**** BEGIN LOGGING AT Tue Jul 16 02:59:58 2013 Jul 16 03:03:58 Is anyone can help me for the Question: http://stackoverflow.com/questions/17648449/where-to-see-the-detail-of-do-rootfs-procedurethe-copied-files-path-in-openemb Jul 16 03:21:15 How to use the SSTATE_MIRRORS in the case of NFS server ? Jul 16 07:53:50 Hi, Jul 16 07:55:06 Is there a way to get a list of packages that are created from recipes? Jul 16 08:16:51 iwamatsu___: bitbake -s > packages.txt Jul 16 08:20:42 ant_work: thank you. However, the method does not take only a list of recipes. I want to get the list of packages that will be created from the recipe.... Jul 16 08:21:03 For example, libacl-dev and libacl is created from recipe of acl Jul 16 08:27:14 iwamatsu___: offhand I'd say to list providers you have to enable debug Jul 16 08:31:26 iwamatsu___: how about bitbake -e | grep "^PACKAGES=" Jul 16 08:35:26 ndec: thanks, it work! Jul 16 08:37:40 morning all Jul 16 09:23:56 pot shot here but has anyone managed to use a Yocto toolchain with Fedora 19? The installer worked on F18 but doesn't work on F19 - any takers? Jul 16 11:43:58 Hi how do i include qemu into my sdk generation? Jul 16 11:44:17 till now I've used "MACHINE="qemuarm" bitbake edna-image-minimal -c populate_sdk" Jul 16 11:51:58 i've read about runqemu-extract-sdk Jul 16 11:52:08 but then I don't have my full sysroot Jul 16 11:52:29 because then I have to go back to the adt-installer instead of "-c populate_sdk" Jul 16 12:12:05 fenrig: I think you would do TOOLCHAIN_HOST_TASK += "nativesdk-qemu" in your image recipe Jul 16 12:13:12 bluelightning: thx, however there is not much information/documentation about the sdk :/ Jul 16 12:13:52 fenrig: I will try to get that improved for the next release Jul 16 12:15:21 bluelightning: thx ;-) keep up the good work Jul 16 12:40:16 bluelightning: what does "TOOLCHAIN_HOST_TASK += "nativesdk-qemu"" do and how does it help me? Jul 16 12:40:41 fenrig: you wanted qemu in the host part of your SDK, right? Jul 16 12:40:53 bluelightning: yeah :) Jul 16 12:41:03 bluelightning: so how do I use it then? Jul 16 12:41:06 fenrig: right, that will accomplish that Jul 16 12:41:30 bluelightning: how? Jul 16 12:42:15 fenrig: it becomes part of the SDK installation in the same way as the cross compiler and other tools Jul 16 12:42:26 Hello. I am trying to build a package, but it seems that no matter what I do I cannot get LD use proper paths during build stage. This is my full recipe (without dependencies, but it is do_compile that fails for this recipe) and the error output: http://pastebin.com/7m6172iD Anyone have some pointers? :) Jul 16 12:43:55 fenrig: once the SDK has been installed you should be able to run qemu-system- Jul 16 12:45:28 Stygia: you shouldn't be touching ${D} until do_install Jul 16 12:45:42 Stygia: that's likely the root of your issue there Jul 16 12:46:34 bluelightning, Hmm alright, I was under the impression that without D prefixed, I would be instructing LD and the other tools to install into my actual root system? Jul 16 12:47:06 bluelightning, And can/should I still do DESTDIR=${D}? Jul 16 12:47:18 Stygia: yes you should, but only within do_install Jul 16 12:47:37 bluelightning: I can't use qemu-system- and I'd like to integrate it into eclipse :/ Jul 16 12:47:58 fenrig: I'm afraid I don't know anything about eclipse integration Jul 16 12:48:27 http://www.yoctoproject.org/docs/current/dev-manual/dev-manual.html#configuring-the-target-options Jul 16 12:48:36 I was following this (incomplete) guide Jul 16 12:49:02 but at the settings it doens't really help at pointing at the right things :o Jul 16 12:49:08 it's really general Jul 16 12:50:22 Stygia: otherwise, any dependencies should be picked up from the sysroot and output just goes to the directory in which you're building (the current directory or ${B} if you need to refer to it explicitly, which you shouldn't in most cases) Jul 16 12:50:27 bluelightning, Ah yes, now I'm back to my original problem, my cross-platform linker gives this error: | arm-poky-linux-gnueabi-ld: warning: library search path "/usr/local/lib" is unsafe for cross-compilation Jul 16 12:51:05 bluelightning, Hmm. I was actually trying to achieve the exact opposite, to prevent LD from using /usr/local/lib and other paths on my system. Jul 16 12:51:07 Stygia: right, it shouldn't be looking at your host system Jul 16 12:51:39 bluelightning, That's why I wanted to set LDFLAGS to ${D}${something}, but that didn't work for me. Jul 16 12:52:00 bluelightning, this is the command that fails: | arm-poky-linux-gnueabi-ld --sysroot=/mnt/dev/fsl-community-bsp/build/tmp/sysroots/imx6qsabresd -shared -L/usr/local/lib -fstack-protector -Wl,-Bsymbolic SQLite.o dbdimp.o sqlite3.o -o blib/arch/auto/DBD/SQLite/SQLite.so Jul 16 12:52:14 bluelightning, I was (mistakenly) hoping that adding ${D} in front of L in LDFLAGS would make it stop doing this. Jul 16 12:52:24 Well actually it _does_ make it stop doing that, giving