**** BEGIN LOGGING AT Thu Jun 06 03:00:22 2013 Jun 06 03:09:57 there must be something weird about how parted handles sdcards Jun 06 08:07:46 morning all Jun 06 08:19:03 morning bluelightning Jun 06 08:19:06 morning all Jun 06 09:17:14 Hi Jun 06 09:18:42 I'm having trouble in do_fetch(). Apparently the environmentvars SSH_AGENT_PID and SSH_AUTH_SOCK is not there. Are there any good explanation as to why? Jun 06 09:21:59 The same is actually the case for all the whitelist vars I have declared. They are just not there in do_fetch() Jun 06 09:23:32 SorenHolm: the first thing to check would be if the variable is getting into the datastore by inspecting bitbake -e Jun 06 09:23:54 those two variables are whitelisted by default btw Jun 06 09:26:07 They are both there. Jun 06 09:39:36 SorenHolm: I just tested here, and values for these two vars set before running bitbake are being exported to do_fetch according to log.do_fetch Jun 06 09:40:12 SorenHolm: which version of BitBake/OE are you using? Jun 06 09:40:54 bluelightning: Everything master from a couple of days ago. Jun 06 09:41:19 ok, I was testing with master Jun 06 09:41:27 I don't think that code has changed in a while Jun 06 09:41:55 SorenHolm: no mention of these vars in log.do_fetch? Jun 06 09:58:49 bg Jun 06 09:58:53 gm Jun 06 10:00:01 bluelightning: no .... actually no environment variables are mensioned in the log. Jun 06 10:03:27 SorenHolm: here I see them being explicitly exported in the command line used for running the fetch Jun 06 10:08:59 bluelightning: Hmmmm.. strange. I can see that for other recipes as well. I'm rolling my own do_fetch() - could that make a difference ? Jun 06 10:09:22 SorenHolm: ah, that would almost certainly explain it Jun 06 10:11:09 SorenHolm: why the need to write your own do_fetch out of curiosity? Jun 06 10:12:15 bluelightning: ok. Because git@myserver:myrepo.git using pure ssh-key authentifaction did not work for me using SRC_URI. Jun 06 10:14:47 SorenHolm: I think the problem there may be that we explicitly don't look at ~/.gitconfig Jun 06 10:15:24 bluelightning: As far as I remember the problem was to specify the username .... Jun 06 10:24:12 SorenHolm: I just successfully fetched a remote repo using ssh with key-based authentication, so that generally should not be a problem Jun 06 10:24:31 bluelightning: How can I get the environment correctly setup when doing my own tasks? Jun 06 10:24:47 SRC_URI = "git://git@git.server.name/repo;protocol=ssh" Jun 06 10:25:30 SorenHolm: if possible I'd like to figure out why the standard git fetcher is not working in your case Jun 06 10:27:38 bluelightning: Ok I have this : Jun 06 10:27:41 SRC_URI = "git://git@myerver.com:MyRepo.git;protocol=ssh;branch=master" Jun 06 10:27:56 bluelightning: And bitbake tryes to do this : Jun 06 10:28:03 git ls-remote ssh://git@myerver.com:MyRepo.git master Jun 06 10:28:40 which obviously does not work, cause git do not accept that kind of reposityry path. Jun 06 10:29:13 The correct command is 'git ls-remote git@myerver.com:MyRepo.git master' Jun 06 10:29:30 SorenHolm: I'm not familiar with that syntax, should it not be /Myrepo.git ? Jun 06 10:30:18 bluelightning: no. The repository is located in the git-users home-dir. Jun 06 10:30:54 bluelightning: Like this : "git://git@myerver.com/home/git/MyRepo.git;protocol=ssh;branch=master" Jun 06 10:31:18 bluelightning: But still the fetcher executes : "git ls-remote ssh://git@myerver.com/home/git/MyRepo.git master" Jun 06 10:35:19 SorenHolm: and the latter doesn't work? Jun 06 10:38:17 bluelightning: correct. the problem lies in the "ssh://"-part. Jun 06 10:38:52 SorenHolm: supposedly those syntaxes should be equivalent Jun 06 10:42:37 Hello everybody. Is it possible to override a bbclass ? I mean creating a new bbclass with the same name in another repository and urges bitbake to use it.. Jun 06 10:43:28 vadmeste: it is, but then it's better to try to get changes to the bbclass into its original location Jun 06 10:44:22 SorenHolm: that's what I can determine from the git-fetch