**** BEGIN LOGGING AT Mon Sep 10 02:59:58 2012 Sep 10 07:24:42 good morning Sep 10 07:38:58 hi! Sep 10 08:02:31 how can i change the includedir in my recipie so it does not point to /usr/include ? Sep 10 08:04:33 hi all Sep 10 08:05:02 hi Sep 10 08:05:39 morning all Sep 10 08:07:36 hi Sep 10 08:09:14 hm... my porting efforts were stopped by haret beeing unable to boot linux on that machine Sep 10 08:09:27 afournier: do you have problem with eglibc recipe? Sep 10 08:10:08 just downgrade make to 3.81, eglibc makefile has some issues with make 3.82. i guess it will be something with denial of mixture generic and non-generic targets Sep 10 08:10:35 it is my own recipe, it looks like autotools;bbclass sets --includedir=${includedir} that expands to --includedir=/usr/include, but it looks like it uses the /usr/include from my system, not from oe Sep 10 09:18:53 morning all Sep 10 09:19:38 hi pb_ Sep 10 09:25:07 hi bluelightning, zecke Sep 10 09:34:51 hey pb_ Sep 10 09:37:33 florian: good morning! Sep 10 09:54:30 what can cause that haret hangs just after message go go go in status box? may the kernel be wrong or haret just doesn't know how to boot it? Sep 10 09:54:52 on my second PDA this particular haret binary works ok Sep 10 09:55:20 machine is detected as generic PXA270 Sep 10 10:02:54 Hello Sep 10 10:04:34 Could somebody help me? I have an issue with "bitbake -c menuconfig virtual/kernel" on Fedora 16 x64. A menuconfig window opens then close without me being able to interact with it. Sep 10 10:05:15 I have read past issues on forums with this, linked to gnome terminal and others, but proposed fixes don't seem to work. Sep 10 10:06:26 Did anybody have such issues? On Fedora 13 (i686), I did not have the problem. However, meanwhile, I have done some "oebb.sh updates", so I can't really compare. Sep 10 10:08:31 erwin: if those proposed fixes involve setting TERMCMD* then it does not work anymore Sep 10 10:08:41 erwin: do you have ncurses-devel installed on your system? unless it's the latest version of the metadata, you need that for menuconfig Sep 10 10:08:47 erwin: now there is OE_TERMINAL Sep 10 10:11:14 Yeah, ncurses-devel is installed Sep 10 10:13:32 erwin: what happens if you do bitbake -c devshell virtual/kernel Sep 10 10:18:13 bluelightning: it opens a gnome terminal that "cd"s me to the git directory of the kernel sources (by the way I'm under kde, and have tried kterm and xfce terminals) Sep 10 10:19:05 erwin: ok, try running "make menuconfig" from inside there Sep 10 10:20:28 It works! Sep 10 10:21:07 Thanks, bluelightning. So... what's the rationale behind this? Sep 10 10:21:26 hmm, no rationale at all; I would have expected it to fail... Sep 10 10:22:49 Well, I have edited the input driver section and the .config was correctly modified. The architecture is also correct (arm, fyi) Sep 10 10:23:32 Is that a terminal problem? Sep 10 10:23:46 erwin: it doesn't sound like a terminal issue no Sep 10 10:26:57 erwin: can you hack oe-core/meta/classes/cml1.bbclass so it always pauses? Sep 10 10:27:34 mickey_office: good morning Sep 10 10:27:49 good morning pb_ Sep 10 10:29:19 did someone make a netsnmp recipe for oe-core ? Sep 10 10:32:32 afournier: there's one in meta-oe yes Sep 10 10:32:42 ah ok Sep 10 10:33:09 I think it will shortly to be moved across to meta-networking once that move happens; that's still within the same repo though Sep 10 10:36:43 bluelightning: pause where? In the "python do_menuconfig()"? Sep 10 10:38:43 erwin: just delete this: "if [ $? -ne 0 ]; then echo 'Command failed.'; " Sep 10 10:44:19 Actually, the "sources/openembedded-core/meta/classes/cml1.bbclass" that I have in my workspace does not test return code. Besides, when I said menuconfig did not work, but it's not accurate: oe says the command succeeded, but there's a window that opens and closes very quickly Sep 10 10:44:47 (this window seems to be one similar to the one obtained through "devshell") Sep 10 10:45:29 erwin: it suggests to me that the command it is running is exiting immediately Sep 10 10:46:02 erwin: you could try OE_TERMINAL = "screen"; that's what I use Sep 10 10:46:25 (I use KDE as well) Sep 10 10:46:31 FWIW Sep 10 