the other error in my pastebin. Jul 16 12:53:56 bluelightning, There is _one_ source file - not mine, part of the module - that directly refers to /usr/local/include/ Jul 16 12:54:07 bluelightning, I have considered just sed'ing this file... but I realize that would be evil. Jul 16 12:55:32 Stygia: it's just that the sysroot is not ${D}, it's ${STAGING_DIR_HOST} Jul 16 12:56:03 Stygia: if the path is genuinely hardcoded and can't be overridden by passing in the path, sed'ing is the only option Jul 16 12:57:58 bluelightning, Well... alright. I will do that then, I think, hoping it helps. Jul 16 12:59:54 bluelightning, Hmm. What should I substitute for /usr/local/lib, ${libdir}? Jul 16 13:00:06 Stygia: I would think so yes Jul 16 13:00:29 bluelightning, Thanks, I'll try. Jul 16 13:00:41 bluelightning, You, by the way, are the single most friendly and helpful person I"ve found on IRC. Jul 16 13:01:00 Stygia: you're welcome :) Jul 16 13:01:13 bluelightning, The last couple of weeks of working with this would have sucked a _lot_ if not for you and rburton Jul 16 13:01:19 bluelightning, So hang in there. ;) Jul 16 13:02:02 Stygia: I have the impression perl isn't really designed with cross-compilation in mind, so you have probably jumped in at the deep end :) Jul 16 13:03:03 Stygia: I feel rewarded when people are successful with the system, so I'd say the same to you :) Jul 16 13:03:48 bluelightning, Hehe, yes, this has been kinda fun. But it has taken me about a week + to even run my script, and I'm still working on it. Jul 16 13:04:02 Okay, so it's roughly 1,000 lines of perl, and it works with our API and does a lot of work, but still. Jul 16 13:06:37 bluelightning, Once I finish all this I'll try and commit what I've done into OE proper... oe-classic? Jul 16 13:07:02 Stygia: if you have a number of additional recipes it might be worth creating a separate layer Jul 16 13:07:59 oe-classic is the old, no longer really maintained version of OE Jul 16 13:08:23 bluelightning, Ah right. Well we do already have a layer, but it's company-specific. Jul 16 13:09:10 Stygia: sure, in this case I figure it would be a generic layer for perl recipes that are additional to the ones already in OE-Core Jul 16 13:12:02 bluelightning, Hmm alright, well I'll look into that once I'm done. Jul 16 13:12:22 bluelightning, Question, is there a reason sed in BB doesn't take / for a delimiter but wants :? It weirds me out, I'm so used to PCRE Jul 16 13:13:20 Stygia: it should accept / but of course if you're dealing with strings that contain / that could be problematic, best to use : or ! or some other symbol Jul 16 13:14:40 hi guys, quick question: at the beginning of my bitbake build, all recipes are parsed. can I then, while the build is running, work on the recipes and layers or could that produce build errors if a recipe changes? Jul 16 13:14:45 Stygia: note that the delimiter in sed is up to the person writing the expression Jul 16 13:15:30 hsychla: I think recipes can get re-read during the build process so you should avoid doing that Jul 16 13:15:40 rburton, Yes... this is what I thought, actually. I just seemingly can't get sed to work from bb when it does work from bash. But I think I"m doing something wrong Jul 16 13:16:30 bluelightning: thanks for the quick answer! not the answer I was hoping for but that's not your fault... :) Jul 16 13:16:58 rburton: Can you please comment here further: http://patches.openembedded.org/patch/53489/ ; sorry about formatting, I did email client update and all settings messed up Jul 16 13:18:54 mshakeel: ah yes, sorry Jul 16 13:19:10 Stygia: feel free to pastebin something in case its a silly typo Jul 16 13:19:27 rburton, I will if I still can't get this to work in a while. But I have read it so many times, I don't think I typo'd again. Jul 16 13:19:41 rburton, Plus... regular expressions is one thing I have _tons_ of experience with, so if I'm doing the wrong I'll be pissy. Jul 16 13:20:09 Stygia: can you give an example sed command that doesn't work? Jul 16 13:22:03 bluelightning, sed -i 's/\/usr\/local\/lib\//${libdir}/g' sqlite3.c Jul 16 13:22:11 bluelightning, But I made this pass by using : for a delimiter. Jul 16 13:22:19 guys, in some cases a brute patch is easier... Jul 16 13:22:22 Stygia: ${libdir} contains slashes though Jul 16 13:22:37 Stygia: surely that's what would break it? Jul 16 13:22:50 bluelightning, Yes, but I have this expression in by do_configure_pretend and it does not complain: sed -i 's:\./configure\(.