**** BEGIN LOGGING AT Thu Nov 15 02:59:59 2012 Nov 15 09:32:21 morning all Nov 15 09:41:05 hi bluelightning Nov 15 09:41:08 hi all Nov 15 09:41:11 hi silvio Nov 15 09:58:02 there is an official pdf of yocto manual? Nov 15 10:18:02 hi Nov 15 10:27:56 hi afournier Nov 15 10:34:43 hi silvio Nov 15 11:36:27 bluelightning: hi, are you around? Nov 15 11:42:14 hi anarsoul Nov 15 11:42:34 bluelightning: do you still own h1940? Nov 15 11:42:38 I do, yes Nov 15 11:42:53 haven't touched it in a while but I still have it Nov 15 11:43:14 did you try 3.1 kernel on it? there's a discussion on lakml that s3c24xx stuff is broken for 3.x kernel Nov 15 11:43:26 but I'm pretty sure that h1940 can run 3.1 Nov 15 11:43:39 I don't recall; probably I didn't yet Nov 15 11:44:04 3.1 is kind of old by now though... Nov 15 11:44:08 I see Nov 15 11:44:18 bluelightning: I know, it's time for update Nov 15 11:44:29 (not that the recipes in meta-handheld are at all up-to-date for ipaqs) Nov 15 11:44:44 so much things to hack on, and so little time :) Nov 15 11:44:50 indeed Nov 15 11:45:02 I would certainly appreciate any advice/assistance you could provide Nov 15 11:46:14 bluelightning: I'll test 3.6 on h1940 and rx1950 machines and report later; need to get aes1660 fingerprint scanner working first Nov 15 11:46:29 anarsoul: ok, great Nov 15 11:54:29 morning Nov 15 12:22:58 anarsoul: hi there Nov 15 12:23:42 anarsoul: I've seen your spi-pxa2xx patch is still in queue Nov 15 14:52:16 how do you do when a package has a license but it does not include it inside ? Nov 15 14:53:55 afournier: what do you mean by "does not include it inside"? Nov 15 14:54:14 the LICENSE file is not present, but the website states it's under GPL Nov 15 15:08:33 afournier: does it have GPL text in one of the source code files? Nov 15 15:09:20 the README has a link to http://www.gnu.org/licenses/gpl.html Nov 15 15:09:53 i use the README as the license and use the line offset trick to do a partial md5 of the paragraph ? Nov 15 15:10:22 if that is the only reference to it's GPL licensing then I guess that's all you can do! Nov 15 15:10:27 ok Nov 15 15:12:51 yes, use the line start/end to capture the description of what the license in and/or copyright statements at a minimum Nov 15 15:13:03 gotcha thanks Nov 15 15:13:08 in some cases recipes patch in a copy of the license text, as well, iirc Nov 15 15:13:36 i was thinking about it, but it looks like a big hack Nov 15 15:13:37 :) Nov 15 15:13:37 but the key is to ensure that any change to the license upstream results in the build aborting, rather than building with an incorrect LICENSE field Nov 15 15:13:40 * kergoth nods Nov 15 15:13:56 you may wish to at the same time tell upstream to include the license text ;) Nov 15 15:14:03 ok Nov 15 15:14:04 good idea Nov 15 16:11:30 hi everybody!! Nov 15 16:12:01 hi chouimat Nov 15 16:25:38 hey Nov 15 16:25:58 what package contains adduser and the rest of them? Nov 15 16:26:09 shadow-utils Nov 15 16:26:30 ty Nov 15 16:29:35 well, shadow seems to contain useradd; i was thinking of adduser, i.e. spelled the other way around Nov 15 16:34:12 avahi-autoipd uses it in postinst Nov 15 16:35:19 presumably that should be updated to use useradd.bbclass Nov 15 16:36:27 okay; thanks Nov 15 16:37:29 correct.. any packages that add/modify users should use the adduser class instead.. Nov 15 16:37:36 otherwise it gets to be a mess Nov 15 16:39:29 it seems it needs to be cleaned up Nov 15 16:45:47 we do have a few recipes that are doing it the old way Nov 15 16:46:11 if you look at useradd.bbclass itself you can see why it's a good idea to use it... Nov 15 17:05:25 seems useradd.bbclass does not handle removal on uninstall Nov 15 17:05:46 policy decision (it's intentional) Nov 15 17:05:55 once a user has been added files on the system could be owned by that user.. Nov 15 17:05:56 and also it seems the avahi recipe was already transitioned to use useradd.bbclass, except for avahi-autoipd there is a comment saying "cannot do this offline" Nov 15 17:06:13 so if you remove the user, you can leave numerically owned files which is usually what you don't want.. Nov 