**** BEGIN LOGGING AT Tue Jun 18 02:59:59 2013 Jun 18 04:17:07 1 Jun 18 04:17:10 gah Jun 18 06:43:21 boveutxein Jun 18 07:09:28 How is the PR variable used? What conseqeuence will I face if I forgot to increment it? Jun 18 07:59:29 some good guy committed my patch to openembedded. Thanks Richard Purdie (RP I think). Jun 18 08:01:32 I'm referring to http://cgit.openembedded.org/openembedded-core/commit/?id=cf59801be372bda962a94e6a406e97d20744ae45 Jun 18 08:01:38 morning all Jun 18 08:19:19 hi, anyone knows if newer yocto kernel than 3.0 is available for intel d2550/nm10 (cedartrail) platform? Jun 18 08:20:16 panda84kde: you're welcome :) Jun 18 08:20:30 morning all Jun 18 08:21:46 noticed that cedartrail was dropped from dylan, just wondering if it will come back, or has been renamed or something.. Jun 18 08:24:54 hbragge: the cdv graphics driver only supports up to kernel 3.1.0 Jun 18 08:26:22 hbragge: it hasn't been updated to be compatible with the newer kernels Jun 18 08:27:58 Net147: ok, any updates to be expected? Jun 18 08:28:29 i would think there is some demand at least, or? Jun 18 08:29:30 hbragge: not sure what Intel's plans are for the driver Jun 18 08:30:44 Net147: ok, many thanks for your info Jun 18 08:32:37 hbragge: I am interested in it as well since I use D2500HN Jun 18 09:00:56 Is there a way to see what package (rpm) installed a specific file in the rootfs? Jun 18 09:12:09 Saur: if you use rpms you can use rpm -qf /path/to/file (on target) Jun 18 09:13:07 Saur: that requires that you kept the package database though, some minimal images delete it to keep size down Jun 18 09:13:13 erbo: Was more thinking of on the host (don't have rpm on the target) Jun 18 09:15:03 Saur: we don't have anything out of the box that will do that, however you could hack it by adding a shell function to ROOTFS_POSTPROCESS_COMMAND that calls ${RPM} -qf ${IMAGE_ROOTFS}/path/to/file Jun 18 09:15:32 ok Jun 18 09:20:38 What is responsible for creating basic devices like /dev/null and /dev/zero that are needed long before udev is started? Jun 18 09:28:36 Ah, figured it out (meta/files/device_table-minimal.txt). Jun 18 09:29:59 Is it possible to enable network access, if I use BB_NO_NETWORK, for certain recipes or host? Jun 18 09:35:24 gonzzor: not that I know of, but you could use BB_FETCH_PREMIRRORONLY and use PREMIRRORS (or INHERIT += "own-mirrors" and SOURCE_MIRROR_URL, which does the same thing) to point to a local server Jun 18 09:36:04 bluelightning: That could work, thanks Jun 18 11:42:53 I am currently switching between building Poky efb8a460d2 and Poky 1.5_M1.rc1. Somewhere in between libtasn1 changes version from libtasn1.so.3 to libtasn1.so.6 (I expect in c880bd0e13). Now when I rebuild it fails on wpa_supplicant because libgnutls is still referring to the other version of libtasn1. Why isn't libgnutls rebuilt so it refers to the current version of libtasn1 (it does DEPEND on it)? Jun 18 12:12:54 Btw, I need to do "bitbake -c clean gnutls" and "bitbake -f -C fetch gnutls" to get wpa-supplicant to build again (after up/downgrading past c880bd0e13 of poky). That doesn't feel right... Jun 18 13:05:14 there is no reason a 2.6.x kernel compiled from Dylan shouldn't work? Jun 18 13:10:48 tasslehoff: no explicit reason I am aware of, although I have not tried building a 2.6 kernel with it Jun 18 13:12:39 bluelightning: ok. I'm stuck at "Starting kernel", with a recipe that worked for denzil. Jun 18 13:14:41 tasslehoff: I can't tell but this may be related: http://comments.gmane.org/gmane.comp.handhelds.openembedded/55701 