**** BEGIN LOGGING AT Thu Mar 10 02:59:58 2016
Mar 10 07:24:23 <johnliu> Has anyone tried adding normal users in the yocto image?
Mar 10 07:25:07 <johnliu> I cannot succeed
Mar 10 07:26:18 <johnliu> I think the useradd command under chroot condition failed to work
Mar 10 07:32:16 <bluelightning> johnliu: adding through what method?
Mar 10 07:35:38 <LetoThe2nd> johnliu: isn't http://www.yoctoproject.org/docs/latest/mega-manual/mega-manual.html#ref-classes-useradd the recommended method nowadays?
Mar 10 07:38:25 <johnliu> I was using the "extrausers" in the local.conf
Mar 10 07:39:36 <johnliu> Just inherit the class and append the command to EXTRA_USERS_PARAMS
Mar 10 09:37:28 <kageja> hi everyone GoodMoorning, Have nice Linuxing :) :)
Mar 10 09:44:10 <AndyP_> Anyone know what the best starting point would be for a Texas Instruments DM355 project?
Mar 10 10:19:37 <Anticom> Hi all. I've just switched to systemd for my image. Most stuff looks just fine except sshd doesn't want to start. Before when using sysvinit, it generated the host files on first boot from how i understand it. Now when i do '/usr/sbin/sshd -d' i get http://pastebin.com/wp0RdYLp
Mar 10 10:19:57 <Anticom> Is this a common issue or is there something wrong with my build config?
Mar 10 10:23:06 <mborzecki> Anticom: imo there should be a helper service that starts before sshd and generates keys
Mar 10 10:23:17 <mborzecki> that's how it works for dropbear
Mar 10 10:23:53 <boucman_work> shouldn't it be implemented as "first boot do_install" ?
Mar 10 10:25:12 <Anticom> mborzecki: we're using openssh
Mar 10 10:25:26 <Anticom> And i just don't know how to track down the issue any further
Mar 10 10:25:36 <Anticom> Any recommendations?
Mar 10 10:27:18 <mborzecki> w8, why are you starting sshd by hand?
Mar 10 10:28:36 <mborzecki> it's socket activated, it'll be started automatically by systemd when you connect to 22
Mar 10 10:28:45 <mborzecki> port 22
Mar 10 10:29:20 <boucman_work> (I meant do_pkg_postinst)
Mar 10 10:29:50 <Anticom> [my-pc]$ ssh -vT targetdevice gives me ssh: connect to host <ip> port 22: No route to host
Mar 10 10:29:56 <Anticom> and i don't see any sshd being started
Mar 10 10:30:12 <Anticom> neither do i see it in ps after booting the device
Mar 10 10:30:32 <boucman_work> can you connect to your device in any way (no route to host isn't "connection refused" iiuc
Mar 10 10:30:35 <mborzecki> Anticom: that's a different thing :) try to ping your board first, i'm guessing it will fail too
Mar 10 10:31:21 <Anticom> mborzecki: ping is failing aswell, but what does that tell me?
Mar 10 10:31:55 <boucman_work> AndersD, that your card's network is not properly configured
Mar 10 10:32:09 <Anticom> .__.
Mar 10 10:32:14 <Anticom> ps aux | grep sshd
Mar 10 10:32:22 <Anticom> there's no sshd running on my device
Mar 10 10:32:43 <mborzecki> as I said, sshd will not be started unless something connects to tcp port 22
Mar 10 10:32:44 <Anticom> before i switched to systemd it worked so i don't think it's my NIC being broken
Mar 10 10:32:54 <boucman_work> can you connect to your board via serial (or a screeen + keyboard) ?
Mar 10 10:32:57 <mborzecki> if you cannot reach the device, there will be no connection obviously
Mar 10 10:33:28 <boucman_work> Anticom, probably not, but we need to connect to the board to diagnose...
Mar 10 10:34:03 <Anticom> boucman_work: i am connected via serial... Also i think i have to apologize mborzecki : ifconfig doesn't show my eth0 anymore :O
Mar 10 10:34:29 <Anticom> i just the the lo
Mar 10 10:34:44 <boucman_work> Anticom, let me guess... what does "ifconfig -a" say ?
Mar 10 10:35:00 <boucman_work> network interface names are different when using systemd
Mar 10 10:36:19 <Anticom> oh i see. We where using this eth0:foo thing before. I'm not into networking that much. However the interfaces aren't there anymore as i've known them
Mar 10 10:36:29 <Anticom> and there's no /etc/network/interfaces anymore
Mar 10 10:37:47 <Anticom> Think there would be quite some fixing to be done in order to being able to use systemd for our image :/
Mar 10 10:37:53 <mborzecki> hmm i thought we don't ship persistent interface naming rules by default
Mar 10 10:38:33 <Anticom> also suddenly my device wants nfsd since i've switched to systemd. Think i'll just roll back for now
Mar 10 10:41:08 <boucman_work> Anticom, did you run ifconfig -a ?
