 Debian is Debian, OpenEmbedded is OpenEmbedded May 10 09:51:55 Umeaboy: RPM as well, but start with the first one May 10 09:52:49 mckoan: because OE/Yocto is ment to build packages/distros and not ne one (if I read and understand the whole thing correctly May 10 09:56:33 hwinkel: I don't understand your sentence, sorry May 10 11:09:59 hwinkel: using OE to build packages for Debian is possible but going that way is waste of time May 10 11:22:09 hrw: any reason for waste of time May 10 11:23:34 hrw: we have the idea to have a single source (recipe) and build packages for all targets May 10 11:23:42 very high level spoken May 10 11:27:44 hwinkel: Oe will generate more granulated packages (which may be +) but you have to take care of all dependencies to be at same versions as in Debian (may be lower but not higher) May 10 11:28:05 you may need to have mangle dependencies of packages to get them installable May 10 11:28:34 cause OE splits libraries to separate packages when Debian may not do the same May 10 11:28:57 tried that years ago with building packages for maemo May 10 11:36:04 hrw, maybe i got not the whole context but why don't you build debs ON debian ? May 10 11:37:48 hi all May 10 11:37:54 lautriv: not me - hwinkel May 10 11:38:37 lautriv: when I was using OE to build debs they were to be installed on maemo running devices (where there was no such thing as proper on board development environment) May 10 11:38:38 hrw ah, ok. May 10 11:38:42 hi pb May 10 11:39:22 lamikr: yesterday I got Tizen device. some people will never learn how to make proper OS. May 10 11:39:31 http://marcin.juszkiewicz.com.pl/2012/05/10/tizen-first-impressions/ May 10 11:42:51 hrw: thanks , thats was somehow expected, currently we use rebuildd plus some scripts around to manage product lines have internal dependencies. May 10 11:43:18 glad that helped May 10 11:44:20 Having the goal of building packages for embedded targets(ipkg) RH(RPM) and Debian/Ubuntu(Deb) and reading the Yocto advertising to support all there of them lead to this idea to have a central build system can handle all of them May 10 11:44:53 a bit like OpenSuse Build service is promising but with lacking the cross build and embedded targets May 10 11:44:56 hwinkel: you confused distributions with package formats May 10 11:45:24 hrw: hhhmm, i try not to do May 10 11:45:27 ;-) May 10 11:45:55 hrw this post is sad and funny together. May 10 11:46:21 hwinkel: OE can generate packages as ipk, deb, rpm, tar files. If you want to add umbaumba format then it is usually a matter of few hours to write code. May 10 11:46:48 Yocto promises to build Distributions which results in images build from packages and provides a package stream which can be used to installed on targets by using the selected packet manager (APT, RPM, opck) May 10 11:47:20 hwinkel: of course it is possible to make OE build packages compatible with non-OE distributions (like Maemo, OpenWRT, Debian etc) but it usually has some incompatibilities involved May 10 11:47:40 hwinkel: and Yocto gives you that May 10 11:48:39 The questions is the amount of work required to know which versions of apps. and libraries are currently on a specific target like let say Ubuntu 12.04 May 10 11:49:44 hrw, do you know about newer (working) kernels for the hx4700 ? (older ipaq) it seems that thingie is somewhat lost since 2.6.21. May 10 11:49:55 on this topic may tools like DistroTracking https://wiki.yoctoproject.org/wiki/Distro_Tracking yould help May 10 11:50:01 lautriv: never used hx4700 May 10 11:50:19 hwinkel: I am more or less 2 years outside of OE now May 10 11:51:09 hrw: BWT I'm not expecting all runs out of the box, we are "just" evaluating the options (Suse OBS, Yocto/OE, HomeGrown) May 10 11:51:39 whith a strong influence on embedded OE is o the short list ;-) May 10 11:52:01 hwinkel: OBS can cross compile May 10 11:52:17 hwinkel: Maemo and MeeGo guys were using OBS for cross builds May 10 11:52:40 We should have a more modern kernel in OE for it. maybe Angstrom has support for it... May 10 11:53:20 hrw: yes, I read that, but as OE its a own universe one of our guys are diving into right now May 10 11:53:31 hrw, I'm not sure that puts OBS in a favourbale light :) May 10 11:53:44 