10:52:23 bluelightning: well, that does not work (same behaviour as before). Well, at least with the "bitbake -c devshell" I have a trick to make the menuconfig thing work. Thanks a lot. I will try and have a correct solution when I have more time. Sep 10 10:54:06 erwin: hmm, well it works here, but I'm not using Fedora so that's at least one difference Sep 10 11:03:40 bluelightning: by doing an 'oe_terminal("${SHELL} -c \"make menuconfig; echo 'Pausing for 5 seconds'; sleep 5\"", '${PN} Configuration', d)', I can see that make menuconfig exits with an error 2 (ncurses problem). However, I have ncurses installed so I don't understand. Sep 10 11:04:15 The fact that the "bitbake -c devshell virtual/kernel" then "make menuconfig" works then does not make sense!! Sep 10 11:05:27 I have both x86 and x86-64 dev libs installed, could that be a source of problems? Sep 10 11:05:33 (for ncurses) Sep 10 11:05:37 I'm not sure, in theory no Sep 10 11:06:28 FWIW, this was changed in a later revision to the one you have to use the libraries installed by ncurses-native rather than relying upon the host Sep 10 11:07:21 I see. Sep 10 11:07:43 so I'm not sure whether it's worth spending too much time on it... Sep 10 11:09:15 Ok, at least I can keep on working safely on my driver (and use menuconfig, instead of *cough* manually editting .config). Thanks! Sep 10 11:09:25 Have a good day, bye. Sep 10 11:13:02 what would be a most ligthweight paint program ? Sep 10 11:14:37 xpaint? Sep 10 11:19:40 no xpaint recipe here Sep 10 11:39:26 bluelightning: finally I got one builder free to run bitbake -g emacs and libx11-native really isn't in emacs dependency tree, but it's pulled by libxft-native libxext-native libxkbfile-native pango-native mkfontscale-native cairo-native xkbcomp-native libxrender-native Sep 10 11:58:35 JaMa: are you able to trace that back further? Sep 10 12:04:29 bluelightning: ? what pulls that to shr-image? Sep 10 12:04:40 bluelightning: e.g. navit/midori through librsvg-native Sep 10 12:05:10 JaMa: I mean, a lot of those are libraries Sep 10 12:05:24 JaMa: it would be useful to know how they are brought in Sep 10 12:10:13 bluelightning: librsvg is used to convert pixmaps in navit and that pulls e.g cairo and pango Sep 10 12:11:03 JaMa: I wonder if we can build cairo-native and pango-native without X, since we're not rendering to X Sep 10 12:11:45 both of those recipes do support building without X as a DISTRO_FEATURE. Sep 10 12:12:22 pb_: right, so it should be possible; I don't think DISTRO_FEATURES should be influencing native though Sep 10 12:12:53 yeah, agreed; I just meant that you could easily enough add another term to the conditional there. Sep 10 12:13:07 right Sep 10 12:17:48 JaMa: the exception handling fixes have been merged btw Sep 10 12:24:53 bluelightning: yup, seen that, thanks! Sep 10 12:25:20 JaMa: let me know if you see any other poorly handled error situations Sep 10 12:35:53 someone use xmlrpc-c in OE-core? Sep 10 12:40:08 silvio_: I'm sure someone was asking about that a few months ago... I'm assuming it wasn't you? Sep 10 12:40:53 bluelightning, You right wasn't me Sep 10 12:42:15 I saw that xmlrpc-c was present in classic, I take the recipe and I m try to adapt it to OE-core Sep 10 12:42:52 silvio_: yes I think if nobody has already done that that's a good idea Sep 10 12:43:49 bluelightning, If it works I can send the recipe as bugtrack or what? Sep 10 12:44:11 silvio_: please send a patch according to the recommendations in the meta-oe README Sep 10 13:02:56 is it possible to mask licenses by image? What we want is a release image with no GPLv3 packages. Then we want a development image, where we can have stuff like gdb. Ideally we would like to NOT have to change local.conf between these two builds. Sep 10 13:10:21 would it be possible to do something in the lines of what HOSTTOOLS_WHITELIST_GPLv3 does? Sep 10 13:13:26 mertsas: I think you can only whitelist it or not; we don't currently have specific provisions for doing what you're attempting to do Sep 10 13:13:58 ok, thanks Sep 10 13:55:33 how do you handle tarball that contains uppercase letters, basically it does not unpack in the good directory