Mar 10 10:41:19 <Anticom> boucman_work: yup
Mar 10 10:41:40 <Anticom> got eth1, ip6tn10, lo and sit0
Mar 10 10:43:05 <Anticom> and eth1 isn't configured probably meaning no ip/mask set
Mar 10 10:43:33 <Anticom> so what's the equivalent to /etc/network/interfaces now?
Mar 10 10:44:41 <boucman_work> weird... that's a mix of persistant and non-persistant names
Mar 10 10:44:48 <mborzecki> Anticom: /etc/systemd/network/*.network files
Mar 10 10:45:53 <mborzecki> mborzecki: I usually add a default setup to devel/testing images https://github.com/open-rnd/meta-openrnd/tree/master/recipes-connectivity/network-profile-systemd
Mar 10 10:46:31 <mborzecki> Anticom: ^^
Mar 10 10:47:04 <Anticom> there are no *.network files
Mar 10 10:47:24 <boucman_work> yes, that's why your interfaces arn't configured...
Mar 10 10:47:45 <Anticom> jeez... I thought switching from sysvinit to systemd just changed an executable basically but my entire system is different now :/
Mar 10 10:48:01 <mborzecki> oh, and btw. networkd is not enabled by default, you need to set PACKAGECONFIG_pn-systemd_append = " networkd"
Mar 10 10:48:18 <boucman_work> Anticom, long term, it's quite cool, but yes. there is some work to be done
Mar 10 10:48:33 <LetoThe2nd> the decision for an init system should not be done easily, and especially not late in a project ;-)
Mar 10 10:48:43 <boucman_work> mborzecki, what's the default network handler when systemd is enabled, then ?
Mar 10 10:48:49 <Anticom> boucman_work: which takes time which we don't have right now i'm afraid :/
Mar 10 10:48:55 <Anticom> LetoThe2nd: agree
Mar 10 10:48:58 <boucman_work> I thought it was ifupdown just like with systemv
Mar 10 10:49:08 <mborzecki> boucman_work: imo there's none, you're free to choose what you want
Mar 10 10:49:35 <boucman_work> so no network out-of-the-box with systemd ?
Mar 10 10:50:04 <LetoThe2nd> AFAIK not without adding connman, nm, networkd, whatever.
Mar 10 10:50:16 <mborzecki> iirc networkd is built by default in upstream, but it's disabled in OE
Mar 10 10:50:52 <boucman_work> ok, there's probably something selected in core-image-* which would explain why i'm confused
Mar 10 10:51:37 <Anticom> And concerning my other question: I suppose systemd recipe requires nfs (at least by default) ?
Mar 10 10:51:43 <mborzecki> ifupdown probably, you should not have networkd unless it's enbled explicitly
Mar 10 10:52:26 <LetoThe2nd> why should systemd require nfs?
Mar 10 10:53:02 <boucman_work> mborzecki, there is still one mistery... if ifupdown was on the systemv image... why did it disapear when switching to systemd...
Mar 10 10:53:33 <Anticom> LetoThe2nd: dunno but suddenly systemd wants to mount nfsd (although it fails since the kernel module is missing)
Mar 10 10:53:43 <Anticom> But i can't see where else it's comming from
Mar 10 10:54:33 <mborzecki> boucman_work: ifupdown likely depends on sysvinit
Mar 10 10:54:46 <boucman_work> hmmm, that would explain it
Mar 10 10:55:31 <boucman_work> (totally unrelated) there seems to be a bug when using both rm_work and buildhistory on image recipes... the workdir is cleaned before buildhistory is run => buildhistory fails
Mar 10 10:56:30 <LetoThe2nd> heh, thats a funny one indeed.
Mar 10 10:57:08 <mborzecki> boucman_work: seems like it's an indirect dependency, ifupdown inherits update-rc.d, that in turn depends on initscripts, iirc initscripts are not installed when using systemd
Mar 10 10:57:41 <boucman_work> systemd has a compatibility layer, though... right ?
Mar 10 10:57:42 <LetoThe2nd> mborzecki: at least not if you actively prohibit it, as suggested in the yocto docs.
