**** BEGIN LOGGING AT Thu Feb 13 02:59:57 2020
Feb 13 03:00:03 <kiwi_29> Hello. all the functions of recipe have their log and run files stored in tmp/work/<PLATFORM>/<application>/temp directory   in files such as log.do_compile , run.do_compile etc. Where is the log and run info for __anonymous function in recipe
Feb 13 03:05:05 <denix> none of the python function code is logger
Feb 13 03:08:19 <kiwi_29> I see files in temp directory like log.do_compile log.do_unpack log.do_configure etc.. and they looks like logs of functions written in recipe... no?
Feb 13 03:08:32 <kiwi_29> denix
Feb 13 03:09:19 <kiwi_29> the real question is , how do I get log for __anonymous function written inside the recipe
Feb 13 03:10:55 <mranostay> khem: no just a McLovin reference
Feb 13 03:15:43 <denix> none of the python function code is logged
Feb 13 03:16:05 <denix> do_compile, do_configure, etc. are all shell functions
Feb 13 03:22:16 <khem> there is python devshell
Feb 13 03:39:19 <khem> mranostay: hmm too much context, compiler bails out
Feb 13 04:41:02 <kiwi_29> khem , how to use devpyshell to run the __anonymous function written inside my application recipe
Feb 13 06:55:45 <beratiks> hi, I try to bug fix almost a few weeks. Please help me ! : https://stackoverflow.com/questions/60202030/why-subdevice-is-null-in-ov5640-mipi-v3-module
Feb 13 07:05:34 <khem> beratiks: seems a kernel thing, perhaps a bug in kernel or module
Feb 13 07:06:27 <yocti> New news from stackoverflow: Why Subdevice is NULL in ov5640_mipi_v3 module? <https://stackoverflow.com/questions/60202030/why-subdevice-is-null-in-ov5640-mipi-v3-module>
Feb 13 07:08:22 <beratiks> khem: Interesting. Everything looks fine but not working. I make kernel manually.
Feb 13 08:01:01 <angelo__> hi, need to have KERNEL_DEVICETREE copied in final rootfs /boot, where should i setup this ?
Feb 13 08:32:12 <LetoThe2nd> angelo__: add kernel-device-tree to  IMAGE_INSTALL
Feb 13 08:34:03 <angelo__> ok, i did   IMAGE_INSTALL += " kernel-devicetree "
Feb 13 08:34:12 <angelo__> thanks !
Feb 13 10:02:31 <ernstp> looking at the new cve-check.bbclass stuff. nice idea. but trying to understand what gets included in the image report. seems to include a lot of extra stuff that's not installed
Feb 13 11:16:08 <gaston68> I am developing a Linux image operating system using Yocto Project.I am using SUMO branch (Yocto Project 2.5).In order to install the boost library in my Linux image, I added this line to the local.conf file :IMAGE_INSTALL_append = " boost"But I discover that the default version installed is Boost 1.66.0.Under meta/recipes-support/boost directory I
Feb 13 11:16:08 <gaston68> can find three files :recipes-support/boost/boost_1.66.0.bbrecipes-support/boost/boost-1.66.0.increcipes-support/boost/boost.incI think that those are the recipes for this library.I need to use version 1.70.0 because I have an application coded using this version.I don't know how to modify these recipes in order to have the desired version, or is
Feb 13 11:16:09 <gaston68> there another method to use ? I don't find any helpful tips on the internet. Can you help me please ?Thank you!
Feb 13 11:17:50 <neverpanic> https://layers.openembedded.org/layerindex/recipe/5268/ has 1.7.20
Feb 13 11:18:01 <neverpanic> Updating to Yocto Zeus (3.0) will give you 1.71.0
Feb 13 11:18:13 <neverpanic> If that's not an option, you can always backport that specific version of the recipe.
Feb 13 11:18:52 <neverpanic> Also, usually you would not IMAGE_INSTALL_append for libraries. Instead, write the recipe for the application you want in the image, add that application to IMAGE_INSTALL, and make sure that recipe has DEPENDS += "boost"
Feb 13 11:20:25 <gaston68> what do you mean by "backport" please ?
Feb 13 11:20:47 <gaston68> I do not want to update my all image to another version
Feb 13 11:21:15 <gaston68> and I want to use 1.70.0 version of boost, not 1.71
Feb 13 11:30:09 <gaston68> neverpanic I see what backport mean. If you do me a favor tell me how can I backport the recipe of boost 1.66.0 version in order to run 1.70.0 version
Feb 13 11:45:05 <LetoThe2nd> gaston68: poky never had a recipe for 1.70, so manual effort is always required. you can either try and grab the state for 1.69, or 171 as a start. but this will probably not be exactly trivial, sorry.