[^-]\):./configure --build=${BUILD_SYS} --host=${HOST_SYS} --target=${TARGET_SYS} --prefix=${prefix} --exec_prefix=${exec_prefix} --bindir=${bindir} --sbindir=${sbindir} --libexecdir=${libexecdir} --datadir=${datadir} --sysconfdir=${sysconfdir} --sharedstatedir=${sharedstatedir} --localstatedir=${localstatedir} --libdir= Jul 16 13:22:50 ${libdir} --includedir=${includedir} --oldincludedir=${oldincludedir} --infodir=${infodir} --mandir=${mandir}\1:' Makefile.PL Jul 16 13:22:59 bluelightning, Uh sorry, was a bit longer than I itended. Jul 16 13:23:44 Stygia: sure, but there you've specified that the separator is : so slashes wouldn't be an issue in that expression Jul 16 13:23:59 bluelightning, This is what I'm currently using: sed -i 's:\/usr\/local\/lib:${libdir}:g' sqlite3.c Jul 16 13:24:15 Stygia: sure Jul 16 13:25:02 Stygia: when ${libdir} gets expanded the original statement will become this: sed -i 's/\/usr\/local\/lib\///usr/lib/g' sqlite3.c Jul 16 13:25:03 bluelightning, But it doesn't seem to prevent the error with LD using /usr/local/lib... Jul 16 13:25:39 Stygia: there has to be another reference somewhere then, or a path you're not overriding by passing it in Jul 16 13:26:15 bluelightning, Yes, I assume so. I'll try and find the local copy of sqlite3.c and see what happens to it. Jul 16 13:26:38 bluelightning, but grep -ri "/usr/local/lib" in the extracted folder only shows one entry in sqlite3.c, though. Jul 16 13:27:31 Stygia: btw, have you been using the OE-Classic recipes as a reference at all? some of these recipes have already been written there and should only need migrating Jul 16 13:28:08 bluelightning, Yes, I have. Jul 16 13:28:42 bluelightning, http://cgit.openembedded.org/openembedded/tree/recipes/perl/libdbd-sqlite-perl_1.31.bb this recipe fails for me, though, complaining about Makefile.PL Jul 16 13:29:01 bluelightning, What happens at times with the CPAN meta class, but I'll work-around it I hope. Jul 16 13:30:23 Stygia: hmm... I would have thought the answer would be to continue to use cpan.bbclass and fix or override it where needed rather than re-doing everything by hand Jul 16 13:32:05 bluelightning, Usually it is. But this recipe is troublesome. Jul 16 13:33:33 bluelightning, And, hmm. Now I've substituted it successfully for ${libdir}. But it still stops and complains that -L/usr/local/lib is insecure for cross-compilation. I can resolve that by doing LDFLAGS = "-L${D}${libdir}", but then it complains that ..../image/usr/lib, in spite of that location existing. Jul 16 13:34:00 Stygia: as I mentioned, ${D} isn't the sysroot Jul 16 13:34:49 Stygia: ${STAGING_LIBDIR} is probably what you want (which includes ${libdir}) Jul 16 13:35:32 bluelightning, Oh wait, right. I'll try that. ${D} is the destination sysroot, the place from which the "real", eventual system is build, which is why I should only bother with it in do_install... yea? Jul 16 13:36:27 Stygia: ${D} is the destination directory yes, and yes Jul 16 13:37:32 Stygia: these are the kinds of details that a central class is supposed to help avoid you having to deal with ;) Jul 16 13:37:51 bluelightning, Heh yea I know. The CPAN class _usually_ works great, but it's harder to troubleshoot when it does complain. Jul 16 13:38:18 bluelightning, Question, is oe-classic maintained? If their recipe for perl-module-dbd-sqlite doesn't build ATM, can I try and commit my own recipe instead? Jul 16 13:38:55 Stygia: no, we prefer to get recipes working on top of OE-Core and find a layer to publish up-to-date recipes in (if one doesn't already exist) Jul 16 13:39:25 bluelightning, Ah, right. I think the "perl" layer layer will be fine, if it still exists in OE-core. Or devtools. Jul 16 13:39:44 bluelightning, I've been told (by rburton, IIRC) that I"ve build practically all of metacpan by now. Jul 16 13:40:32 well, a start of it :) Jul 16 13:40:34 Stygia: OE-Core is one layer; but we probably wouldn't add these to OE-Core itself since they aren't needed by most folks, a separate layer is probably the way to go Jul 16 13:41:25 i can see merit in a meta-cpan or something like that Jul 16 13:45:24 rburton, Yea, I could create that layer I suppose. I wouldn't even mind maintaining it, not by this point. Jul 16 13:46:13 And hmm. I have still not been able to make it work, using STAGING_LIBDIR. It gives the same error as using LIBDIR in LDFLAGS: http://pastebin.com/CabBU7YJ (recipe and errors included). Not tampering with LDFLAGS makes it try and -L/usr/local/lib which generates an error. Jul 16 13:47:26 The folder most definitively exists, it contains lots of files. Jul 16 13:49:59 Stygia: one sec, i'm just getting that statement witnessed by a lawyer Jul 16 13:50:17 stygia: does Makefile.PL take an -L flag? Jul 16 13:53:05 zibri, I don't think it takes options at all, it has no help and when I skimmed the source I didn't see opt parsing. Jul 16 13:53:06 i don't think so at least... i can't find it in any reference to in the source. Jul 16 13:53:17 zibri, Heh same conclusion I made. Jul 16 13:53:23 stygia: i does take stuff like PREFIX=/lolol Jul 16 13:53:30 but that's all i think Jul 16 13:54:00 zibri, Hmm okay, I missed that, then. Jul 16 13:54:34 zibri, But for now, it says that a directory which most certainly does exist doesn't. I'm probably misinterpreting the error output, but I"m sorta running out of ideas on where to look/what to try. Jul 16 13:54:45 stygia: note that -L Jul 16 13:55:01 ehrm, it complains about a path beginning with -L Jul 16 13:55:41 zibri, But I'm pretty sure this is how LD works, isn't it? Before I even tampered with LD at all it said complained about -L/usr/local/lib being insecure Jul 16 13:56:00 zibri, It's LD which fails, and I am/was under the impression that -L/path/something and -R/some/path is the right way to do it Jul 16 13:56:01 yes, but you do "perl Makefile.PL ${LDFLAGS} ..." Jul 16 13:56:34 ehm, sorry my bad. that's do_configure Jul 16 13:56:48 zibri, I'll try, but IIRC I did this already, and it will complain about -L/usr/local/lib? Jul 16 13:57:06 stygia: ah, in do_compile you start the command by doing ${LDFLAGS}, i.e. trying to execute -L${STAGING_LIBDIR} Jul 16 13:57:22 zibri, I tried LDFLAGS=${LDFLAGS}, too Jul 16 13:57:36 zibri, With it just dies saying that -L/usr/local/lib is insecure for cross-compliation Jul 16 13:57:43 look at line 32 in your paste, it's missing the actual command Jul 16 13:57:59 ah Jul 16 13:58:30 zibri, Which is why I tihnk this LD stuff is sorta right, it does prevent that error and instead creates _another_ error that -R/mnt/dev/fsl-community-bsp/build/tmp/sysroots/imx6qsabresd/usr/lib: not found Jul 16 13:58:30 LDFLAGS=${LDFLAGS} is at least more correct than the line in the paste Jul 16 13:58:37 zibri, Ah well. Jul 16 14:01:41 bluelightning, Hmm. Doing do_compile, should I be using STAGING_X instead of X varibales? Jul 16 14:02:00 Stygia: I assume so yes Jul 16 14:02:39 are there any good sdk documentation, cause I'm really stuck :/ Jul 16 14:03:25 fenrig, http://docs.openembedded.org/usermanual/usermanual.html These is this manual, which is the best I've found Jul 16 14:03:42 that manual is quite out-of-date by now though Jul 16 14:03:45 fenrig, IMHO OE/BB documentation is in need of some TLC. Jul 16 14:03:48 bluelightning, Ah. Jul 16 14:03:58 you really want to be using the Yocto Project documentation Jul 16 14:04:01 bluelightning, Well I always, always have problems finding proper documentation for it, though. Jul 16 14:04:07 bluelightning, Yocto project docs? Hmm alright. Jul 16 14:04:16 fenrig, Just listen to bluelightning he knows this way better than me Jul 16 14:04:20 oh yeah, and where does the difference from angstrom en yocto come from. Angstrom SDK generated a "native" tool directory and Yocto generetad some scripts :/ Jul 16 14:04:27 https://www.yoctoproject.org/documentation/current Jul 16 14:04:52 bluelightning: Those documentation aren't really complete enough Jul 16 14:05:26 fenrig: I realise that for your purposes, but they really are much more up-to-date than the OE manual linked above Jul 16 14:05:32 bluelightning: sometimes they are unclear. they seem to look nice but in reality they are everything but complete :/ Jul 16 14:05:48 fenrig: we do accept bug reports for the documentation Jul 16 14:06:01 in fact we would really appreciate them Jul 16 14:06:24 bluelightning: yes well I can report some bugs, but that won't really get my much farther at this moment. Jul 16 14:06:29 *further Jul 16 14:06:56 fenrig: I really want to help you but the problem I have is I don't regularly use the SDK nor do I use Eclipse... Jul 16 14:07:32 fenrig: if nobody else here can advise you I'd suggest you try posting the complete issue to the mailing list Jul 16 14:08:12 bluelightning: okay will do, hope this will help me :) Jul 16 14:14:57 bluelightning, If I want to write my own do_install etc while using cpan/autotools, this is when I set BBCLASSEXTEND = "native", yes? Jul 16 14:16:07 Stygia: er, no, BBCLASSEXTEND = "native" creates a -native variant of your recipe for building for the host Jul 16 14:16:29 (building for the build host that is) Jul 16 14:16:57 Stygia: if you want to specify your own do_install, just specify it after the inherit line Jul 16 14:17:56 bluelightning, Ah, right. Thanks. Hmm is there a system variable for /usr/share/, or /usr/local? I haven't found any in the manuals I have available Jul 16 14:21:51 Stygia: /usr/share is ${datadir}; /usr/local isn't a normal install path so we don't have a variable for that Jul 16 14:24:44 bluelightning: I mailed the mailing list, thx for helping me out Jul 16 14:30:47 bluelightning, Thanks, I will do that, and then use sed for /usr/local which _is_ being installed to by this program. Jul 16 14:31:01 bluelightning, I'm gonna go file a bug for /usr/local not having a variable Jul 16 14:31:16 Stygia: I don't think that's a bug Jul 16 14:31:53 Stygia: when pieces of software install stuff into /usr/local instead of ${prefix}, that is a bug in the way the software is being built Jul 16 14:32:34 bluelightning, But now, instead of having to the _option_ of redirecting /usr/local, I have to sed the makefile Jul 16 14:32:47 Stygia: having a variable won't fix that though Jul 16 14:33:31 Stygia: I would recommend filing a bug with upstream to make it not hardcode /usr/local Jul 16 14:45:05 bluelightning, Wouldn't it? Wouldn't