Jun 18 13:16:25 I can only assume that if the kernel config is exactly the same the issue must be with building the kernel with a newer gcc Jun 18 13:18:48 bluelightning: that's what I thought as well. the uImage looks sane, the config is the same, and "mkimage -l" looks correct. Jun 18 13:26:04 Hello, is possible to configure a client IRc to access direct this channel? Jun 18 13:27:14 bluelightning: first attempt tomorrow is to compile with -mno-unaligned-access Jun 18 13:32:52 Sam__: depends on your client, quassel should support that Jun 18 13:37:38 Hello, i'm trying to create a image of Yocto using the kernel 3.8.13 to Gumstix Overo, who has tried this? Jun 18 13:40:24 Sam___: I'm sure many people, however there might not be any in this room at this precise time Jun 18 13:40:36 Sam___: are you having any difficulties? Jun 18 13:48:51 yes, with the machine, i look the versions of machines in the git of the project (kernel 3.8.13), if i use the arm is sufficient Jun 18 13:48:51 ? Or i need to create one new BSP? Jun 18 13:53:17 bluelightning: yes, with the machine, i look the versions of machines in the git of the project (kernel 3.8.13), if i use the arm is sufficient Jun 18 13:53:17 ? Or i need to create one new BSP? Jun 18 13:54:36 Sam___: you shouldn't need to create a new BSP unless there isn't a BSP for your hardware; meta-gumstix should provide BSPs for the gumstix machines I think Jun 18 14:01:18 Hi all - has anybody worked with armv6k tunings? I'm looking for patches for ARM tunings for the arm1176jz-s. Jun 18 15:00:08 YPTM: Björn Stenberg is on the call Jun 18 15:01:58 YPTM: Welcome to the technical team meeting. Please let me know who's on the bridge. Thanks! Jun 18 15:02:13 Song_Liu: I'm on the call Jun 18 15:02:21 YPTM: Paul Eggleton is on the call Jun 18 15:02:27 YPTM: tom z on the call Jun 18 15:02:35 YPTM: Björn Stenberg is on the call Jun 18 15:02:46 YPTM: Kevin Strasser is here Jun 18 15:03:00 YPTM: ross here Jun 18 15:03:30 YPTM: Michael is on the call Jun 18 15:03:54 Hi, all. David Wolfe of Navigation Solutions here. Jun 18 15:04:08 (Formerly of Magneti Marelli.) Jun 18 15:04:28 evadeflow: welcome back. I was looking at one of the bugs you filed recently funnily enough :) Jun 18 15:04:49 YPTM: Any opens? Jun 18 15:04:54 Really? Wow, those things have a long tail. :-) Jun 18 15:05:10 YPTM: nitin is on the bridge Jun 18 15:05:17 YPTM: bruce will dial in late. Jun 18 15:06:02 evadeflow: the one about fetching all the sources (I'm hoping I'm remembering the right thing! :) ) Jun 18 15:06:11 YPTM: Mark is here.. sorry I'm late Jun 18 15:06:46 RP: Oh, yeah, I remember that one. Jun 18 15:08:26 Song_Liu: here for a short moment :) Jun 18 15:10:00 Song_Liu: here is the current 1.4 test report Jun 18 15:10:00 https://wiki.yoctoproject.org/wiki/1.4_QA_Status Jun 18 15:10:51 AlexG: thanks Jun 18 15:12:24 YPTM: Beth just joined Jun 18 15:18:25 pidge want me to share the SPDX stuff? Jun 18 15:18:42 oh, we're a boring crowd :) Jun 18 15:19:10 Zagor: any ptest developments? Jun 18 15:19:57 RP: ah, yes. I can say a few words. Jun 18 15:21:05 Zagor: sounds good, look forward to seeing the patches! Jun 18 15:25:18 hi walters Jun 18 15:26:11 hey rburton Jun 18 15:28:02 bluelightning: perhaps you could give a summary of what is being planned for automated QA? Jun 18 15:30:01 urgh who pushed something giant to poky-contrib Jun 18 15:30:12 Receiving objects: 8% (10800/125958), 4.16 MiB | 184 KiB/s Jun 18 15:30:24 for a repo last fetched this morning at 5am Jun 18 15:32:31 RP: thanks for summarising Jun 18 