manual page at any rate Jun 06 10:44:45 bluelightning: could you tell me how to do that ? I can't find that in google.. Jun 06 10:45:37 SorenHolm: the fetcher could be made to understand the alternative syntax though; if you really need that I would suggest modifying the fetcher in bitbake/lib/bb/fetch/git.py rather than trying to hack around it using your own do_fetch Jun 06 10:46:17 vadmeste: if BBPATH is set for your layer then just put your version of the class in classes/ Jun 06 10:49:15 bluelightning: well - I guess I have to specify the full path to the repository then. Cause my latter example actually works, but it seems strange to have to specify the full path when all I wan't is to make it just use $HOME as reference. Jun 06 10:55:06 SorenHolm: I guess nobody has tried to use the other syntax previously Jun 06 10:55:23 SorenHolm: patches welcome - it probably wouldn't be hard to make the scp-like syntax work as well Jun 06 11:19:17 bluelightning: ok. But regarding the original issue. How do I setup the environment in my own do_fetch() ? Jun 06 11:31:25 morning all Jun 06 11:32:22 hi all, Jun 06 11:32:42 I have a question related to fglrx driver in openembedded platform Jun 06 11:32:57 I am having a screen tearing issue which I couldn't solve Jun 06 11:33:07 is there anybody who can shed some light? Jun 06 12:08:16 SorenHolm: probably you would use bb.fetch2.runfetchcmd() like the fetcher does Jun 06 12:10:34 bluelightning: yeah, but actually I have some additional command also so it would be inconvinient I think. Around 1 year ago this actually did work. Jun 06 12:14:19 SorenHolm: sorry I'm not catching your meaning... Jun 06 12:16:54 .wi22 Jun 06 12:16:59 ups Jun 06 12:21:52 SorenHolm: we may well have tightened up the behaviour of which variables get exported; normally we don't expect people to be writing their own custom do_fetch Jun 06 12:39:08 bluelightning: If I leave SRC_URI empty and add a do_fetch_prepend will I then have a more complete environment to work with ? Jun 06 12:39:43 SorenHolm: no, that will not help Jun 06 12:40:06 SorenHolm: we have explicitly not exported those variables to all tasks Jun 06 12:41:20 SorenHolm: look at how runfetchcmd in lib/bb/fetch2/__init__.py does it Jun 06 12:47:49 bluelightning: ok .. thanks. Jun 06 13:37:55 bluelightning: bitbake didn't use the bbclass found in vadmeste_layer/classes which has a high priority.. However, if I delete the original bbclass, my file is used.. Maybe I need to setup a new config... Jun 06 13:38:41 vadmeste_: bbclasses and config files are loaded based on BBLAYERS order in bblayers.conf, not layer recipe priority, unfortunately Jun 06 13:38:50 vadmeste_: move your layer before the others in bblayers.conf Jun 06 13:39:20 BBPATH order, isn't it? Jun 06 13:39:59 which usually is the same as BBLAYERS order, but some layers.conf weren't respecting it Jun 06 13:40:04 yep, but most layers append to it, and the layer.conf files are parsed in the order of BBLAYERS Jun 06 13:40:08 yeah, ones that prepend are problematic Jun 06 13:40:11 its a good point Jun 06 13:49:42 kergoth: it works.. thanks you for help Jun 06 14:08:20 vadmeste_: the complexities here are why I strongly recommend you try to get the changes into the original class... Jun 06 15:46:28 bluelightning: ping Jun 06 15:51:10 bluelightning: buildhistory_get_sdk_installed doesn't work well when 2 sdk images are built at the same time pkgcache="$1/installed-packages.tmp" and other temporary files are shared and build can fail with | rm: cannot remove '/OE/shr-core/buildhistory/sdk/oecore-i686-cortexa8-vfp-neon/target/installed-packages.tmp': No such file or directory Jun 06 15:51:20 or generate invalid buildhistory info Jun 06 16:00:06 JaMa: hmm... I hadn't tested that case I'll admit Jun 06 16:00:20 JaMa: can you please file a bug? Jun 06 16:03:36 yup Jun 06 16:03:51 I just wanted to ask you if you don't have such bug already Jun 06 16:08:43 Can OE help in creating an SDK for the generated Linux distribution, that runs on Win32 ? Right now I'm struggling