Mar 10 10:59:05 <mborzecki> LetoThe2nd: which I do https://raw.githubusercontent.com/open-rnd/meta-openrnd/master/conf/distro/openrnd-poky-systemd.conf :) not even sure if this is the canoncal way of switching distro to systemd
Mar 10 11:05:03 <Anticom> Well i was kind of counting on that sysvinit compatibility so we could gradually port to systemd's way of doing things
Mar 10 11:06:28 <Anticom> LetoThe2nd: Also i was wrong. nfs seems to have been in our old image already. How do i find out what's causing it to be in our image?
Mar 10 11:06:42 <Anticom> I'm not that fit with yocto tools yet when it comes to such tasks
Mar 10 11:07:19 <LetoThe2nd> bitbake -g depexp or such, maybe? google-fu to the rescue.
Mar 10 11:07:49 <Anticom> LetoThe2nd: is this the thing, that's giving me a couple of dot files?
Mar 10 11:09:06 <LetoThe2nd> see: https://wiki.yoctoproject.org/wiki/BitBake/GUI
Mar 10 11:10:54 <Anticom> Okay thank's
Mar 10 11:14:15 <Anticom> Just to make sure. So when i select nfs-utils in "Package" "Reverse Depends" is showing me, which packages depend on nfs-utils and hence cause it to be included in my build?
Mar 10 11:32:21 <Lucretia> Hi, gcc-cross has "--disable-bootstrap" on configure and calls "make all-host." I'm trying to enable Ada support it's failing as it's misconfiguring for some reason. gcc/ada/gcc-interface/Makefile is getting the following line: "tmake_file =  $(srcdir)/config/mips/t-mips $(srcdir)/config/t-slibgcc $(srcdir)/config/t-linux $(srcdir)/config/t-glibc ./config/mips/t-linux64" where the "./config/mips/t-linux64" should be "$(srcdir)/config/mips
Mar 10 11:32:22 <Lucretia> /t-linux64" - is there something I'm missing when gcc is configured and built in this way?
Mar 10 13:09:26 <Anticom> Can i get information about the overall size of all rpm's as well as their overall space requirement in rootfs? I know there where some files with some buildstats but i can't remember anymore where they where and what exactly they contain
Mar 10 13:09:44 <boucman_work> if I want to submit a patch for meta/ agains what branch should I work ? master or master-next ? (I havn't found anywhere in the wiki that explains the branch policy, but I might have missed it
Mar 10 13:10:51 <boucman_work> Anticom, you need to add INHERIT += "buildhistory" to your local.conf, rebuild the image, and you should have the info you are looking for in a buildhistory/ subdirectory of your build directory
Mar 10 13:13:03 <Anticom> Also i don't know how to get rid of nfs-utils entirely. I did PACKAGE_EXCLUDE += "nfs-utils" in my local.conf and now `smart install` is complaining: Computing transaction...error: Can't install packagegroup-core-full-cmdline-sys-services-1.0-r6@all: unable to install provider for nfs-utils: error:     nfs-utils-1.3.1-r0@cortexa9hf_vfp_neon is to be excluded
Mar 10 13:13:24 <Anticom> (smart install being performed by do_rootfs)
Mar 10 13:15:50 <Anticom> nfs-utils is not in the packages that are being installed. Moreover the only reverse dependency of nfs-utils is nfs-utils
Mar 10 13:15:56 <Anticom> So what's the issue there?
Mar 10 13:16:32 <boucman_work> you would need to remove nfs-utils from packagegroup-core-full-cmdline-sys-service's dependencies
Mar 10 13:17:04 <boucman_work> look for the corresponding .bb to see how it's added, then add the correct line to remove it in your local.conf
Mar 10 13:22:02 <Anticom> boucman_work: i tried that aswell. I've copied the recipe to our layer and removed nfs-utils from the RDEPEND
Mar 10 13:22:27 <Anticom> Or do i have to do both PACKAGE_EXCLUDE and remove it from RDEPEND?
Mar 10 13:23:40 <boucman_work> you should do it in a bbappend rather than copying the .bb, but that's not important
Mar 10 13:24:13 <Anticom> However that failed aswell :/
Mar 10 13:24:32 <boucman_work> first thing, you shoud run "bitbake -e packagegroup-core-full-cmdline-sys-services" (piped to a pager or text editor) and check how RDEPEND is constructued and that your change is taken into account
Mar 10 13:29:43 <Anticom> Well I'm not having nfs-utils removed from RDEPENDS_packagegroup-core-full-cmdline-sys-services anymore but just in PACKAGE_EXCLUDE in my local.conf currently
Mar 10 13:30:16 <Anticom> bitbake -e packagegroup_core-full-cmdline shows nfs-utils is currently still in nfs-utils
Mar 10 13:31:00 <Anticom> However the yocto manual states: PACKAGE_EXCLUDE: Use this variable to prevent specific packages from being installed regardless of whether they are "recommended-only" or not. You need to realize that the build process could fail with an error when you prevent the installation of a package whose presence is required by an installed package.