it allow me to ensure /usr/local didn't get used directly? Jul 16 14:45:14 bluelightning, Like with the other variables. Jul 16 14:45:48 Stygia: just having the variable wouldn't affect building your recipe, no Jul 16 14:46:26 Stygia: you'd have to pass in the alternative path somehow, and if that's possible you can just do that now Jul 16 14:48:36 bluelightning, Ah. Well fair enough. AFAIK it isn't, we're SED'ing like crazy here. Jul 16 14:52:23 Stygia: this doesn't make sense though, when I install these perl libraries on my normal distro, they certainly do not go into /usr/local Jul 16 14:52:47 bluelightning, Hmm. DBD::SQLite? Jul 16 14:53:10 bluelightning, This _is_ the first CPAN module that has been bothering me about this particular thing, though. Jul 16 14:53:29 bluelightning, Which is weird, since its' by Stephen Mueller, one of the 3 people who apparently wrote like 80% of all the modules on metacpan. :P Jul 16 14:53:39 Gisle Aas being the absolute king. Jul 16 14:55:09 Stygia: http://pastebin.com/UpWPRPUY Jul 16 14:55:43 (this is on Ubuntu) Jul 16 14:57:55 bluelightning, Well... the Makefile here does explicitly mention /usr/local/, and it does try to go there Jul 16 14:58:07 bluelightning, But I fixed it with these sed's though: sed -i 's:\/usr\/local:${D}/usr/local:g' Makefile Jul 16 14:58:07 sed -i 's:\/usr\/share:${D}/usr/share:g' Makefile Jul 16 14:58:35 Stygia: I'm tempted to suggest looking at how Debian / Fedora /etc. build this specific piece of software Jul 16 14:59:17 YPTM: AlexG here Jul 16 14:59:22 anybody on the bridge? Song_Liu ? Jul 16 14:59:33 bluelightning, But now I get: http://pastebin.com/5UK1ypXw, which I thought I could resolve by making a recipe for perl-module-extutils-typemap, which I did and added to RDEPENDS_${PN}, but no luck. Jul 16 14:59:36 YPTM: FYI, I'm going to be a few minutes late, but I will be on the call Jul 16 15:00:35 YPTM: Scott Rifenbark joined the call Jul 16 15:01:06 YPTM davest is in the house Jul 16 15:01:18 YTPM: Saul is here Jul 16 15:01:25 YPTM: Tom Z here Jul 16 15:01:27 YPTM: polk is here Jul 16 15:01:33 YPTM: Welcome to the technical team meeting, please let me know who's on the bridge Jul 16 15:01:57 YPTM: belen joined Jul 16 15:02:33 YPTM: Michael here Jul 16 15:02:38 YPTM: Paul Eggleton joined Jul 16 15:02:40 YPTM: Scott Rifenbark joined call Jul 16 15:02:40 YPTM: nitin joined the bridge Jul 16 15:02:42 YPTM: I'm here Jul 16 15:02:51 YPTM: Tom here Jul 16 15:03:01 Matthew Weigel here Jul 16 15:03:15 YPTM: Any open? Jul 16 15:03:26 * fray is going on vacation and won't be around for a week or so.. Jul 16 15:03:29 YPTM: Dialing in Jul 16 15:03:45 YPTM: Cristian joined Jul 16 15:05:02 YPTM: The QA report for last week is ready: https://wiki.yoctoproject.org/wiki/WW28_-_2013-07-10-1_-_Weekly_Yocto_1.5_M2 Jul 16 15:05:12 YPTM: jzhang on the call Jul 16 15:05:25 YPTM? Yocto project team meeting? Jul 16 15:05:28 YPTM: Bruce Ashfield on the call. Jul 16 15:05:40 Stygia: yes Jul 16 15:05:51 Stygia: yes Jul 16 15:05:58 Thanks. :) Jul 16 15:06:02 open to the community -- 972-995-7777 x 42001078# Jul 16 15:06:18 Stygia: your welcome to join, fray beat me to the number Jul 16 15:07:02 YPTM: Ramana here Jul 16 15:07:04 zeddii: you on? Jul 16 15:07:14 YPTM: Bruce Ashfield on the call. Jul 16 15:11:30 Song_Liu: can you please post the link to the M2 release criteria? Having proxy problems Jul 16 15:15:24 YPTM: https://wiki.yoctoproject.org/wiki/Yocto_Project_v1.5_Status#Milestone_2 Jul 16 15:18:33 FYI, M2 suffered from toolchain transition issues and other issues.. which we've since corrected (I think).. so release note is "good enough" Jul 16 15:21:38 I agree an preview/RC0? is a good idea if they have the cycles while avoiding further M2 work Jul 16 15:25:36 YPTM: need to drop the call. See you next week! Jul 16 15:26:38 AlexG: sure. thank you Jul 16 15:29:15 Song_Liu: thanks for running the YPTM Jul 16 15:39:13 zibri, And as you wish. Jul 16 16:00:16 stygia: is your layer available somewhere? using the cpan bbclass seems to work-ish, but i lack DBI, so i can't Makefile.PL to generate a Makefile. (Also, the Makefile.PL exits with 0 in that case, so do_configure() doesn't fail :-( ) Jul 16 16:00:47 zibri, Not yet, but I could easily upload it (we do intend to open it). I could also just send you the DBI recipe, would that work? Jul 16 16:01:24 yes, probably, as long as it doesn't have any other dependencies not available :) Jul 16 16:04:40 zibri, http://pastebin.com/CbdfCft8 Jul 16 16:04:42 thx Jul 16 16:04:44 zibri, I think it _may_, hang on Jul 16 16:05:02 zibri, Huh, no. DBI has no dependencies on CPAN Jul 16 16:05:30 zibri, Ah, but DBD::SQLite does, one of which (Tie::Hash) I have no recipe for. Jul 16 16:05:50 heh, i'll just comment it out for the time being ;) Jul 16 16:06:09 zibri, IT's not mentioned in the recipe, though. Jul 16 16:06:22 zibri, I just RDEPENDS_${PN} all of them in my "main" recipe (the actual script all this