15:33:00 evadeflow: that is good the hear, thanks! :) Jun 18 15:33:46 halstead: a fix has merged for that now, I think the ABs are using a workaround Jun 18 15:39:34 RP: It really was super easy. I followed the Task#1 and Task#2 tutorials from https://community.freescale.com/docs/DOC-94849 and got an image for i.MX6 Sabre LITE building in a few hours with no problems. Jun 18 15:39:46 It wasn't nearly so easy just 6 months ago. Jun 18 15:40:40 (For that board, at least. meta-fsl seems to have gotten lots better.) Jun 18 15:41:09 evadeflow: I think it has, yes Jun 18 15:55:46 anyone know of a package that provide ipv6 dhcp client services? Jun 18 16:01:09 Garibaldi|work: the "dhcp" recipe in master looks like it has support for ipv6 Jun 18 16:01:24 (without having actually tried it that is) Jun 18 16:03:17 ah, ok :-) Jun 18 16:06:20 I looked in the conf files and didn't see any ipv6 addresses Jun 18 16:06:41 so I thought maybe not, but I'll investigate more. If it does, that'll simply life greatly :-) Jun 18 16:08:21 yeah, their website says it supports both Jun 18 16:10:24 interestingly it does not configure itself based on the "ipv6" DISTRO_FEATURES item... I don't know if that's an oversight or would be unnecessary Jun 18 16:10:53 I'm guessing oversight.. I've seen a number of recipes that have that issue.. when I find them, I try to point people to fix them Jun 18 16:15:46 I'll give it a shot, if it seems it ought to work and doesn't, I'll file a bug Jun 18 16:19:18 Garibaldi|work: thanks Jun 18 16:20:40 bluelightning: which devs are developing hob? Jun 18 16:21:27 ant_work: you mean the existing gtk hob? I think Cristiana Voicu is doing the maintenance work atm Jun 18 16:21:44 I've seen again the issue where the hob script created its own tmp-eglibc-eglibc Jun 18 16:27:21 ant_work: definitely cause for filing a bug then if you wouldn't mind Jun 18 16:28:27 sure, I'll just want to see what happens if I run hob on build-from-scratch 'cause I've run it after bitbake invocations Jun 18 16:30:36 bbl Jun 18 20:21:58 There may be some pseudo work upcoming. We've got an experimental idea for performance improvements, and at some point in the next month or three I may have an update which cleans up and improves diagnostics/debugging. Jun 18 20:22:24 This is sort of a back-burner task, but it's a thing to plug away at when I'm waiting for other stuff. Jun 18 20:23:15 Goals are (1) reducing the computation done for messages that won't be displayed, (2) making it easier to get the debugging information you care about, (3) making it easier for people who aren't me to read the logs. Jun 18 20:48:53 seebs: sounds interesting Jun 18 20:50:02 My other theory, of developing a retrovirus which makes other people's brains enough more like mine that the existing diagnostics would work, was not approved by the ethics committee. Jun 18 20:50:18 In retrospect, I should have gone for a degree in *mad* psychology, their ethics committees are a lot more visionary-friendly. Jun 18 20:50:49 heh Jun 18 21:08:05 hah Jun 18 21:11:54 * fray finds Fedora make 3.82 is busted.. wonderful Jun 18 21:14:42 wow.. fc14 and newer are -all broken.. neat.. Jun 18 21:14:55 I wonder if anyone actually reads GNU savannah bugs exceptme... Jun 18 21:17:06 fray: ;) Jun 18 21:17:34 fray: the GNU make mailinglist was quite responsive when building WebKit got noticable slower (they didn't fix it though) Jun 18 21:19:41 heh Jun 18 21:20:16 seebs: not sure if you saw, there was a pseudo patch for compilation on rhel 4.7 sent to the oe-devel mailing list; I told the