Mar 10 13:31:19 <Anticom> so i thought adding PACKAGE_EXCLUDE += "nfs-utils" to my local.conf would be sufficient
Mar 10 13:31:37 <Anticom> bitbake -u depexp -g my-image shows, that there are no reverse dependencies of nfs-utils
Mar 10 13:31:41 <Anticom> that's why i don't get it
Mar 10 13:36:30 <boucman_work> PACKAGE_EXCLUDE will prevent a package from being installed on an image. That's ok if nobody DEPEND on it, but if someone needs that package, then it will fail
Mar 10 13:36:45 <boucman_work> so someone, somewhere needs nfs-utils, we just need to figure out who
Mar 10 13:37:38 <Anticom> wait what? DEPENDS or RDEPENDS? Why does bitbake file if it's a build dependency and i just prevent it from being installed to rootfs?
Mar 10 13:37:41 <Anticom> how does that make sense?
Mar 10 13:38:00 <boucman_work> you are right, I meant RDEPENDS here
Mar 10 13:38:01 <Anticom> s/file/fail/
Mar 10 13:38:37 <boucman_work> and the answer is in meta/recipes-extended/packagegroups/packagegroup-core-full-cmdline.bb line 154
Mar 10 13:38:53 <Anticom> boucman_work: when i do `bitbake -e my-image` and search for nfs-utils it's only in the PACKAGE_EXCLUDE
Mar 10 13:38:58 <boucman_work> where we see that packagegroup-core-full-cmdline-sys-services RDEPENDS on nfs-utils
Mar 10 13:39:10 <Anticom> oh now i get it
Mar 10 13:39:19 <Anticom> the packagegroup has it as dependency
Mar 10 13:39:21 <Anticom> .__________.
Mar 10 13:39:21 <Anticom> doh
Mar 10 13:39:40 <boucman_work> yes, bitbake -e gives you the environement of a particular recipe, not in general
Mar 10 13:39:44 <Anticom> Okay so how would i do that in a bbappend instead of copying the recipe to my layer?
Mar 10 13:40:10 <boucman_work> I got the name from your error message, but you could also have found it by grep in the buildhistory directory (probably, I havn't checked)
Mar 10 13:40:47 <boucman_work> create a file recipes-extended/packagegroups/packagegroup-core-full-cmdline.bbappend in your meta
Mar 10 13:40:52 <boucman_work> and add a single line in it
Mar 10 13:41:16 <boucman_work> RDEPENDS_packagegroup-core-full-cmdline-sys-services_remove += " nfs-utils"
Mar 10 13:41:42 <boucman_work> (and do a bitbake -e packagegroup-core-full-cmdline to check that it did work as expected)
Mar 10 13:42:03 <boucman_work> you can also safely add that line to your local.conf for faster testing
Mar 10 13:45:30 <Anticom> boucman_work: ah, the _remove suffix was the one i wasn't aware of
Mar 10 13:45:39 <Anticom> i was looking for something like -=
Mar 10 13:46:03 <boucman_work> Anticom, there are subtle differences between the two. I highly recommand reading the syntax section of the bitbake usermanual
Mar 10 13:46:28 <boucman_work> note that the bitbake usermanual is the only yocto manual that is not part of the mega-manual => often missed
Mar 10 13:47:36 <Anticom> boucman_work: Where is the "official" documentation on the bb syntax? i can't find it in the mega manual. I've been using http://elinux.org/Bitbake_Cheat_Sheet#.bb_file_syntax as refference
Mar 10 13:48:16 <boucman_work> https://www.yoctoproject.org/docs/2.0/bitbake-user-manual/bitbake-user-manual.html
Mar 10 13:48:28 <boucman_work> as I said, it's the only one that is not part of the mega-manual
Mar 10 13:48:56 <Anticom> oh, overlooked that, sorry
Mar 10 13:50:41 <Anticom> boucman_work: btw even in the bitbake manual there's no -=
Mar 10 13:50:42 <Anticom> (:
Mar 10 13:51:02 <boucman_work> hmm
Mar 10 13:51:43 <Anticom> And why _remove += and not _remove =
Mar 10 13:51:44 <Anticom> ?
Mar 10 13:52:43 <boucman_work> future-proofing
Mar 10 13:53:10 <Anticom> so _remove += is legit?
