**** BEGIN LOGGING AT Tue Dec 06 02:59:57 2011 Dec 06 13:36:23 Hello ... Dec 06 13:36:32 I need some help with dosfstools Dec 06 13:36:38 can someone help with it? Dec 06 13:39:02 is there any problem in using gplv3 tools for building? not installing it onto target but at build process Dec 06 14:02:46 otavio: I guess that's a question for your lawyers :) Dec 06 14:02:56 I think the general assumption is no though Dec 06 14:03:22 bluelightning: I am concerned about adding the newer dosfstools onto oe-core Dec 06 14:03:36 bluelightning: then as it is gplv3 i'd like to know about it Dec 06 14:04:11 otavio: you might want to ask sgw when he arrives Dec 06 14:05:52 bluelightning: ah ok thx Dec 06 14:23:53 It seems nitink is the one responsable in fixing the dosfstools to work with direcotires. Dec 06 16:33:48 RP__: There are a few recent requests that I would like to see make it into 1.2 M1, if you are willing. Dec 06 16:34:11 sgw: ok, which ones? Dec 06 16:34:21 Lei's GIT protocol change, Koen's buildhistory, and my wget/distro_tracking Dec 06 16:40:16 sgw: any others? Dec 06 16:40:20 * RP__ is ok with those Dec 06 16:41:14 RP__: at this point no, I think others are master, pending more testing and bug, I think that's RC1 Dec 06 16:41:38 RP__: Oops, maybe Martin's sysstat set Dec 06 16:42:24 sgw: the ones not in master yet? Dec 06 16:42:28 RP__: I am not sure if you took incandescant's checksum for RC1 or not, but I think that can wait. Dec 06 16:42:55 RP__: right, I have not full read email, was scanning. Dec 06 16:43:40 sgw, RP__ I have a patch for jffs summary images. Will send in a couple of hours. Do you think "sum.jffs2" extension is ok? Dec 06 16:44:42 ant_work: we had jffs2.nosummary before why not jffs2.summary? Dec 06 16:44:54 used ie here http://build.shr-project.org/shr-core/images/om-gta02/ Dec 06 16:47:40 I'd prefer to end with jffs2 being this is the real image format Dec 06 16:48:24 some examples use jffs2.sum or -sum.jffs2 Dec 06 16:48:38 that's why I'm, asking Dec 06 16:52:00 * incandescant is happy for his checksum patch to go in m2 Dec 06 16:53:09 om-gta02 has summary in .jffs2 and then extra jffs2.nosummary see http://paste.pocoo.org/show/517130/ Dec 06 16:55:04 JaMa: http://paste.debian.net/148334/ Dec 06 16:55:30 and http://paste.debian.net/148335/ Dec 06 16:56:23 I'll send to the ML when at home Dec 06 17:00:38 ant_work: don't you need --no-cleanmarkers also for sumtool? Dec 06 17:01:33 oh, right Dec 06 17:02:47 I'll fix that. This allows us to do http://paste.debian.net/148336/ Dec 06 17:03:49 I know it will make om-gta0* configs a bit shorter too, but I'll probably add mv to keep the names the same as they are now Dec 06 17:06:03 would be great to make it default... rename old IMAGE_CMD_jffs2 to IMAGE_CMD_jffs2.nosummary (and output too) and IMAGE_CMD_jffs2 producing .jffs2 with summary included Dec 06 17:06:15 hm.. JaMa, no, seems no fixes are needed Dec 06 17:07:02 ? Dec 06 17:07:15 about "--no-cleanmarkers also for sumtool" Dec 06 17:07:48 ok.. I didn't looked into it.. just noticed that we're using that now Dec 06 17:08:29 well, I'll add a -n just in case Dec 06 17:09:01 I see your doubt...I think that being the jffs2 is already created w/out cleanmarkers sumtool should not add any Dec 06 17:09:19 need a bit more testing then I'll send it. Dec 06 17:10:07 sgw: pushed updated 1.2_M1 Dec 06 17:11:39 RP__: thanks, I will start pulling stuff to MUT and testing a master Dec 06 17:26:56 gm all Dec 06 17:27:14 have we fixed a race condition in rpm since the edison release? Dec 06 17:27:41 e.g, the BUILDSPEC function fails due to "Unable to open temp file" Dec 06 17:28:02 This happened in the update-rc.d packaging on another person's system, I'm trying to debug remotely. :-/ Dec 06 17:34:52 is the pseudo guy here? Dec 06 17:36:15 oddly, yes, but I'm about to have to leave Dec 06 17:36:30 seebs, you might be interested in http://git.gnome.org/browse/ostree/tree/src/ostbuild/ostbuild-user-chroot.c Dec 06 17:36:53 seebs, basically on linux SECBIT_NOROOT | SECBIT_NOROOT_LOCKED allows one to do chroot(2) safely as a user Dec 06 17:37:22 as you might imagine, this approach is a *lot* faster than pseudo =) Dec 06 17:37:43 unshare(2) allows one to make private bind mounts Dec 06 17:37:46 Yes, but I believe it requires setuid for the initial setup. Dec 06 17:37:48 so i bind mount /proc and /dev in Dec 06 