Sep 10 14:00:23 morning Sep 10 14:00:40 afournier: are you sure it's the case that's the issue and not the internal directory structure of that specific tarball? Sep 10 14:02:40 bluelightning: the internal directory has the same name than the tarball, and both contains uppercase letters Sep 10 14:03:27 afournier: ok, so it sounds like all you need to do is set S to match the extracted directory name Sep 10 14:03:45 so i just have to redeclare S Sep 10 14:03:52 I think so yes Sep 10 14:03:58 thanks for the tip Sep 10 14:04:06 S = "${WORKDIR}/SomeAnnoyingName" Sep 10 14:04:08 np Sep 10 14:54:09 bluelightning: i got a 3.5.2 kernel working on the hx4700 Sep 10 14:54:18 DarkPlayer: great! Sep 10 14:54:57 bluelightning: the error seems to be related to the kexec syscall, if it is enabled, the kernel is unbootable Sep 10 14:57:59 bluelightning: do you have working wifi with the acx100 driver ? Sep 10 14:58:44 DarkPlayer: no, I never looked into that Sep 10 14:59:09 well, not beyond searching for updated already available patches anyway Sep 10 15:00:36 hmm i got it almost working Sep 10 15:01:36 can search for wifi networks and connect to them, but the module does not allow me to send data Sep 10 15:02:14 ah, hmm... sounds like progress but not quite there Sep 10 15:02:59 cmd=ACX1xx_CMD_ENABLE_TX ... status=Channel invalid in current regulatory domain: FAILED Sep 10 15:03:32 the regulatory domain is set correct, but the driver block all traffic Sep 10 15:06:58 sigh Sep 10 18:54:22 pb_: what is this READ_ONLY_ROOTFS? is it a real variable? Sep 10 18:58:29 msm: base_contains("IMAGE_FEATURES", "read-only-rootfs" Sep 10 18:59:02 lower case - ok Sep 10 18:59:02 msm: used only in rootfs_ipk.bbclass Sep 10 19:07:43 JaMa: pb_: ok this is my fix: https://gist.github.com/3693022 Sep 10 21:16:58 msm: oh, yeah, sorry. it used to be a real variable, but now it is part of IMAGE_FEATURES Sep 10 21:17:14 like O_P_M, which bluelightning reminded me the other day no longer exists in its own right. :-} Sep 10 21:18:59 pb_: i submitted something that might clean things up Sep 10 21:19:19 thanks Sep 10 21:36:15 why is there both a meta-systemd on git.yoctoproject.org and one in meta-oe? Sep 10 21:36:21 as if things weren't confusing enough already Sep 10 22:15:57 that does seem faintly nuts. Sep 10 22:16:31 NIAI, I guess Sep 11 00:35:34 any advice on working in the devshell, looks like the wiki and the manual both state to be done Sep 11 00:36:05 I've gotten into it and my goal is to step through the do_configure task as that is was was failing Sep 11 00:36:42 ./configure doesn't seem to be the way to go and it fussed about a --host option Sep 11 01:02:37 oh so ./configure --host=arm appears to succeed, which means I didn't do something right. Sep 11 01:05:23 I noticed that printenv indicates libdir=/usr/lib, which is the root of /usr/lib/cmplrs/cc2.11, and not in the oe tree. Does anyone know if libdir should point to /usr/lib or a lib directory within OEDIR Sep 11 01:06:21 —libdir is passed as exactly what it should be Sep 11 01:11:46 ok. I'm trying to track down how this line gets constructed during the configure phase of gcc_4.5: arm-angstrom-linux-gnueabi-gcc -march=armv7-a -fno-tree-vectorize -mthumb-interwork -mfloat-abi=softfp -mfpu=neon -mtune=cortex-a8 --sysroot=/home/one/setup-scripts/build/tmp-angstrom_v2012_05-eglibc/sysroots/beagleboard -o conftest -pipe -g -feliminate-unused-debug-types -Wl,-O1 -Wl,--hash-style=gnu -Wl,--as-needed conftest.c Sep 11 01:11:46 -L/usr/lib/cmplrs/cc3.11 Sep 11 01:11:50 doh Sep 11 01:12:17 the relevant part is: arm-angstrom-linux-gnueabi-gcc …. -L/usr/lib/cmplrs/cc3.11 Sep 11 01:12:46 It then bias out as /usr/lib/cmplrs/cc3.11 isn't safe for cross-compiling Sep 11 01:12:52 bails* Sep 11 01:37:54 looks like a targeted -c clean may have fixed it, ugh Sep 11 01:43:04 Would someone look at my attempt at a kernel bbapend (http://pastebin.com/sTisGcKT) and point out what I've done wrong that building can't find the patches (http://pastebin.com/XFTHRpYB) Sep 11 01:43:35 FTR I also tried FILESEXTRAPATHS_prepend := "${THISDIR}/${PN}:" Sep 11 02:08:15 nm, stupid typo **** ENDING LOGGING AT Tue Sep 11 02:59:58 2012