**** BEGIN LOGGING AT Tue Oct 04 02:59:58 2016 Oct 04 06:13:52 khem: my difficulty is that my user's application A is composed of their own libraries X, Y, and Z. They're used to modifying X, Y, and Z, installing the new versions to their local sysroot, building a new version of the application (against that same sysroot), and installing that application and the modified X, Y, Z libraries to a rootfs dir, then mounting that rootfs over nfs. Oct 04 06:14:25 The issue is that it doesn't seem like modifying the sysroot is really supposed to happen. Oct 04 06:14:35 Or is it? I guess I'm looking for guidance. Oct 04 06:15:23 [the application also happens to be installed to the sysroot, but that's not really something they rely on, it's just incidental] Oct 04 07:09:25 Hi. Is there a way to list extra packages I'd like built & added to the package repo, but not included in the image? Something like a CORE_IMAGE_EXTRA_NOINST ? Oct 04 07:23:34 I guess I can bitbake individual packages or package groups and update the index. Or I could have an "extras" image and bake that, but it might be nice to have it all in the one image recipe. Oct 04 08:00:51 probably easiest to go the packagegroup route Oct 04 08:10:14 ok, ta. Oct 04 08:36:20 When I add the archiver task gcc-source fails to unpack_and_patch with Applying patch 0001-gcc-4.3.1-ARCH_FLAGS_FOR_TARGET.patch Oct 04 08:38:16 |--- a/configure Oct 04 08:38:16 |+++ b/configure Oct 04 08:38:16 -------------------------- Oct 04 08:38:16 No file to patch. Skipping patch. Oct 04 08:46:23 hello Oct 04 08:47:09 what is the recomended way in yocto to manage dependencies between package version - not only names Oct 04 08:47:52 eg. my_package_1.0 depend on second_packege_1.1 or above? Oct 04 08:48:42 krzeminski: there are versioned dependencies: http://www.yoctoproject.org/docs/current/ref-manual/ref-manual.html#var-RDEPENDS Oct 04 08:50:48 ernstp: did not expect that this is so simple Oct 04 08:50:54 ernstp: thanks Oct 04 08:51:26 haven't tested it :-) Oct 04 08:53:53 no problem, will test it Oct 04 08:54:09 just I have not found this in the docs Oct 04 12:29:47 Hi! Has anyone got an idea how to debug strange segfaults with Qt5.5.1 on an i.mx6 system within libGLES. See: https://paste.ee/p/9kEJY. Oct 04 12:36:27 hello everybody. How can I change the root password in my generated linux system? Oct 04 12:44:08 HyP3r: if you're just developing you can set IMAGE_FEATURES += "debug-tweaks" to log in as root without a password Oct 04 12:46:03 frsc: have you allocated memory for the graphics card? latest drivers for everything? Oct 04 12:48:12 ernstp: I'm using FSL Community Branch Jethro. I have never changed anything regarding memory allocation, but I'll check this... Oct 04 12:49:41 frsc: if your kernel CONFIG_CMA_SIZE_MBYTES is too small it wont work. You could try running Qt with env var QT_EGLFS_IMX6_NO_FB_MULTI_BUFFER=1 Oct 04 12:51:06 That should use less mem, so it might run then. Not the proper solution, but good for verifying. Oct 04 12:54:37 ernstp: I'm /finished/ with debugging and want now that my rootfs has a root password set Oct 04 13:02:10 HyP3r: you could use `ROOTFS_POSTPROCESS_COMMAND` in `your-image.bb` Oct 04 13:03:53 Hi all. How can i instruct a recipe to compile using c++11 using a bbappend? i tried BUILD_CXXFLAGS += "-std=gnu++11" but that does not work Oct 04 13:04:35 Oh i see, it's target_cxxflags, right? Oct 04 13:23:18 jubr: I have used this here: https://wiki.yoctoproject.org/wiki/FAQ:How_do_I_set_or_change_the_root_password Hope this is correct Oct 04 13:24:55 HyP3r: that looks fine, haven't tried it myself though Oct 04 13:41:00 how does bbappend work? For example I