submitter to send it upstream Jun 18 21:21:02 Oh, I didn't see. Uhm. Jun 18 21:21:20 I am starting to be very nervous about the possibility that I may not yet be free of RHEL 4.x patches. Jun 18 21:21:52 * fray files a new bug Jun 18 21:22:01 Huh, I think I'm not actually *on* oe-devel, just oe-core. Jun 18 21:22:19 yeah, it was an odd place to send it Jun 18 21:22:46 seebs: it's here FYI: http://patchwork.openembedded.org/patch/51757/ Jun 18 21:23:45 fray: I thought we tested on F17 and F18 for the last release Jun 18 21:27:45 Thanks! That's… well, it's right, I think. What's weird is, the oldclone branch exists because we were building for RHEL 4.x at the time. Jun 18 21:27:57 So I totally thought it had been tested. Maybe it used to have a … and they cleaned it up. Jun 18 21:29:39 ahh Fedora bugzilla 975597 Jun 18 21:29:49 bluelightning .. it turns out Fedora "almost" fixed it Jun 18 21:30:03 see description on the oe-core list (or the Fedora bug) Jun 18 21:30:07 https://bugzilla.redhat.com/show_bug.cgi?id=975597 Jun 18 21:30:07 Bug 975597: was not found. Jun 18 21:31:05 AFAIK, it's broken all the way back to Fedora 14.... :P Jun 18 21:31:20 sorry.. Fedora -16- Jun 18 21:33:13 * fray waits for the Fedora people to say it's invalid and ignore that bug report Jun 18 21:34:47 * fray goes to verify the make 3.82 version we have in the YP is fixed.. :P Jun 18 21:34:51 just dawned on me Jun 18 21:35:03 fray: probably worth checking... Jun 18 21:35:21 I thought that rburton and walters had checked it already though Jun 18 21:35:34 looking at the Fedora version.. it appears that they have about 2 or 3 patches.. for 3.82 issues Jun 18 21:35:42 i only backported one patch which fixed systemd's build Jun 18 21:36:16 good.. the 30653 is one of them Jun 18 21:36:19 (and that is there) Jun 18 21:36:40 our 30612 patch might be broken as well.. Hmm. Jun 18 21:37:42 ya.. we're missing it as well.. Jun 18 21:37:46 now I have to go find it Jun 18 21:38:29 looks like it was fixed Sun 12 Jun 2011.. but easn't tagged in the commit message with the right ID number Jun 18 21:38:46 what does the syntax test.a( test_a.c test_b.c) mean? Jun 18 21:39:02 it defines test.a with the dependencies of test_a.c and test_b.c Jun 18 21:39:09 that is the bit that was fixed. Jun 18 21:39:14 what is missing is the leading space.. Jun 18 21:39:25 it's easy for a leading space to get in there due to variable substitution.. Jun 18 21:39:34 doesn't test.a: test_a.c test_b.c already say that test.a depends on those? Jun 18 21:39:35 http://savannah.gnu.org/bugs/?30612 Jun 18 21:40:02 ya.. just a test case.. ;) Jun 18 21:40:13 confusing... Jun 18 21:40:18 try this.. Jun 18 21:40:32 so you can inject dependencies on other targets inside the dependency list of a different target? Jun 18 21:40:36 acidfoo: bar.c foobar.a(foo.c) Jun 18 21:40:49 where did acid come from?! Jun 18 21:41:13 anyway.. that is a legal way to declare a dependency in foobar.a on foo.c, while defining something else Jun 18 21:41:47 seems like a pretty crack feature to me, although i guess i tend to use makefile-generating programs rather than writing raw make typically Jun 18 21:42:13 found the fix http://git.savannah.gnu.org/cgit/make.git/commit/?id=b06b8c64a29a5ba3a8daecd829fa2f98d42cb285 Jun 18 22:03:22 seebs: performance and other improvements sound good Jun 18 22:04:25 RP, re the make 3.82 -- I'd suggest we add make to the buildtools set.. :P Jun 18 22:04:50 fray: yes, that would be an easy fix Jun 18 22:05:05 once my current world build is done.. I'll be trying tht Jun 18 22:05:09 