Crofton|work: ;D May 10 11:54:07 Crofton|work: both maemo and meego are dead now but they left us zombie named tizen May 10 11:54:14 yes May 10 11:54:37 compared to maemo or meego I see one plus for tizen May 10 11:54:52 does the meta-handhelds layer have a newer hx4700 kernel? May 10 11:54:53 they gave me developer device for free May 10 11:54:58 rofl May 10 11:55:05 what is the device? May 10 11:55:20 when free == 6.20$ for public transport and 1h of my time May 10 11:56:02 Crofton|work: big plasticky phone with samsung dualcore arm, 1gb ram, 14.6gb emmc, 4.65" 720p screen, wifi, bt, gps May 10 11:56:11 sounds fun May 10 11:56:47 sounds like a good deal :) May 10 11:57:47 Crofton|work: hhm OBS seems messed up if it comes to the advertised foreign distros like Debian and needs some love fixing as well May 10 11:58:09 so either spend time the left or rigth May 10 12:04:27 can someone me explain, what is the difference between PACKAGE_INSTALL and IMAGE_INSTALL? May 10 12:13:17 Nobody knows the difference between PACKAGE_INSTALL and IMAGE_INSTALL? May 10 12:18:33 it is about packagegroups May 10 12:19:12 http://patches.openembedded.org/patch/1539/ May 10 12:19:23 iirc this was merged long ago May 10 12:24:21 ant_work: thx! May 10 12:26:35 lautriv: I'm maintaining meta-handheld were we have support for hx4700 but it's way out of date :/ May 10 12:26:56 lautriv: I understand quite a bit of work has been done at the kernel level for that machine recently May 10 12:30:07 bluelightning, i saw that 3.0 from koen on angstrom-distribution but nowhere sources and the online-builder seems also to be broken. tried OE a half year ago and also no success. May 10 12:35:59 lautriv: I managed to build 3.1 I think; things have moved on quite a bit since then May 10 12:36:29 bluelightning, vanilla ? May 10 12:36:42 lautriv: vanilla plus some patches May 10 12:36:59 bluelightning, where can i see those patches ? May 10 12:37:45 lautriv: they are basically ones that Paul Parsons sent me which he has already submitted to the linux-arm-kernel mailing list for review May 10 12:38:02 lautriv: some of them have been merged since the last set I had from him and when I tried to build with them I didn't get a booting kernel May 10 12:38:14 lautriv: what I need to do is go back to Paul and ask him about the current status May 10 12:39:40 bluelightning, would be nice to hear something in future :) i had a running one booting with kexecboot but CF-card died. May 10 12:42:15 bluelightning, wife called for lunch, just one more question......i have also a HC-700 (pda from motorola, mainly the same parts like hx4700 + a b/w-cam for barcode and lower res. ) did you see any details about this one to get it flashed ? May 10 12:43:44 lautriv: I've not heard of that device I'm afraid... if it's similar it's probably not impossible to imagine we could build a kernel for it, but atm I have trouble finding time to support the machines we currently have, so it would have to be something someone else took on May 10 12:46:07 bluelightning, i assumed that because the hc-700 is usually for commercial use ( transportation / logistics ) i would investigate a bit time if there were some tool running from embedded winblows, haret could not tell me much. May 10 12:46:28 <-- eat, bbl May 10 12:46:31 lautriv: from a quick browse of linux-arm-kernel it looks like a few of Paul's hx4700 patches may have made it into 3.4; if so that'll make things a lot easier May 10 12:50:10 or maybe not... I will have to check with him directly May 10 13:01:51 bluelightning, regardless of the actual state, where may i pull the arm-related port in future ? ( i was used to get them by distro ) May 10 13:02:54 lautriv: in terms of an actual built distribution I hope this will end up in angstrom; right now all we really have are recipes May 10 13:09:24 bluelightning, ok, so hopefully they fix the broken online-builder soon. May 10 13:13:43 lautriv: see announcement on angstrom ML May 10 13:16:40 JaMa, will do May 10 13:17:54 JaMa, bluelightning hints for any tool to dig a bit deeper in hardware/resources like haret does/did from a given winCE ? May 10 13:18:33 lautriv: I think haret is the only tool to do that sort of thing; unfortunately the project seems