have multiple bbappend files for one package in the several meta layers, which one has highest priority? Oct 04 13:41:15 depends on the layer priority Oct 04 13:41:36 try `bitbake-layer show-appends` Oct 04 13:42:25 Is there any good resource on how to set up remote debugging for software that was build using bitbake? Oct 04 13:42:31 I'm having issues getting it to work in eclipse Oct 04 13:44:14 jubr: so, if BBFILE_PRIORITY_layername has priority 1, the directives from bbappend file for a package will be taken from this layer, correct? Oct 04 13:48:16 bbappends across multiple layers can apply Oct 04 13:49:40 0x4: I'm not sure, but I think higher level is more important, my `bitbake-layers show-layers` did not help much there either Oct 04 13:50:49 In general, if you add a layer yourself, chances are you want to override stuff from the other (poky) layers, so you probably want to give it the highest prio Oct 04 13:51:28 CTtpollard: ok, if I rewrite do_install function in bbappend file in several layers, which one has the highest priority? Oct 04 13:52:05 Determined by the output of `bitbake-layers show-layers` Oct 04 13:52:17 as in: it will use that order Oct 04 13:53:24 random-google-session: https://lists.yoctoproject.org/pipermail/yocto/2014-November/022382.html -- first lines confirm: higher-prio-is-more-important Oct 04 13:53:41 Ox4: why would you completly rewrite do_install in different layers? Oct 04 13:54:27 Anticom: good q. 0x4: Are you maintaining multiple architectures in multiple layers? Oct 04 13:54:58 jubr: that's the smell i wanted to point out ;) Oct 04 13:55:45 Anticom, jubr, no. Just interesting Oct 04 13:56:53 Ox4: Well in general bbappends should not completly redefine a task but only _append to it Oct 04 13:57:00 true Oct 04 13:59:05 understood, thank you Oct 04 13:59:40 jubr: I have 320 MiB of CMA and it was running with this setting with Dizzy-Branch and Qt5.3.2, now with Jethro and Qt5.5 I'm getting these segfaults. Oct 04 14:00:08 hello Oct 04 14:00:25 is there a method to use a layer with the exception of one bbappend? Oct 04 14:00:45 jubr: But it really looks like a memory issue, because the segfault happens in libGLES at different lines. Oct 04 14:01:36 c0rnel: Yes, exclude the pattern with BBMASK Oct 04 14:02:17 sveinse, can i say something lie BBMASK += "myrecipe.bbappend" ? Oct 04 14:02:37 The interface has changed in recent versions of poky, but the old syntax is BBMASK .= "|.*/meta-custom/recipes-packages/networkmanager" Oct 04 14:04:14 sveinse, thank you Oct 04 14:04:20 http://downloads.yoctoproject.org/releases/eclipse-plugin/2.2/neon described in http://www.yoctoproject.org/docs/latest/mega-manual/mega-manual.html#neon-new-software seems to be broken (404) Oct 04 14:04:23 c0rnel: np Oct 04 14:04:23 Any suggestions? Oct 04 14:06:33 anyone who has some experience with cef and their new GN system in yocto? Oct 04 14:09:01 I have one more question: I would like to import my own database schema dump to mariadb after intallation, how can I do that using bbappend file? Oct 04 14:10:50 Ox4: Probably belongs to do_install task Oct 04 14:12:39 I see the pkg_postinst_${PN}-server function in the mariadb.inc Oct 04 14:13:31 post-package-install would work too i guess Oct 04 14:14:12 so just append to this task (pkg_postinst_mariadb-server_append() { /* import schema dump */ } Oct 04 14:14:36 but I cannot use mysqlimport because the server is not running yet Oct 04 14:14:40 rburton: Can you help with the broken eclipse link? Oct 04 14:14:59 Ox4: can't you just copy the files to sysroot? Oct 04 14:15:08 i'm not sure whether anything else is needed Oct 04 14:16:57 I'm trying to write a recipe for