Mar 10 13:53:18 <boucman_work> _remove variables are "normal variables" you can add stuff to them multiple times via += and the content will only be used at the end
Mar 10 13:53:24 <boucman_work> so if you type
Mar 10 13:53:28 <boucman_work> X_remove = "a"
Mar 10 13:53:34 <boucman_work> X_remove = "b"
Mar 10 13:53:42 <Anticom> get it
Mar 10 13:53:43 <boucman_work> the variable will only contain b at the end
Mar 10 13:53:48 <boucman_work> :)
Mar 10 13:53:54 <Anticom> thank's for explanation
Mar 10 13:53:57 <boucman_work> np
Mar 10 13:54:21 <Anticom> oh btw: is it considered good practise to have a NL at the EOF in bb's and bbappend's ?
Mar 10 13:54:32 <boucman_work> if you read french, I wrote a rather complete article on how it works here : http://www.linuxembedded.fr/2015/12/yocto-comprendre-bitbake/
Mar 10 13:54:44 <Anticom> nop, i'm german
Mar 10 13:54:50 <boucman_work> good question, no idea :P
Mar 10 13:56:34 <Anticom> At least i'm slowly getting a grip on how to use bitbake
Mar 10 13:57:38 <rburton> new line is sensible but not mandatory
Mar 10 13:57:48 <rburton> (sensible as git moans if you don't have a newline at the end of a file)
Mar 10 13:58:07 <boucman_work> yeah, when you understand the bitbake core principles and use bitbake -e aggressivley, you're on your way to understanding the yocto internals...
Mar 10 13:58:28 <Anticom> boucman_work: i also learned to love depexp
Mar 10 13:58:32 <Anticom> rburton: thank's
Mar 10 13:58:48 <LetoThe2nd> boucman_work: hey, we're in the it industry. understanding things is completely overrated.
Mar 10 13:59:49 <boucman_work> hehe
Mar 10 14:00:17 <Anticom> LetoThe2nd: le me every single day @ work http://starecat.com/content/wp-content/uploads/my-code-doesnt-work-i-have-no-idea-why-my-code-works.jpg
Mar 10 14:00:33 <Anticom> :)
Mar 10 14:03:49 <rburton> there's a print of http://memesvault.com/wp-content/uploads/Dog-Meme-I-Have-No-Idea-What-Im-Doing-03.jpg in the office
Mar 10 14:04:18 <LetoThe2nd> about nothing beats: http://www.bonkersworld.net/images/2011.11.15_life_of_a_swe.png
Mar 10 14:41:03 <Lucretia> Hi, gcc-cross has "--disable-bootstrap" on configure and calls "make all-host." I'm trying to enable Ada support it's failing as it's misconfiguring for some reason. gcc/ada/gcc-interface/Makefile is getting the following line: "tmake_file =  $(srcdir)/config/mips/t-mips $(srcdir)/config/t-slibgcc $(srcdir)/config/t-linux $(srcdir)/config/t-glibc ./config/mips/t-linux64" where the "./config/mips/t-linux64" should be "$(srcdir)/config/mips
Mar 10 14:41:04 <Lucretia> /t-linux64" - is there something I'm missing when gcc is configured and built in this way?
Mar 10 14:44:30 <boucman_work> anybody here is a yocto contributor ? I would need a little guidance on patch-submission...
Mar 10 14:46:28 <btooth> hi.. i know we can use "inherit extrausers" to add users and groups, but how can i set file permissions in do_install to a specific user?
Mar 10 14:57:43 <btooth> ping bluelightning
Mar 10 14:59:51 <LetoThe2nd> oO('ping the blue lightning with a string to tell a thing about patching'... ok. never ever again listen to comedy hiphop music when being overly tired.)
Mar 10 15:11:02 <jonatan> Hello. I have 2 recipes providing similar functionality. I want my image to provide one or the other, so I thought I would PROVIDE and RPROVIDE a "virtual/" package, and set the PREFERRED_PROVIDER in my local.conf, but when I put this virtual package in my IMAGE_INSTALL, it is "not in my base feeds" - is my thinking wrong here? Thanks.
Mar 10 15:25:54 <Anticom> boucman_work: not sure whether this is what you were looking for: https://wiki.yoctoproject.org/wiki/Poky_Contributions#Patch_Submission_Process
Mar 10 15:28:28 <boucman_work> thx I had found a couple of other pages, but not that one...
Mar 10 15:37:45 <btooth> anyone experience with extrausers?
**** ENDING LOGGING AT Thu Mar 10 17:12:40 2016

**** BEGIN LOGGING AT Thu Mar 10 17:14:17 2016
Mar 10 17:22:10 <Lucretia> Hi, I'm getting a failure in a link: /mnt/data/yocto/poky/build/tmp/work/x86_64-linux/gcc-cross-arm/4.9.3-r0/sysroot-destdir/mnt/data/yocto/poky/build/tmp/sysroots/x86_64-linux/usr/include/gcc-build-internal-arm-poky-linux-gnueabi/arm-poky-linux-gnueabi/libstdc++-v3/include/bits/stamp-bits
Mar 10 17:22:18 <Lucretia> stamp-bits points to itself.