to have died May 10 13:18:46 lautriv: it would be worth poking the developers to see what they are doing May 10 13:21:12 bluelightning , yes i had to use some haret still on my backups and google didn't find anything. May 10 15:50:46 *tumbleweed* May 10 17:07:17 khem: Hi, are you around by any chance? I'm backporting the GCC 4.7 support from master to a layer that is supposed to work with the denzil release. I'm seeing the following when bitbake gets to libgcc: http://pastie.org/private/hdw0hntlo9eqrklzt4w . It works when running against current master. So, I suppose I'm missing something. Any Ideas what this could be? May 10 17:08:59 Hm, I guess part of the problem is that I do not entirely understand what's shared and when the gcc sources are duplicated : / May 10 17:17:16 anyone else run into an inability to clone openembedded-core-contrib May 10 17:19:47 kergoth`: at least the clone starts for me; Receving objects... May 10 17:20:59 just for the record, I hate bbappends .... May 10 17:22:11 hmm, odd May 10 17:22:26 ah, maybe its my gitconfig url replacements May 10 17:22:28 * kergoth` mutters May 10 17:22:29 kergoth`: `git clone git://git.openembedded.org/openembedded-core-contrib` finished successfully on my end May 10 17:22:42 k, thanks May 10 17:22:57 Crofton|irssi: what about them? May 10 17:23:13 when the original recipe changes versions May 10 17:23:19 the append has to change May 10 17:23:28 running into branch syncing issues May 10 17:23:58 we have talked about wildcards (%) in bbappend names to overcome that issue May 10 17:24:17 hasn't gone any further than that though May 10 17:24:49 yeah, I recall May 10 17:26:51 it would reduce maintenance, but the one downside would be it would be less obvious when the contents of the bbappend might be out of date May 10 17:27:04 problem solved by getting things back on denizl and cherry picking the recipe I had on master May 10 17:27:37 i.e. instead of an immediate error it would be some less obvious breakage some time later (perhaps not even until runtime) May 10 17:28:12 heh May 10 17:28:16 always a trade off May 10 18:18:16 khem: wrt my libgcc issue - it's 7cf9f0597648c0bdaa080976d74acfbfc4c8443d that I forgot to backport May 10 18:21:11 well, not exactly forgot to but I've copied the gcc-configure-runtime.inc gcc-cross-initial.inc and gcc-package-cross.inc but I suspect the build picks the ones from oe-core rather the layer. May 10 18:21:56 Strange, even after applying cb01b2a4fc82e62eacc4441fddc3fdfb04a3acf3 which changes the gcc recipes to use the complete paths in require directive May 10 18:22:44 Doesn't bitbake take the *inc files from the layer first? May 10 19:08:18 kenws: it depends on how layer appears in bbpath May 10 19:08:28 it has to appear before oe-core May 10 19:08:38 bbpath is traversed left to right like path May 10 19:08:45 first hit is taken May 10 19:09:12 kenws: do u plan to keep 4.6 in linaro layer ? May 10 19:09:28 I think you should punt it and start from 4.7 onwards May 10 19:09:47 since I think I all users of 4.6 are not moving to 4.7 May 10 19:10:00 s/not/now May 10 19:10:07 * mr_science finally finished a small set of hardware tests for the motherboard manufacturer May 10 19:10:39 now i have to get back on customizing the ltp stuff May 10 19:22:51 * mr_science swears at putty May 10 19:24:44 mr_science: you should rather swear at windows May 10 19:25:54 since computers and airconditioners are similar in some sense - they both work well when you shutdown the windows May 10 19:56:09 ok, still unsure how to go further...............is it recommented to use oe-core right now or is this just a "theoretical" branch for future releases ? May 10 19:57:52 lautriv: denzil branch is just week or 2 old.. so it's recommended to use that for "products", master for new development May 10 19:58:04 s/branch/release/ May 10 19:59:48 JaMa, i purged any earlier OE and will start from scratch again, is denzil oe-classic or oe-core ? May 10 20:00:46 oe-core May 10 20:02:29 thanks, will see what the wiki gives for further actions ;) May 10 20:08:12 lautriv: if you look at oe-classic git log you'll notice that it's pretty much dead now and only 2011.03-maintenance is worth using for existing products.. but for new it's