canfestival, and I get this linker error: https://bpaste.net/show/3a0319b4a893 Oct 04 14:17:47 Does someone here happen to know what this can be? I've tried quite a few arguments/permutations in devshell, but I can't get it to link Oct 04 14:21:58 sveinse: probably forgott to link against some library. Ask the guys over at ##c or ##c++ Oct 04 14:22:59 Anticom: come a bug please Oct 04 14:23:40 rburton: ?? I've looked into 2.1.1 etc. directories but there was no "neon" subdirectory Oct 04 14:24:19 Anticom: literally no idea re eclipse, but if there's a problem then file a bug and CC tim orling Oct 04 14:24:58 rburton: Not sure whether it's a bug. I'm guessing the latest docs is just ahead of the actual builds (?) Oct 04 14:25:02 Anticom: do you mean the datebase? Oct 04 14:25:08 Ox4: yes Oct 04 14:25:33 I think the mysqlimport is the better approach Oct 04 14:25:48 Ox4: I don't know mariadb but usually they just store their stuff in some rather portable files. So if you just copy the files to your sysroot it should work out just fine Oct 04 14:28:13 Anticom: yes, I can create the dump file of the database but I would like to store this dump back after the building mariadb package Oct 04 14:32:05 Anticom: I doubt it. Upstream code works when cross compiling under other build envs. It's when I take it to Yocto it stops. This "libc.so.6: error adding symbols: DSO missing from command line" which perplexes me Oct 04 14:32:47 I've checked that the missing symbol is indeed present in libc.so.6 (albeit under a weak symbol) Oct 04 14:34:13 sveinse: inspect log.do_compile in path/to/build/tmp/work////temp and check whether everything is build/linked as you do it on your dev machine or wherever it does work Oct 04 14:35:21 sveinse: also you could add TARGET_LDFLAGS += "-Wl,-t" to get a trace of what libs exactly are linked Oct 04 14:35:47 (not sure, whether you need the -Wl, there or not) Oct 04 14:36:58 Anticom: I'm am manually working in devshell and testing various options (including -t) Oct 04 14:37:24 It does not reveal anything more than the above failure Oct 04 14:38:18 I was inclined towards adding -lc, but then it fails on missing _start, so apparently it seems the linker startup has gone haywire Oct 04 14:40:59 hah, got it! Oct 04 14:41:36 upstream is using ld for linking. Change it to gcc and it links properly Oct 04 14:42:32 marka: hey I've tried extrausers.bbclass but I get this error: opkg_install_cmd: Cannot install package racechip-password Oct 04 14:42:40 marka: http://pastebin.com/iDKcCBCq <- this is the recpie Oct 04 14:42:58 marka: I really don't know why, and I don't know how to troulbeshoot that Oct 04 14:43:59 * marka looks Oct 04 14:48:25 HyP3r: the -P should be lowercase. no? Oct 04 14:48:35 sveinse: #blametheprogrammer :D Oct 04 14:49:15 outside of that I would attempt to get any additional error logging since "Cannot install package..." is not all that useful Oct 04 14:49:47 Anticom: heh, yes. #they are to blame, we're perfect. #pridebeforethefall :P Oct 04 14:49:50 possibly get a system up and running with smartpm and attempt to install the package, as opposed to including it in rootfs creation Oct 04 14:51:00 but first look at the capital P, pretty sure that should be p Oct 04 14:53:31 marka: this 's/-P/-p/g' could be the solution but I'm confused take a look at that: https://wiki.yoctoproject.org/wiki/FAQ:How_do_I_set_or_change_the_root_password Oct 04 14:54:03 YPTM armin is on Oct 04 14:54:24 marka: how can I give you more information like that? Oct 04 14:54:30 marka: or what log file do you want Oct 04 14:55:19 YPTM: Ready-Access Number: 8007302996 Access Code: 2705751 Oct 04 14:56:05 HyP3r: ya, I would