depends on) Jul 16 16:06:27 zibri, I do intend to fix that. Jul 16 16:09:32 zibri, I'll be leaving for the day BTW Jul 16 16:09:38 zibri, Heh, I get paid to do this ATM. ^_^ Jul 16 16:09:42 heh Jul 16 16:10:02 i just love perl i guess :) Jul 16 16:10:34 zibri, Heh, perl is nice. Although... I will honestly admit I have lost my love for it after meeting Python. Jul 16 16:11:08 zibri, There are certain things just missing from Perl that, IMHO, make it fit for small-ish projects but not for larger programs. Jul 16 16:39:01 hey anyone about? Jul 16 16:39:04 I have a quick question regarding recipes for Yocto/BitBake Jul 16 16:39:16 go ahead, spit it out... Jul 16 16:39:20 I am building an application which requires a "normal" user account to run Jul 16 16:39:22 like apache would do, running in "www-data" for example Jul 16 16:39:37 yes... Jul 16 16:39:47 so how would I get my recipe to add this into the system Jul 16 16:39:50 you can inherit useradd Jul 16 16:40:07 genius.. so inherit useradd, then presumably some variable to specify the details? Jul 16 16:40:10 (I am very very new to this, btw) Jul 16 16:40:24 yup Jul 16 16:41:24 me too, in terms of the latest meta-* stuff Jul 16 16:41:25 mostly still use oe-classic Jul 16 16:41:25 I think I have it... Jul 16 16:41:41 in meta-skelleton is an example, duh - appologies for the dumb question :D Jul 16 16:41:55 RichBayliss: basiclly it takes useradd parameters so you can set everything, no dumb questions here! Jul 16 16:42:03 mr_science - thankyou very much, I will go try it Jul 16 16:42:14 cheers guys Jul 16 16:42:20 fingers crossed! :D Jul 16 16:45:42 it builds! time to do an image and test it on my RPi! Thanks guys - been smashing my head into the keyboard over this for a couple of hours :( Jul 16 16:46:58 i usually grep for examples... Jul 16 16:49:51 so is there any kind of yocto/poky setup guide for remote gdb? Jul 16 16:51:32 seeing a little bit of mailing list traffic... Jul 16 16:58:06 mr_science: there's nothing yocto/oe specific about it at all. install gdbserver, run it as you would on any linux systemm, connect with gdb as with any remote connection Jul 16 17:16:37 BRILL! Image compiled and working fine - thanks guys, much appreciated! Jul 16 18:13:29 kergoth: nothing about installing *-dbg packages or other setup stuff specific to OE? Jul 16 18:14:33 Good morning! Can I get some pointers patching an existing .bb file? This recipe has multiple patches already. I simply want to patch the .bb file from a .bbappend in my own layer Jul 16 18:15:20 just make a .bbappend and add what you need Jul 16 18:15:53 normally you can just d_foo_append() {} your own functions, etc Jul 16 18:16:13 mr_science: ya.. I realized as soon as i hit the enter key Jul 16 18:16:18 eg, do_install_append() { add your code } Jul 16 18:17:09 you can _append most things... Jul 16 18:17:42 mr_science: i'm bitbaking again with the list of files removed in my bbappend Jul 16 18:18:05 the list of patch files? Jul 16 18:18:34 is there a newer upstream version than the existing recipe? Jul 16 18:18:59 mr_science: hmm yes. The story is this - I have a recipe that contains some patches. And I simply want to patch the recipe's bb file. This recipe is in a different layer from my own. Jul 16 18:19:15 is there any other way to find dependencies chain instead of "bitbake -g -u depexp"? Jul 16 18:19:29 because the dependency in -u depexp is wrong.. Jul 16 18:19:37 ftonello: just bitbake -g will write dot files, so you can see the raw deps Jul 16 18:19:49 ftonello: but if you've found a bug in depexp, mail me with a way to reproduce :) Jul 16 18:19:53 libnss-mdns triggers avahi, but nothing triggers libnss-mdns :( Jul 16 18:20:03 its not a bug there I think.. Jul 16 18:20:17 adam__: what specifically do you want to change from the existing recipe? Jul 16 18:20:37 so avahi is been installed in my image without no one trigering it.. make any sense? Jul 16 18:20:58 several things could pull in avahi as a dep... Jul 16 18:21:30 pretty much anything that needs zereconf support... Jul 16 18:21:43 *zeroconf even Jul 16 18:21:47 mr_science: i need to modify the pkg_postinst behavior of the original recipe Jul 16 18:21:56 ftonello: you might find it coming in through an image feature that doesn't always appear in depexp Jul 16 18:22:15 mr_science: can i simply overwrite the entire pkg_postinst fuction from my bbappened? Jul 16 18:22:21 adam__: so you want to override the existing function? Jul 16 18:22:51 yes, that's what i was going to say... Jul 16 18:23:05 just replace the fucntion rather than _append to it Jul 16 18:23:56 ftonello: specifically, DISTRO_FEATURES defaults to "zeroconf", so packagegroup-base will depend on libnss-mdns -> avahi. Jul 16 18:24:11 ftonello: if you don't want any zeroconf support, drop that from your distro Jul 16 18:24:26 (yes, this needs to be better expressed in depexp) Jul 16 18:24:49 in this case, chasing the deps through "git grep" works Jul 16 18:25:21 