15 17:06:18 fray, my thoughts exactly Nov 15 17:06:22 so if a user is created, it lives until it's manually removed Nov 15 17:06:38 I'd like to know why it can't be done offline Nov 15 17:06:52 (I haven't seen any cases where it can't be done offline -- useradd that is) Nov 15 17:07:07 i have less of a clue than you :) Nov 15 17:11:48 the issue might have actually been handling for multiple packages, I think it can do that now though Nov 15 17:12:00 I believe it can Nov 15 21:55:05 how is BBCLASSEXTEND ment to be used Nov 15 21:55:14 read the many, many example usages Nov 15 21:55:23 done it Nov 15 21:55:34 it's just a word separate list of variants to produce. if native is there, then a second version of the recipe is constructed which inehrits native Nov 15 21:55:41 hence the 'BBCLASS' part of the name Nov 15 21:56:13 is that automatically consturcted Nov 15 21:56:53 bitbake does it Nov 15 21:57:07 all you do is put the name of a class there, it does the rest Nov 15 21:57:11 or classes Nov 15 21:57:37 right, so my dilemma is that I need to pass different EXTRA_OECONF params depending on if it is compiled for the native side or for the target Nov 15 21:58:04 yes, overrides are provided for that, and you can see many many examples of that too Nov 15 21:58:10 grep for class-native Nov 15 21:58:13 :) Nov 15 21:59:17 surprising how many developers don't want to hear "grep is your friend..." Nov 15 22:03:35 In this I was actually baffled with my editor (Sublime Text) which finds only 1 matchm but grep gives maaany... go figure Nov 15 22:03:57 misconfiguration of the project probably :) Nov 15 22:04:07 * kergoth isn't very good with sublime text 2 yet Nov 15 22:06:14 Yeah, so EXTRA_OECONF_class-target is perhaps what want Nov 15 22:07:44 use of the target override is extremely rare, it's almost always sufficient to define EXTRA_OECONF and then EXTRA_OECONF_class-native, and the former will be used for everything but native Nov 15 22:08:05 right Nov 15 22:09:27 in my case there is a config option --disable-scanner and I want that to be present in the target, but the native part needs to run it Nov 15 22:09:34 so that the binary is available Nov 15 22:09:44 when the target parts are compliled Nov 15 22:10:00 does that make sense Nov 15 22:10:15 yep, makes sense Nov 15 22:10:38 building something native and then using that to build the target is an extremely common thing to have to do when crosscompiling Nov 15 22:12:32 kergoth: yeah Nov 15 22:15:29 Does the native bits get triggered automatically? i.e. if I want the user of the component to only specify DEPENDS += "package" and not DEPENDS += "package-native package" Nov 15 22:16:57 DEPENDS and RDEPENDS are automatically mangled by native.bbclass to refer to the native versions Nov 15 22:19:02 ok Nov 15 22:19:28 but for running on the cmd line bitbake package, this still fails Nov 15 22:19:36 ? Nov 15 22:20:08 so if I do only bitbake the_packge on the cmd line Nov 15 22:20:25 is does not automagically build the native bits first Nov 15 22:20:31 i.e. I need to build it manually Nov 15 22:20:35 of course it won't Nov 15 22:20:39 it can't read your mind :) Nov 15 22:20:44 ;) Nov 15 22:20:47 DEPENDS += "package-native" in package*.bb Nov 15 22:21:12 hmmm... wont that induce a circular dependecy Nov 15 22:21:32 either native.bbclas is ssmart enough to exclude ${PN}, or you can use DEPENDS_class-native = "" Nov 15 22:22:12 but try without that first, it might recognize that it shoudlnt' depend on itself, not sure if the class checks for it Nov 15 22:26:50 kergoth: needs the override, otherwise creates a loop Nov 15 22:27:01 a dependency loop that is Nov 15 22:27:11 does it come with a barf-bag? Nov 15 22:28:05 barf-bag? Nov 15 22:28:13 error report or Nov 15 22:28:42 ERROR: Nov 15 22:28:42 Dependency loop #1 found: Nov 15 22:28:54 and so on... Nov 15 22:30:58 kergoth: thanks for the help! Nov 15 22:31:10 np Nov 16 01:56:31 hey Nov 16 01:56:43 where does one send patches for bitbake? on the oe-core mailing list? Nov 16 02:04:56 CMoH: bitbake-devel@lists.openembedded.org Nov 16 02:07:59 ty **** ENDING LOGGING AT Fri Nov 16 02:59:58 2012