say a quick glance the wiki might have a typo for /P/p/ Oct 04 14:56:24 EXTRA_USERS_PARAMS is calling the commands Oct 04 14:56:46 and usermod AFAIK only has '-p' Oct 04 14:56:59 YPTM: Stephen Joined Oct 04 14:57:39 HyP3r: as for getting more information I can attempt to reproduce your issue here Oct 04 14:58:02 but rather than wait for me I definitely encourage you to see what you can get Oct 04 14:58:39 again, I think attempting to install with smart on a running system will yield more useful information, gut feeling Oct 04 14:59:10 YPTM: Joshua joined Oct 04 14:59:27 This upstream code does not separate on ${bindir} and {libdir}. Is there a common prefix variable I can use? Or do I strictly need to patch its build system here? Oct 04 15:00:44 sveinse: have a look at bitbake.conf. ${prefix} is likely what you want. Oct 04 15:00:45 why do people keep adding a redundant [oe-core] to their patch submission subject lines? Oct 04 15:00:57 it's pointless, the list adds it anyway, and the readme in the layer says nothing about that crap Oct 04 15:02:02 that sort of thing is the reason dvhart wrote tools for creating and submitting patches Oct 04 15:03:36 marka: I guess the problem is: Unknown package 'racechip-password'. Oct 04 15:03:56 marka: a package with no files is no package... I added this extrausers to another packages which is also delivering files Oct 04 15:05:23 HyP3r: I was curious about that Oct 04 15:06:23 rburton: thanks Oct 04 15:06:26 Are there any examples of any recipes that generates both target packages as well as native packages from the same source/recipe? Oct 04 15:06:30 you can probably confirm that easy enough by looking at the packages-split directory Oct 04 15:06:35 YPTM is over. Oct 04 15:06:44 wow all ARs are assign to ross Oct 04 15:07:13 best way to keep the meeting short "ross said he'll do it" Oct 04 15:07:14 hang on what Oct 04 15:07:16 HyP3r: anyways, lots of info out there for you to debug work around 'empty' package rules Oct 04 15:07:31 sveinse: everything that uses BBCLASSEXTEND Oct 04 15:07:37 oh it was YPTM Oct 04 15:07:59 oh, sorry guys Oct 04 15:08:24 i guess i should read the minutes Oct 04 15:09:29 marka: ok I run it now seems to work :D Oct 04 15:09:39 good stuff Oct 04 15:50:35 What's the right way for a layer to provide a configuration file intended for use with populate_sdk? It looks like meta-qt5 is providing some bbclasses to do this (populate_sdk_qt5.bbclass & populate_sdk_qt5_base.bbclass), is that the right way to handle it? Oct 04 15:55:19 fishey1: that depends on what the file is Oct 04 15:55:23 RP1: ping Oct 04 15:55:38 kergoth: setting environment variables for qmake to build for target Oct 04 15:55:51 dropped in environment-setup.d Oct 04 15:56:13 just write a new recipe, either nativesdk or target, depending on which sysroot you want to have the environment-setup.d in Oct 04 15:56:27 then add that to TOOLCHAIN_HOST_TASK or TOOLCHAIN_TARGET_TASK, respectively Oct 04 16:00:02 can I print out my custom variables definition from bbappend file to the output? Oct 04 16:00:40 that makes sense. so instead of using ${SDK_OUTPUT}${SDKPATHNATIVE}/environment-setup.d (which is what qt4 is doing right now in a `create_sdk_files_apped`, just have a recipe with a do_install that uses ${D}/environment-setup.d (and add to the TOOLCHAIN_*_TASK as appropriate)? Sounds... sane. Is there a guideline on which sysroot's (host or target) environment-setup.d should be used? I'm leaning towards Oct 04 16:00:41 Ox4: just use bitbake -e, the same way you'd examine any recipe or metadata variables Oct 04 16:00:43 TARGET as this config references some paths in target. Oct 04 16:01:27 fishey1: yeah, it