libnss-mdns -> avahi -> packagegroup-base -> adds deps on DISTRO_FEATURE Jul 16 18:25:50 rburton: hmm Jul 16 18:26:10 mr_science: add dbg-pkgs to IMAGE_FEATURES if you want the debug symbols on the target Jul 16 18:26:22 rburton: how do I chase deps using git grep? Jul 16 18:26:43 rburton: also, how do I remove this from my DISTRO_FEATURES? because my distro features I use += operator Jul 16 18:26:54 ftonello: you'll have to set it outright Jul 16 18:27:14 use bitbake -e to get the full current value Jul 16 18:27:21 rburton: yeah Jul 16 18:27:25 thanks Jul 16 18:33:17 mr_science: cool i am replacing the functions entirely. Thanks! Jul 16 18:48:13 np Jul 16 20:22:49 Good afternoon everyone, I'd like to add a user for my desktop image. Should I create a recipe that uses adduser or is there a class that handles this in OE/Yocto? Jul 16 20:29:39 adam__: so atm the only way is to use a separate recipe; however patches are in review to allow adding a user outside of recipes for the kind of purpose you're presumably needing it for Jul 16 20:31:38 bluelightning: ahh great thanks! Jul 16 20:44:07 bluelightning: as in create a non-root user with a known password? Jul 16 20:50:18 mr_science: i hope so? Jul 16 20:53:20 mr_science: b1gtuna: correct Jul 16 20:54:10 i currently have that hacked into my custom_rootfs_postprocess Jul 16 20:54:44 would be much nicer with real bbclass support... Jul 16 20:55:30 i should go look at those patches... Jul 16 20:55:37 mr_science: do you mind sharing it with me? I can't login through GDM at the moment haha Jul 16 21:00:56 bigtuna: something like this => http://paste2.org/cEy60fh5 Jul 16 21:01:29 where is the name of your user account Jul 16 21:01:54 you'll also need to hash the password and replace Jul 16 21:02:46 er,, b1gtuna ... Jul 16 21:03:01 mr_science: nice Jul 16 21:03:48 mr_science: why could you not use the useradd class for this? Jul 16 21:04:01 that's inside a "fakeroot custom_rootfs_postprocess() {}" function in my image recipe Jul 16 21:04:08 it's oe-classic Jul 16 21:04:25 don't think arago has that Jul 16 21:04:37 mr_science: I think only maybe the sudoer's file would be an issue. Ah oe-classic, not oe-core. Jul 16 21:05:02 yup, in my rpi build i would definitely use it... Jul 16 21:05:29 * mr_science makes a mental note to add a non-root user to the rpi build Jul 16 21:48:04 bluelightning: Can you point me to the patch set that adds user creation? Jul 16 21:48:10 bluelightning: thanks Jul 16 21:48:13 b1gtuna: sure one sec Jul 16 21:48:47 bluelightning: :) Jul 16 21:50:00 b1gtuna: http://lists.openembedded.org/pipermail/openembedded-core/2013-July/081320.html Jul 16 21:50:19 b1gtuna: there's still some feedback on it though so it may need further work Jul 16 21:53:30 bluelightning: Ah I see. This will be really nice once it's completed. It's a little pain configuring users manually Jul 16 21:54:18 yeah, the image hack is exactly that . a hack... Jul 16 21:55:34 mr_science: btw are you using systemd on your images? Jul 16 22:01:32 not currently Jul 16 22:02:25 don't think it's in arago (oe-classic) and i specifically made my rpi image "light" so i don't have systemd in any features Jul 16 22:02:57 just old-school init scripts and update-rc.d Jul 16 22:07:20 mr_science: cool thanks Jul 16 22:10:15 it's all on github if you really want to see it... Jul 16 22:10:31 https://github.com/sarnold/meta-raspberrypi Jul 16 22:29:06 hmm, is hob stable? Jul 16 22:29:06 it freezes for me when selecting the architecture. Jul 16 22:29:25 oh, it just takes long time to parse the corresponding recipes. **** BEGIN LOGGING AT Tue Jul 16 22:42:58 2013 Jul 16 22:47:16 any reason why hob does not have its own repository? Jul 16 22:48:58 lpapp_: its code is heavily tied to the interface of bitbake Jul 16 22:53:17 bluelightning: means? Jul 16 22:54:04 lpapp_: meaning, when bitbake's internals change hob needs to be updated to match, and during development this happens frequently Jul 16 22:55:01 bluelightning: I am still not sure how that limits the repository separation. Jul 16 22:55:13 lpapp_: it's a lot easier for users grabbing the hob code to get something that correctly matches up if bitbake and hob are in the same repo Jul 16 22:55:42 bluelightning: why ? Jul 16 22:55:49 btw, users should use releases? Jul 16 22:56:04 lpapp_: a lot of users track master Jul 16 22:56:12 more than in other projects Jul 16 22:56:15 is that really the majority? Jul 16 22:56:24 usually, it is the very minor portion in projects, really. Jul 16 22:56:34 it's a significant percentage, I don't have numbers Jul 16 22:56:42 but even then, they are on their own. Jul 16 22:56:54 if it is documented what to fetch, it is probably not a big deal. Jul 16 22:56:57 or with git submodules. Jul 16 22:57:29 as a user, I think, it would actually make it clearer to use with proper separation. Jul 16 22:58:07 wrt master, we have been rarely left with broken bitabke and that couple of times it was on weekend ;) Jul 