with getting gcc compiled with mingw, using the sys-root from the linux created sdk Jun 06 16:08:48 JaMa: I was not aware of the issue, no Jun 06 16:18:28 ... "bc" which is a turing-complete programming language that's so seldom used busybox hasn't got it Jun 06 16:19:08 as distinct, of course, from the many turing-complete programming languages which are frequently used but, er, also not in busybox Jun 06 16:19:10 heh Jun 06 17:03:00 fun of vendor kernel trees... one repo 335 branches Jun 06 17:05:46 hrw nice Jun 06 17:05:51 its aurora? Jun 06 17:06:26 woglinde: yes Jun 06 17:06:29 hrw where are you working now inside cannonical again? Jun 06 17:06:45 woglinde: now nowhere - searching for new place Jun 06 17:06:54 woglinde: hi! Jun 06 17:07:13 woglinde: can you start your IM? Jun 06 17:07:19 zecke yes Jun 06 17:07:23 hrw oh Jun 06 17:07:30 so its the worst case Jun 06 17:11:13 jkroon_, I think you need a Canadian cross toolchain? Jun 06 17:11:24 Crofton, yes Jun 06 17:12:48 maybe try asking in #yocto, i think some of the more sdk focused people will be there Jun 06 17:13:00 I've been seeing things about adt on windows Jun 06 17:15:41 Crofton, ok, thanks Jun 06 18:02:29 hrw: we should have at least one embedded software spot Jun 06 18:02:35 *open Jun 06 18:02:59 in central california... Jun 06 18:04:13 two in fact Jun 06 18:04:21 http://www.arthrex.com/job-seeker/open-positions/941/senior-engineer-embedded-software Jun 06 18:04:44 http://www.arthrex.com/job-seeker/open-positions/1047/sr-software-engineer Jun 06 18:18:00 mr_science I do not think hrw will move to california Jun 06 18:19:47 mr_science ugh thats a heavy profile Jun 06 18:20:24 I do not think your company will find someone easily Jun 06 18:21:18 they don't generally ask me that stuff first... Jun 06 18:22:56 the other side is for that skill level you have to pay a lot Jun 06 18:38:52 woglinde: feel free to respond and help educate them Jun 06 18:39:59 i've done that before, ie argued with the HR person's initial response and actually convinced them that i was a better fit than they thought... Jun 06 18:40:15 sort of similar... Jun 06 19:11:49 Dear client, at the end of last week, Hetzner technicians discovered a "backdoor" in one of our internal monitoring systems (Nagios). Jun 06 19:12:16 ynezz yes Jun 06 19:12:19 only two security incidents in past 6 months, priceless Jun 06 19:12:24 :) Jun 06 19:12:37 only 2 you know off Jun 06 19:13:17 yep, but I use it only for git and jenkins, so I don't care that much Jun 06 19:14:31 unless sensitive data leak (credit card etc) Jun 06 19:16:50 yep, but once you enter this into the internet, it's not sensitive nor private anymore :) Jun 06 19:17:36 Good point. :-) Jun 06 19:30:31 fray, the end game for what recipe belongs in what is layer is one layer per recipe :) Jun 06 20:43:17 in a recipe, how do i specify a depend on native libstdc++ ? Jun 06 20:43:53 you dont Jun 06 20:44:23 either you install it on host or not Jun 06 20:45:02 host toolchain is independent of bitbake and the external cross toolchain Jun 06 20:45:16 different versions and all that Jun 06 20:47:46 is that really the best(only?) option? Jun 06 21:03:06 yes Jun 06 21:03:19 put it in the readme of your layer Jun 06 21:03:57 i was hoping to distribute the native tool in the sdk i give to local developers Jun 06 21:04:18 sounds like that's not going to work... Jun 06 21:34:37 where is i2cdump in angstrom? Jun 06 21:35:07 i don't see any recipe for it, but i've had angtstrom prebuilts which had it. so i'm wondering where they've got it from Jun 06 21:59:48 muhoo: maybe i2c-tools recipe? Jun 06 22:00:55 ah, fair. i was doing find . -name i2cdump, not find . -name '*i2c*'. thanks Jun 06 22:31:44 well, *that* was interesting... Jun 06 22:47:02 fun with native/cross packaging... Jun 07 01:18:47 I'm having problems with locale support in uClibc: http://pastebin.com/rUfar3Ww Does anyone know why locale_t isn't recognized? Jun 07 01:19:31 uClibc config: http://pastebin.com/KF3pdEcV **** ENDING LOGGING AT Fri Jun 07 02:59:58 2013