really depends on what variables you need. if you need host/target stuff, you won't have access to those in a nativesdk recipe Oct 04 16:02:25 kergoth: nice. Thank you Oct 04 16:03:02 specifically bitbake -e yourrecipe Oct 04 16:03:06 np Oct 04 16:05:55 kergoth: hi Oct 04 16:06:26 RP1: any tips on debugging a taskhash mismatch encountered by an automated build? the only sigdata in stamps is one of the two, so can't compare Oct 04 16:08:29 fun Oct 04 16:11:37 hmmm Oct 04 16:12:28 if I create a variable outside any function in the bbappend file like MYVAR = "SELECT * FROM database;" Oct 04 16:12:45 should I use it in the function like ${MYVAR}? Oct 04 16:12:58 that's how bitbake variables work, yes Oct 04 16:13:02 see the bitbake user manual Oct 04 16:13:25 Ox4: https://www.yoctoproject.org/docs/2.2/bitbake-user-manual/bitbake-user-manual.html, https://www.yoctoproject.org/docs/2.2/bitbake-user-manual/bitbake-user-manual.html#basic-syntax Oct 04 16:14:19 very helpfull, thank you again :) Oct 04 16:21:14 kergoth: ok, so what if the sdk needs a symlink from SDKPATHNATIVE to SDKTARGETSYSROOT? This seems like it's probably completely wrong to do, yet meta-qt4 does it today. I suppose I'll keep that around in a bbclass used for the sdk and look at the root cause later. Oct 04 16:22:06 kergoth: you probably need to have it dump the stamps at parse time Oct 04 16:22:39 kergoth: The line "Slow but can be useful for debugging mismatched basehashes" in siggen.py Oct 04 16:23:22 kergoth: there is an open bug about this, is performance verses debugging issue :/ Oct 04 16:23:49 okay, thanks. wonder if we should add an argument for that.. should we have too many args already.. Oct 04 16:57:16 guys, what is wrong with this bbappend file: http://dpaste.com/3F1963C ? Oct 04 16:57:47 The mariadb package was built successfully, but I don't see my database :-( Oct 04 17:00:51 sorry, this one is correct version: http://dpaste.com/30D13HH Oct 04 17:06:50 and another question: how to make escaped quotes in the recipe? The "\"" is enough? Oct 04 17:07:38 So looking at the rules you have there, it seems to me that the mysql -e command is unlikely to do what you want. Oct 04 17:08:11 Because there's no quoting involved, so you end up just invoking: mysql -e CREATE DATABASE [...] Oct 04 17:08:32 In particular, the semicolon after ${DATABASENAME} would be the end of the mysql command. Oct 04 17:08:45 yes it is Oct 04 17:08:53 So it would then execute GRANT USAGE ON *.* [...] as a new command. Oct 04 17:08:55 that's is why I have asked about quotes Oct 04 17:09:14 I think what you might want is mysql -e "${SQL_CREATE_DATABASE} ... " Oct 04 17:09:20 with the quotes in the shell command. Oct 04 17:09:30 But the other thing is... It seems to me that this should have *failed*. Oct 04 17:09:49 So if you didn't get a failure when the shell tried to invoke a nonexistent command named GRANT, it's probable that the append didn't work as expected. Oct 04 17:19:53 seebs: what about "< database_dump.sql", is it correct? Oct 04 17:21:04 Possibly? You might need to specify the path you installed it in. Oct 04 17:21:25 Oh, hmm. That's a postinst. That may not run until the system boots, and then you'd get a log from it. Oct 04 17:24:43 hm, interesting Oct 04 19:17:59 Hello - what would I add to my DEPENDS= line to compile against libbfd? Oct 04 19:18:10 binutils? Oct 04 19:19:53 bluelightning: thoughts on changing bitbake-layers show-layers output more script-friendly? either automatically based on stdout/stderr being a tty, or an argument? i.e. tab-seperated, no header, rather than space separated with a header and header separator Oct 04 19:28:34 Circuitsoft: that