16 22:58:13 RP,^ Jul 16 22:59:25 I am not sure why unmodularization is thought as simpler rather than more complex. Jul 16 23:00:08 we've never provided a stable interface for UIs, and we do not wish to do so Jul 16 23:00:32 well, I would like to write a Qt/QML ui, but that kinda scares me. :) Jul 16 23:00:32 splitting out the UIs will just bring demands for that stable interface Jul 16 23:00:34 off* Jul 16 23:00:50 bluelightning: why? Jul 16 23:00:58 it should be an internal detail for ui releases. Jul 16 23:01:25 provided, users do not upgrade bitbake. Jul 16 23:01:33 just together with the ui Jul 16 23:01:40 lpapp_: cf the Linux kernel's internal interfaces, which are similarly guaranteed not to be stable Jul 16 23:01:56 IMO, this repository separation is irrelevant to stability. Jul 16 23:02:56 it would just help developers to have a clear separation. Jul 16 23:03:10 and those who wanna work with both, they can fetch both submodules with git, then. Jul 16 23:05:24 I'd be in favor of a proper UI API which would make it easier to create separate projects with separate executable bitbake interfaces, personally, but we've historically not done a good job of maintaining bitbake's APIs and API versioning Jul 16 23:06:06 I do not see what a UI api would mean or help with, in here. Jul 16 23:06:19 bluelightning: so what is the recommended way of creating a qbitbake frontend? Jul 16 23:07:36 lpapp_: as far as I'm concerned, there should be nothing stopping you from doing so; just use how hob works as an example Jul 16 23:08:54 bluelightning: meaning, putting stuff into the poky repository? Jul 16 23:09:18 lpapp_: well, that's the pattern we've been following up to now with UIs Jul 16 23:09:36 lpapp_: to be precise, the bitbake repository (poky just copies bitbake in almost verbatim) Jul 16 23:11:43 bluelightning: http://cgit.openembedded.org/bitbake/ this one or berlios? Jul 16 23:12:04 lpapp_: the one on openembedded.org Jul 16 23:12:25 berlios was closed if I recall correctly. Jul 16 23:13:51 bluelightning: hmm, I need to call python API from C++... Jul 16 23:14:02 or just calling the executable directly with the relevant options. Jul 16 23:14:20 from QML, to be precise, but I likely need some C++ helper. Jul 16 23:18:53 * bluelightning needs sleep... goodnight all Jul 16 23:56:23 If I change an RDEPEND of a package.. do I have to do bitbake -c cleansstate and rebuild it in order that an image that installs that package installs with the correct REPENDS? Jul 17 00:04:38 hmm, interesting that no one has written a ninja class yet. Jul 17 00:05:06 ftonello: no, if you are using dylan or master, it will notice the signature difference and compute the correct dependencies when the image is created Jul 17 00:09:15 sgw_: thanks Jul 17 00:34:05 is this statement, RDEPENDS_${PN}_append_pulseaudio-server = " package1 package2", going to append package1 and package2 to RDEPENDS_pulseaudio-server ? Jul 17 00:34:35 so basically I'm adding those new packages as pulseaudio-server package run time depend? Jul 17 00:39:25 ftonello: what you want to do is RDEPENDS_pulseaudio-server += "package1 package2" Jul 17 00:40:51 sgw_: but can I do that in any recipe? Jul 17 00:41:17 the ${PN} is not pulseaudio for example Jul 17 00:41:33 I want to add the extra dep from a nother recipe Jul 17 00:42:31 add the extra dep for what package? or only when pulseaudio-server is used? Jul 17 00:46:12 the thing is that Jul 17 00:47:06 add the extra dep for pulseaudio-server for example Jul 17 00:47:15 is that possible? Jul 17 00:47:27 like, if I install this package, add this dep to pulseaudio-server as well Jul 17 00:47:39 ftonello: from another recipe or a bbappend? Jul 17 00:48:23 sgw_: another recipe Jul 17 00:50:02 ftonello: you could add a pulseaudio.bbappend to your layer that RRECOMMENDS the package your building, that will cause it to be installed if it's build and available in the package feed. Jul 17 00:50:59 ok.. i need to do in a bbappend then Jul 17 00:54:14 you could also try the RDEPENDS_pulseaudio-server_append = " pacakge1 package2" in your recipe, but I have not really seen recipes do that, so I don't really want to reccommend it. (remember the leading space, since append does not add spaces while += does). Jul 17 01:58:27 sgw_: yes.. i will test and see Jul 17 01:58:29 thanks Jul 17 02:00:46 ftonello: sure, thanks for those patches, they are appreciated Jul 17 02:50:14 after I add a new package in the local .conf , (for example add ethtool) how can I use bitbake to rebuild the new image including the new package? I tried "bitbake core-image-basic", but can't find the added package after the target boot Jul 17 02:53:34 what did you change in local.conf? Jul 17 02:53:43 bitbaking the image should be sufficient. Jul 17 02:54:41 add "CORE_IMAGE_EXTRA_INSTALL = "ethtool " Jul 17 02:56:10 how to check if any wrong configuration or typing in the local.conf with bitbake? **** ENDING LOGGING AT Wed Jul 17 02:59:59 2013