17:37:51 that's correct Dec 06 17:37:59 however i think we could eventually upstream a system call to do this Dec 06 17:38:15 The reason pseudo is the way it is, at the moment, is that we have to be able to do everything without ever having root. :) Dec 06 17:38:25 yep i totally get that Dec 06 17:38:49 That is sorta awesome, though. Hmm. Dec 06 17:39:00 we'd need pseudo for old kernels forever, but i think i'll try to see if i can argue for an unprivileged system call Dec 06 17:39:38 It's an interesting thought. I can't immediately see obvious problems, but I'm not a security guy. Dec 06 17:43:19 hm, i should make the bind mounts configurable so we can bind mount a result directory outside the chroot Dec 06 17:43:54 seebs, failing getting a kernel system call in, i could probably arm twist kzak into getting this in util-linux Dec 06 17:51:19 zenlinux: There is a bugfix in master for that Dec 06 17:51:37 zenlinux: look at patches to package_rpm to do with the localstatedir iirc Dec 06 17:52:06 zenlinux: http://git.yoctoproject.org/cgit.cgi/poky/commit/?id=c2ce6387bd6e0a05a7e2b13741ccc20de96eaf49 Dec 06 17:52:08 RP__, just found it - thanks! Dec 06 18:20:59 sgw: I'm testing out the merge of mine and kergoth's changes to the license bbclass. ugly merge due to some of my restructuring. not sure how RP wants to deal with it, but I'll have a branch with the changes that need to be made to get both of them in. Dec 06 21:00:28 what may I do to add debian to tested distros list? Dec 06 22:47:43 Jay7: Can you reply and ask that on the mailing list please? Dec 06 22:48:01 RP__: on which thread? Dec 06 22:48:24 Jay7: I assume you were reading the thread on the yocto list? Dec 06 22:48:37 "Sanity tested distributions list"? Dec 06 22:48:50 yeah, but I remember lot of such threads :) Dec 06 22:48:57 ah, ok Dec 07 01:56:57 Hey guys, I have a workflow question. Lets say I have a recipe, say AppABC that points to a local SVN:// url. Is there a way I can tell yocto to build from a local checked-out copy of SVN? Dec 07 01:57:41 Or, to put another way, what is the best workflow for a developer who is working on AppABC directly? Dec 07 01:58:00 They want to make changes and test them on our hardware (sh4 arch). Dec 07 01:58:14 mebrown: good question, I unfortunately don't have a good answer handy! Dec 07 01:58:30 Is there somebody who may? Any ideas? Dec 07 01:58:43 This is going to become an exceedingly important question for me to answer this week. Dec 07 01:59:01 mebrown: and all the smarties are probably gone for the day, we will get an answer for you in 24 hours Dec 07 01:59:17 I'm porting our build system over to yocto. I have completed porting all the open source apps, and now am going to start porting our actual application Dec 07 01:59:34 and there are some people concerned about the workflow, and I'm a yocto newbie and dont have a great anser. Dec 07 01:59:38 *answer Dec 07 02:29:24 mebrown: when you have svn:// uri it will checkout the sources locally into DL_DIR Dec 07 02:29:55 mebrown: those will be what build system will use to build the recipe Dec 07 02:31:18 mebrown: I think one way is that you prepare an SDK using yocto's build system and hand it over to app developer who then uses it to develope the app Dec 07 02:31:44 secondly then app can be added to build system using the bitbake recipe Dec 07 02:40:34 mebrown: you can sort of do what you want Dec 07 02:40:43 mebrown: just make a layer Dec 07 02:40:46 with a bunch of bbappends Dec 07 02:40:55 that modify, let's say linux Dec 07 02:41:06 linux_3.0.bbapend: Dec 07 02:41:14 S = /full/path/to/linux Dec 07 02:41:18 msm: his concern is for active developers who will work on software Dec 07 02:41:25 then it will unpack, patch, and compile in that source directory Dec 07 02:41:25 not build system Dec 07 02:41:38 khem: well we sort of do that too Dec 07 02:42:13 yes changing S is also a way Dec 07 02:42:29 there were srctree classes in oe.classic Dec 07 02:42:38 I think they did not make into oe-core yet Dec 07 02:44:15 i think those are used in our attempt at doing it Dec 07 02:44:22 i dont like using it, so its not well tested Dec 07 02:46:54 mebrown: the recommend way is to deploy your app via ssh or tcf and debug with gdbserver Dec 07 02:47:04 then you can just tweak/debug until its ready Dec 07 02:47:07 then add it in a recipe **** ENDING LOGGING AT Wed Dec 07 02:59:57 2011