Mar 10 17:27:03 <Lucretia> http://pastebin.com/rXyU4PTq <- complete log for gcc-cross-arm
Mar 10 18:53:06 <nerdboy> stupid perl...
Mar 10 19:17:17 <kergoth>  Anyone have any thoughts on https://bugzilla.yoctoproject.org/show_bug.cgi?id=9171 ? Do we care?
Mar 10 19:17:19 <yocti> Bug 9171: normal, Medium, 2.2, ross.burton, NEW , Consider limiting the scope of depchains to runtime deps
Mar 10 19:30:25 <Lucretia> can anyone help re cross-gcc?
Mar 10 19:35:32 <Lucretia> nerdboy: you don't have ada on yocto where you work, do you?
Mar 10 19:43:52 <MDavis> Would this be a good place to ask a quick question about psplash?
Mar 10 19:45:31 <bluelightning> MDavis: yes; I can't guarantee I'll be able to answer but please ask anyway
Mar 10 19:46:38 <MDavis> I am trying to get a full screen pslash to work.  It looks fine except there is a horizontal break in the image right above the progress bar.  It appears white not sure if it is the background bleeding through or what.
Mar 10 19:47:05 <MDavis> Its about 1/2 the progress bar in height.
Mar 10 19:56:36 <nerdboy> Lucretia: not atm
Mar 10 19:56:55 <Lucretia> you any good at debugging the gcc-cross recipe?
Mar 10 19:57:02 <nerdboy> just the old v6/v7 stuff
Mar 10 19:57:28 <nerdboy> haven't had a need to look at it
Mar 10 19:57:34 <Lucretia> thought I'd try to update the meta-ada stuff to jethro/gcc-4.9, but it's turning into a bitch
Mar 10 19:57:55 <Lucretia> the denzil stuff was abit of a hack, tbh
Mar 10 19:57:57 <bluelightning> MDavis: hmm, I haven't observed that to be honest... the psplash code is pretty trivial though so if you need to debug it it shouldn't be hard
Mar 10 19:58:03 <Lucretia> and it's changed a lot since then
Mar 10 19:58:25 <nerdboy> last time i punted and used debian to hand-build cross/native gnat
Mar 10 19:58:38 <nerdboy> *native arm
Mar 10 19:59:00 <Lucretia> all I did was add ",ada" to "--enable-languages"
Mar 10 19:59:05 <nerdboy> made a bin recipe to install it
Mar 10 19:59:08 <Lucretia> and it breaks
Mar 10 19:59:24 <Lucretia> https://github.com/Lucretia/meta-ada/blob/jethro/recipes-devtools/gcc/gcc-cross_4.9.bbappend
Mar 10 19:59:41 <nerdboy> ask dbotton maybe?
Mar 10 19:59:52 <Lucretia> he use yocto?
Mar 10 20:00:12 <Lucretia> yocto has a bit of a weird build.
Mar 10 20:00:15 <nerdboy> not that i know of
Mar 10 20:00:33 <nerdboy> everybody's build is "weird"
Mar 10 20:03:37 <Lucretia> just adding that language to it, shouldn't break it like this, unless i'm missing something
Mar 10 20:06:06 <fledermaus> did LANGUAGES_append have something in it before?
Mar 10 20:06:41 <nerdboy> it's a little weird enabling fortran
Mar 10 20:07:09 <nerdboy> Ada would be similar but with native bootstrap req?
Mar 10 20:08:03 <Lucretia> fledermaus: LANGUAGES is defined in meta/recipes-devtools/gcc/gcc-configure-common.inc
Mar 10 20:08:24 <Lucretia> nerdboy: apparently fortran is weirder
Mar 10 20:08:30 <Lucretia> I dunno
Mar 10 20:08:49 <nerdboy> well, that's a given...
Mar 10 20:09:03 <Lucretia> but using an bappend file should work
Mar 10 20:09:26 <Lucretia> I do wonder whether I should --disable-libada and build the tools+lib separately.
Mar 10 20:09:35 <fledermaus> check the log files to see what broke?
Mar 10 20:10:04 <Lucretia> fledermaus: http://pastebin.com/rXyU4PTq <- nothing obvious
Mar 10 20:13:03 <fledermaus> probably a circular symlink that sysroot_strip is too dumb to notice.