sounds correct Oct 04 19:29:15 kergoth: I'd be in favour... the headers there and for some of the other commands have always been a bit bothersome for scripts Oct 04 19:31:32 bluelightning: The package in question (ipxe) uses straight Make, rather than configure. Does it need to pull an include path from the environment somehow? Oct 04 19:31:45 With binutils in DEPENDS, it still can't find bfd.h Oct 04 19:32:18 Circuitsoft: depends entirely on what the makefile is doing Oct 04 19:32:41 CC := $(CROSS_COMPILE)gcc Oct 04 19:32:48 Hmm. Oct 04 19:32:49 bluelightning: k, will have to give it some thought re the semantics, ideally we want to look good for the user, but still be nicely scriptable. thanks for the input. Oct 04 19:33:46 aside, just found another situation where you can use bitbake-layers to get yourself into a situation you can't use it to get out of. missing layer dependency = bblayers no longer parsable. hrmph. might add a dep check to add-layer to prevent a user from getting into that situation Oct 04 19:33:50 hrmph Oct 04 19:34:10 Circuitsoft: you know there's a recipe for ipxe in meta-virtualization, right? http://layers.openembedded.org/layerindex/recipe/33112/ Oct 04 19:34:28 I'm using that. Oct 04 19:34:32 kergoth: yes, I'd noticed that too... unfortunate side-effect Oct 04 19:34:45 It fails when building util/elf2efi32 Oct 04 19:34:47 Circuitsoft: ok, great, just checking Oct 04 19:35:18 Circuitsoft: yes, that CC value is not what you want, it should be using the CC value from the environment Oct 04 19:35:41 It's hardcoded in the Makefile. I'll have to come up with a patch. Oct 04 19:36:05 Circuitsoft: well, maybe not Oct 04 19:36:22 Acutally, I think it's using HOST_CC which is hardcoded as gcc Oct 04 19:36:25 Circuitsoft: in previous versions that was forced, we always passed -e to make, as of the last major release we stopped doing that Oct 04 19:36:46 EXTRA_OEMAKE Is our friend. explicitly pass it in Oct 04 19:37:06 right, that's what I was alluding to Oct 04 19:37:52 So, installing libiberty-dev and binutils-dev on my host machine fixed it. Oct 04 19:37:57 The recipe already requires binutils-native. Oct 04 19:38:13 Where should it find the native dev packages? Oct 04 19:39:50 in the native sysroot Oct 04 19:40:03 which it should already be looking in... Oct 04 19:43:38 Looks like it should be using BUILD_CFLAGS for native programs? Oct 04 19:48:38 Circuitsoft: you should use ${CFLAGS} and allow the native.bbclass to set that var as appropriate Oct 04 19:48:43 iiuc Oct 04 19:49:15 fishey1: It's not a native package, but it has one native executable it has to build, to finish the target. Oct 04 19:51:29 in that case, BUILD_CFLAGS may work, though I'd also consider (and perhaps recommend) depending on a foo-tools-native package instead of having the target package build native tools. I suppose it would depend on whether the native tools are normally installed. u-boot does something like this. Oct 04 19:52:04 s/may/should/ Oct 04 19:52:36 The tool in question is elf2efi32 and elf2efi64 Oct 04 19:54:22 Okay. So, the magic invocation is: make HOST_CFLAGS="$BUILD_CFLAGS $BUILD_LDFLAGS" Oct 04 19:57:06 you'll want to obey BUILD_CC for HOST_CC too Oct 04 19:57:10 if youaren't already Oct 04 19:57:46 So far, BUILD_CC seems to be just "gcc", but I should add that too. Oct 04 19:58:35 it is, but isn't guaranteed to be. in some cases folks override it to e.g. force use of a compatible gcc version on distros running the bleeding edge Oct 04 20:28:17 is it not supported, to specify a branchname as SRCREV ? Looks like nothing will trigger a check-if-iI-need-to-fetch Oct 04 20:31:06 yann: in that case you want to use