There's a couple of notions for performance hackery, not sure whether they'll work. But the diagnostics stuff could be massively improved, and I can have a pass over the docs to make sure they're reasonably coherent. Jun 18 22:05:34 (oh, and fray, the "acid" came from an IRC client figuring that x: where x is the end of a user's name is intended to be addressing them, so it expanded the name.) Jun 18 22:05:46 Biggest win for performance now would likely be to find ways to make configure faster Jun 18 22:06:12 33% of build time in do_configure :/ Jun 18 22:06:26 rp, obviously we need to use native compilation Jun 18 22:06:34 so people undertand how much better we are Jun 18 22:06:50 Crofton|work: how about we replace autoconf ;-) Jun 18 22:06:58 heh Jun 18 22:07:10 autoconf is pretty atrocious. Jun 18 22:07:12 or find a way to cache answers or something Jun 18 22:07:22 we have the answer cache (site cache) in place already.. Jun 18 22:07:44 You could get noticable improvements from making an autoconf variant generate configure files that aren't trying to be completely portable shell. Jun 18 22:07:47 we just need to populate more answers.. we should get performance improvement from looking at the cross section of configure runs and finding other places we can pre-answer questions Jun 18 22:07:59 seebs: right... Jun 18 22:08:01 But there's also a lot of just plain stupid tests. Jun 18 22:08:19 that too Jun 18 22:08:23 There's a lot of stuff that's tested by default even though it doesn't need to be, either for most things, or for anything. Jun 18 22:08:24 I suspect more site cache would be the lowest hanging fruit.. Jun 18 22:08:41 next is get rid of stupid tests.. followed by portable shell.. Jun 18 22:08:54 * fray loathes working in m4 though Jun 18 22:09:15 I think the shell stuff is mostly just startup time stuff like deciding whether configure needs to implement an "echo" that bypasses the limitations of particular systems, whether it's working in zsh, stuff like that. Jun 18 22:09:32 ya, so it might be easier to fix then I'm thinking Jun 18 22:09:33 from what I remember the time is 50% reautoconf and 50% running configure Jun 18 22:09:41 But for instance, I seem to recall noting that the default behavior of configure is to test whether an F77 exists. Jun 18 22:09:53 Or, say, trying to determine sizeof(char). Jun 18 22:10:15 the char (and related) we should already be caching Jun 18 22:10:31 I dunno about the bitbake/oe-core stuff. I believe the WR build system used to have pre-cached results that we just hand-populated into our config cache so they wouldn't get checked. Jun 18 22:10:50 ya, we moved that over to Poky (at the time) back in the 0.9 days Jun 18 22:10:56 seebs: we do have that functionality bit we don't use it much Jun 18 22:11:04 I suspect there are better ways to win Jun 18 22:11:19 anyhow, /me -> Zzzz Jun 18 22:11:22 'night all Jun 18 22:11:23 'night! Jun 18 22:11:39 That said. I still find the sizeof(char) test vaguely offensive, because the language spec *defines* sizeof in terms of char. Jun 18 22:11:45 When applied to an operand that has type char, unsigned char, or signed char, (or a qualified version thereof) the result is 1. Jun 18 22:11:59 So basically, if sizeof(char) != 1, you are out in the woods on your own, and we cannot help you. Jun 18 22:12:02 Sleep good! Jun 18 23:01:29 bluelightning: can you mark my duplicate patch on oe patchwork as superseded? keep the newer one. Jun 18 23:03:04 net147: done Jun 18 23:03:42 bluelightning: thanks. **** ENDING LOGGING AT Wed Jun 19 02:59:58 2013