Mar 10 20:14:48 <fledermaus> '/mnt/data/yocto/poky/build/tmp/work/x86_64-linux/gcc-cross-arm/4.9.3-r0/sysroot-destdir/mnt/data/yocto/poky/build/tmp/sysroots/
Mar 10 20:14:50 <fledermaus> '
Mar 10 20:14:58 <fledermaus> looks suspicious
Mar 10 20:15:06 <fledermaus> anyway, time to flee the workateria.
Mar 10 20:20:08 <Lucretia> I'm taking meta-ada out and building to see what gets built under /mnt/data/yocto/poky/build/tmp/work/x86_64-linux/gcc-cross-arm/4.9.3-r0/sysroot-destdir
Mar 10 20:21:54 <MafiaInc> I have an configuration issue with matchbox-window-manager I guess .. I'm on yocto based thin clinet which runs such WM. I want to be able to change the mouse cursor theme. May I get some help from here? :)
Mar 10 20:23:11 <Lucretia> fledermaus: without meta-ada, there is no libstdc++ dir in that dir, only  ../build/tmp/work/x86_64-linux/gcc-cross-arm/4.9.3-r0/sysroot-destdir/mnt/data/yocto/poky/build/tmp/sysroots/x86_64-linux/usr/include/gcc-build-internal-arm-poky-linux-gnueabi/arm-poky-linux-gnueabi/libgcc/
Mar 10 20:23:53 <MDavis> I took a look at the code of psplashy and found my issue.  The text box for the start up message is what I was seeing.  It is displayed even if the message is defined as an empty string.
Mar 10 20:24:14 <MafiaInc> The thing is I have only default arrow cursor and even if I star some app which takes some time to load the cursor doesn't change to show as a progress indicator for example
Mar 10 20:24:33 <MDavis> Don't know if that is the intended behavior or a bug
Mar 10 20:24:50 <rburton> MafiaInc: the progress indicator on starting apps isn't a WM thing, the launcher needs to do that
Mar 10 20:25:03 <rburton> (ie matchbox-desktop does that)
Mar 10 20:26:02 <MafiaInc> ok .. can you point me if this is configurable e.g. a matchbox-desktop related configuration or should be Xorg configuration .. I have no clue where to look at
Mar 10 20:26:51 <rburton> MafiaInc: as i said, it's not config.  the program that you use to start apps needs to do the notifications, it sounds like yours doesn't
Mar 10 20:26:58 <fledermaus> Lucretia: maybe that's it then
Mar 10 20:27:05 <fledermaus> anyway, time to go.
Mar 10 20:27:12 <Lucretia> yeah, but i don't get it
Mar 10 20:27:14 <MafiaInc> hm .. ok thanks I'll try to clarify that
Mar 10 20:29:41 <MafiaInc> rburton, actually it's a TigerVNC client that should inherit cursors form the remote X server guess (or at least this is the behavior if I use LXDE for example), but here in matchbox the icon stays rock solid never change at any action ..
Mar 10 20:30:18 <rburton> maybe your matchbox isn't built with xfixes so it can't change cursor
Mar 10 20:30:36 <MafiaInc> yes .. that could be the reason
Mar 10 20:30:46 <MafiaInc> thanks for the point .. I'll verify! :)
Mar 10 20:52:28 <eraineri> %n
Mar 10 21:03:20 <RP> halstead: any idea why the fedora22 builder would be much slower than the others?
Mar 10 21:05:51 <halstead> RP: some of them are an older generation motherboard with slower CPUs. We usually schedule fewer builds at a time.on those. I can check into that.
Mar 10 21:06:50 <RP> halstead: it could just be that but I've noticed it being noticeably slower than anything else
Mar 10 21:07:07 <Lucretia> seems that gnat requires libstdc++
Mar 10 21:07:16 <RP> halstead: e.g. https://autobuilder.yoctoproject.org/main/builders/nightly-fsl-arm-lsb/builds/686 - everything else has finished,
Mar 10 21:08:38 <halstead> I'm  afk for a little bit. I'll look at it as soon as I'm back.
Mar 10 21:12:59 <agile_prg> my bitbake -k is failing on do_configure saying sdl devel is not found please install it
Mar 10 21:13:04 <agile_prg> any ideas what I might be doing wrong?
Mar 10 21:14:23 <agile_prg> hello anyone here?
Mar 10 21:24:06 <agile_prg> hi dmoseley you here?
Mar 10 21:24:32 <dmoseley> yes
Mar 10 21:48:38 <agile_prg> dmoseley I get an error trying to compile the yocto kernel under qemu task saying install sdl devel do_configure failed
Mar 10 21:48:44 <agile_prg> any ideas what that means?