branch= and SRCREV = "${AUTOREV}" (or specify a specific revision on that branch, if you prefer) Oct 04 20:31:27 where ;branch= goes on the end of the SRC_URI entry Oct 04 20:31:57 I had the branch already AUTOREV was what I was missing Oct 04 20:32:03 thx Oct 04 20:33:06 wouldn't it be a good place to have AUTOREV mentionned in the " Fetching Code" part of the dev-manual ? Oct 04 20:36:46 maybe it's just missing an xref to "using an external scm", in fact Oct 04 20:46:23 yann: hmm, yes, we should try to cover that somehow - the external SCM section probably needs some reworking too Oct 04 20:48:00 yes - I'm currently trying to find how I activate AUTOINC to be handled, but just can't find it anymore Oct 04 20:50:00 ah, found it Oct 04 20:54:11 yann: er - AUTOINC is just a placeholder, you shouldn't need to "activate" it Oct 04 20:54:24 yann: it gets replaced when the packages are created Oct 04 20:54:49 yes, I meant using the PR service - if I had the name in mind I would have got to it immediately :) Oct 04 20:54:56 ah, right, I see Oct 04 20:55:21 we really ought to be clearer that you need to enable the PR service to get properly incrementing versions Oct 04 20:55:41 maybe we should just turn on a local one by default Oct 04 20:56:07 btw, how open is 2.2 for changes ? Oct 04 20:56:59 I have an embryo of a useful addition, allowing to produce RO images in some cases where qemu does not fully support the target Oct 04 20:58:16 "just" need to find a reliable test to activate this in preference to qemu in the postinst-intercepts Oct 04 21:00:57 in my case, working with an amdfalconx86, which defaults to using the dbfp4 tune, which is not supported in qemu - but for which running native binaries in postinst-intercepts ought to be good on an x86_64 host Oct 04 21:01:43 yann: at this point that will probably have to wait for 2.3, but you could still RFC it now Oct 04 21:02:49 just not sure how to test for the compatibility in the postinst-intercept script. Maybe it would be better to have the "compat" checked at OE level, and the intercept just check that Oct 04 21:05:04 ok, will do that Oct 04 21:05:06 seebs: did you see the renameat patch for pseudo? any thoughts on that?(https://bugzilla.yoctoproject.org/show_bug.cgi?id=10349) Oct 04 21:05:08 Bug 10349: normal, Undecided, ---, mark.hatle, NEW , renameat() loses extended attributes Oct 04 22:17:26 Does the eclipse yocto plugin allow you to configure your yocto image? Oct 04 22:21:17 gnac: it really is only for editing, compiling, debugging applications right now Oct 04 22:21:45 gnac: for image configuration one option would be toaster Oct 04 22:22:25 moto-timo: okay, thanks. Oct 04 22:22:57 Second question, is it possible to generate an SDK with crosscompilers for windows hosts (to a yocto target)? Oct 04 22:23:13 gnac: you need to use meta-mingw Oct 04 22:23:14 Don't flame me on needing a Windows host, its an Enterprise requirement. Oct 04 22:23:24 gnac: otherwise, you want to use CROPS Oct 04 22:23:56 are those recipes/options? or separate downloads? Oct 04 22:23:56 gnac: https://github.com/crops/docker-win-mac-docs/wiki Oct 04 22:24:55 it's rather bleeding edge, but it works Oct 04 22:25:17 It will be more fleshed out in 2.3 Oct 04 22:25:35 I recognize the need to build the yocto "OS" image in a linux host. There is an enterprise packager of yocto that also generates an SDK for use on compiling for linux yocto targets from within windows. Oct 04 22:26:06 so either you builds the windows tools on meta-mingw or you need another hammer (CROPS) Oct 04 22:26:26 Okay, great,, I'll look at meta-mingw, thanks! Oct 04 22:26:35 sure **** ENDING LOGGING AT Wed Oct 05 02:59:58 2016