better to start with oe-core than migrate to oe-core few months later May 10 20:09:30 JaMa, if i have to choose between old and new i would mostly select new but the wiki states "only qemu based" where my taget(s) are real machines. May 10 20:17:40 and oe-core is not the only layer :) May 10 20:18:11 actually there is quite a few and e.g. real phones are supported well enough in meta-smartphone layer May 10 20:18:43 lautriv: http://www.openembedded.org/wiki/LayerIndex - if you think qemu is all there is, i suggest you look there May 10 20:19:45 JaMa, found the layers-list now it's time to jump over my shadow because i hate that versioning-crap (git) the inventor is probably hanging on a limb. May 10 20:20:22 git and mercurial are the best scm tools in existence May 10 20:20:32 if you're stuck in the old client/server model, that's your own issue May 10 20:20:38 lautriv: You around? May 10 20:21:03 Umeaboy, nah :P May 10 20:21:11 OK. May 10 20:21:18 I'll wait til you ARE. ; May 10 20:21:20 ; May 10 20:21:21 ) May 10 20:21:22 ;) May 10 20:21:33 Umeaboy, ok, now i am May 10 20:21:38 hehehe May 10 20:21:39 OK. May 10 20:22:05 kergoth`, but nobody ever got it managed to deliver a proper structure. May 10 20:22:09 One thing is weird with my home entertainment system's WLAN-connected bluray-player. May 10 20:22:16 lautriv: huh? May 10 20:22:36 Umeaboy, what ? May 10 20:22:39 the separate layer structure would exist whehter we used git or not May 10 20:22:55 If you set it to a static IP-adress you don't get any answer at all from it when trying the same nmap-command again from my laptop which I've set to another IP-adress as well. May 10 20:23:13 kergoth`, whenever i git something, it looks like my FS exploded. May 10 20:23:16 But if I change it back & do the test it tells me that port 5353/udp is open. May 10 20:23:25 Why doesn't it do that all the time? May 10 20:23:32 A bug in nmap? May 10 20:23:36 lautriv: making blind generalizations makes you look like an idiot. May 10 20:25:06 kergoth`, some things can be generalized and git matches. (maybe it will change) May 10 20:28:00 Umeaboy, the regular port-usage is mdns (multicats domain name service) i can't ensure your box is using this port for the regular use. May 10 20:28:18 OK. May 10 20:28:26 So, what's next? May 10 20:32:00 Umeaboy, did you look on the chip-numbers while open ? May 10 20:34:00 Which ones? The one close to the flat white cable? May 10 20:34:04 ones May 10 21:44:40 Umeaboy, any chip you could find, whenever you have to compile something for this box, you should know how/what glues together. you can't expect lspci ;) May 10 21:45:28 How stupid of me. May 10 21:47:05 lautriv: Hmmmmmmmmm. Would it help if I built an update for uboot with the cvs-code? May 10 21:47:22 Umeaboy, however, this kernel is already specialized to some grad, getting details in .config is worth a bunch wjile knowing applied patches would be even better. May 10 21:47:47 Yes. May 10 21:47:51 I agree. May 10 21:48:13 So, where do I start? May 10 21:48:44 Umeaboy, actually you can't compile/update anything, you can't even know how it boots ( or was there something visible on screen ? ) May 10 21:50:28 Nope, but as the ROM says you can see the boot-session. May 10 21:51:17 Umeaboy, ROM said that ? May 10 21:51:22 Yes. May 10 21:51:34 http://dl.dropbox.com/u/56935780/ROM's/LG_HB_6200M55.rom May 10 21:51:43 Umeaboy, how ? did you talk with it ? May 10 21:51:44 Use hexedit. May 10 21:52:13 Search for a string like boot or uboot or similar. May 10 21:55:15 Umeaboy, ever tried to stick a usb-keyboard in ? May 10 21:55:50 Nope. I have one, but can't find it atm. May 10 21:55:55 I'm still looking. May 10 21:56:36 i found a VERY interresting hint in the blob .......... May 10 21:56:59 Umeaboy, this thingie is able to tftp-boot (PXE) May 10 21:59:24 Umeaboy, and you have obv. a (maybe hidden) option to update the image from USB, where you can select boot/primary/backup May 10 22:10:09 Right. May 10 22:10:18 Have to find a working keyboard. May 10 23:41:22 lautriv: I did it! May 10 23:41:26 With help thou. May 10 23:41:53 I managed to find a port to connect to. Now the only remaining issue is how to talk to it. **** ENDING LOGGING AT Fri May 11 02:59:58 2012