Mar 10 21:52:06 <dmoseley> agile_prg: Not specifically but it sounds like there is a configure issue with the sdl package which is presumably a dependency of linux-yocto.  The error message should also list a log.do_configure file for that particular package which will have the details of what failed.
Mar 10 21:52:19 <rburton> agile_prg: it means you need to install sdl-devel
Mar 10 21:52:36 <agile_prg> well supposedly it was "installed"
Mar 10 21:52:45 <agile_prg> anyway, just trying to help my friend
Mar 10 21:52:52 <rburton> recent ubuntu?
Mar 10 21:52:54 <agile_prg> I wish I knew how to see what it is actually looking for
Mar 10 21:52:57 <agile_prg> yest 15.10
Mar 10 21:53:03 <rburton> yeah its broken there
Mar 10 21:53:07 <rburton> the release notes have the workaround
Mar 10 21:53:09 <agile_prg> well 15.4
Mar 10 21:53:25 <agile_prg> can you link me to those notes pretty please?
Mar 10 21:53:57 <seebs> so apparently I somehow failed to *notice* a bug report for pseudo not working under qemu.
Mar 10 21:54:06 <seebs> but it's actually pretty fixable, probably.
Mar 10 21:54:23 <seebs> It turns out, I'd not considered the possibility that bind(2) could create a directory entry, so I wasn't checking for, say, successful binds in AF_UNIX.
Mar 10 21:54:38 <rburton> agile_prg: comment #1 on https://bugzilla.yoctoproject.org/show_bug.cgi?id=8553
Mar 10 21:54:39 <yocti> Bug 8553: normal, Medium+, 2.1, ross.burton, VERIFIED FIXED, qemu-native to build on Ubuntu 15.10
Mar 10 21:57:26 <agile_prg> I heart you rburton
Mar 10 21:57:48 <agile_prg> I am confused though, is it fixed or do you still need the workaround?
Mar 10 21:58:32 <rburton> its fixed in git master, but not yet in 2.0.x
Mar 10 21:58:42 <rburton> (the patches for 2.0.1 didn't quite make it)
Mar 10 21:58:50 <rburton> so it will be really fixed in 2.02
Mar 10 21:59:00 <agile_prg> oh ok, I heart you
Mar 10 21:59:08 <rburton> (and is fixed in 2.1, due to be released next month)
Mar 10 22:15:16 <seebs> woot, I believe I've now fixed that bug with pseudo not working under qemu.
Mar 10 22:15:37 <seebs> This has been a really informative exercise in the importance of recognizing that, if I don't know why something happens, it's probably dangerous.
Mar 10 22:15:43 <Lucretia> is libgcc_4.9.bb built after gcc-cross?
Mar 10 22:15:52 <seebs> http://thecodelesscode.com/case/224
Mar 10 22:26:21 <seebs> RP, fray mentioned something about an issue with pseudo not running in qemu environments.
Mar 10 22:26:47 <seebs> I just found a thing which kept it from starting successfully if root was userspace NFS (also running under pseudo), and I'm wondering if that might be it, but I can't find anything under bugzilla.
**** BEGIN LOGGING AT Thu Mar 10 22:35:18 2016
Mar 10 22:57:34 <RP> seebs: it was user mode qemu and we worked around it by unloading it. No open bug
Mar 10 22:59:41 <stwcx> RP: Is there a process for trivial oe-core changes other than sending to the mailing list and waiting?
Mar 10 23:02:18 <RP> stwcx: no, that is the right approach. Something I've missed?
Mar 10 23:02:53 <stwcx> RP: http://patchwork.openembedded.org/patch/117281/ , sent on Monday.  But I was sending it for another author so I didn't know if I messed something up
Mar 10 23:04:56 <RP> stwcx: that change looked fine, I'll make sure its in the next -next build. I think it did make ross/mut already
Mar 10 23:05:47 <stwcx> RP: Thank you.  What are ross/mut?
Mar 10 23:09:08 <RP> stwcx: Its rburton's staging branch, he helps with the merging of patches
Mar 10 23:09:28 <RP> stwcx: I've just merged it to master, will appear shortly
Mar 10 23:48:14 <seebs> Okay, user-mode qemu is a different thing.
Mar 10 23:48:57 <seebs> So, FWIW, if you're running a system under qemu, using a user-mode NFS server (under pseudo) for the rootfs, pseudo should now be able to run in the emulated system. Before, it would have had issues (both because of the lock file not working, and because the sockets wouldn't work.)
Mar 10 23:52:17 <bluelightning> stwcx: for future reference that patch ought to have also been signed-off-by the original author, btw
**** ENDING LOGGING AT Fri Mar 11 02:59:58 2016