Feb 13 11:45:16 <LetoThe2nd> boost is... "complicated" at times.
Feb 13 11:49:47 <nacknick> Hi. I tried to build image for Web OS (Yocto based) following the instructions here: https://www.webosose.org/blog/2019/09/23/webos-ose-1-10-1-release-and-intermediate-branch/
Feb 13 11:50:16 <nacknick> For some reason, I get errors:
Feb 13 11:50:20 <nacknick> `ERROR: Worker process (9133) exited unexpectedly (-9), shutting down...`
Feb 13 11:50:53 <nacknick> `Makefile:42: recipe for target 'webos-image' failed``make: *** [webos-image] Error 1`
Feb 13 11:51:05 <rburton> a pastebin of the full log would be a lot more helpful
Feb 13 11:51:11 <nacknick> OK
Feb 13 11:51:17 <rburton> also paging JaMa ^^^
Feb 13 11:52:20 <nacknick> https://pastebin.com/VGm1Hara
Feb 13 11:56:00 <gaston68> LetoThe2nd I tried to modify the .inc file and put the path of the boost 1.70.0 in SRC_UR and modified the checksum of the other SRC_UR variables. I tried to bitbake but errors related to the patch occured
Feb 13 11:57:26 <nacknick> rburton: Have you seen it?
Feb 13 11:57:38 <nacknick> Maybe someone else knows what is the problem?
Feb 13 11:58:45 <LetoThe2nd> gaston68: recipes do not work that way, sorry
Feb 13 11:59:23 <LetoThe2nd> gaston68: 1) the version comes from the file name
Feb 13 11:59:42 <LetoThe2nd> gaston68: 2) boost usually carries patches that often are modified for version changes.
Feb 13 12:00:05 <LetoThe2nd> gaston68: between 1.66 and current have been changes with repect to py2-py3
Feb 13 12:00:05 <rburton> gaston68: copy the entire boost/ folder
Feb 13 12:00:23 <rburton> that *might* just work
Feb 13 12:00:40 <LetoThe2nd> gaston68: and as you are explicitly asking for 1.70 instead of 1.71 which is a state you can just grab like rburton said, you are in for the pain yourself.
Feb 13 12:05:08 <gaston68> LetoThe2nd rburton see. Thank you for you suggestions
Feb 13 12:06:09 <gaston68> LetoThe2nd by the way, I modified the file name in order to have the required version :)
Feb 13 12:07:25 <yocti> New news from stackoverflow: Couldn't create image from VU+ official repository in Ubuntu 18.04.1 LTS <https://stackoverflow.com/questions/60207224/couldnt-create-image-from-vu-official-repository-in-ubuntu-18-04-1-lts> || Remove locale files corresponding to packages? <https://stackoverflow.com/questions/60207032/remove-locale-files-corresponding-to-packages>
Feb 13 12:15:51 <JaMa> nacknick: why are you building this old version? Do you have enough ram/swap? check dmesg in case the bitbake worker was killed by OOMK
Feb 13 12:17:50 <LetoThe2nd> JaMa: i actually didn't say that because i thought it would bee too obvious to be true....
Feb 13 12:18:34 <JaMa> obvious are the worst kinds of issues :)
Feb 13 12:21:48 <LetoThe2nd> probably
Feb 13 12:41:26 <test123> hello
Feb 13 12:42:29 <test123> I'm trying to build poky, but I'm getting following error: https://pastebin.com/T0iYJ6RC
Feb 13 12:42:53 <mckoan> test123: No space left on devic
Feb 13 12:43:06 <test123> yes, which device has no space??
Feb 13 12:43:14 <test123> accidently two questionmarks
Feb 13 12:43:26 <mckoan> test123: run df -h
Feb 13 12:43:45 <test123> I have 315gb left
Feb 13 12:43:50 <test123> Filesystem      Size  Used Avail Use% Mounted on
Feb 13 12:43:51 <test123> /run/user/1000
Feb 13 12:46:10 <LetoThe2nd> run user?
Feb 13 12:46:25 <LetoThe2nd> thats sounds like something dynamic. which filesystem?
Feb 13 12:48:00 <test123> tmpfs
Feb 13 12:48:38 <test123> Filesystem     Type      Size  Used Avail Use% Mounted on
Feb 13 12:50:08 <LetoThe2nd> you are building in tmpfs and have 416gb free?
Feb 13 12:50:20 <LetoThe2nd> that sounds... too good to be true!
Feb 13 12:51:05 <test123> lol, why do you think im building im tmpfs?
Feb 13 12:52:22 <test123> I thought I'm building it in / ext4 partition
Feb 13 12:52:44 <LetoThe2nd> 12:48 < test123> tmpfs
Feb 13 12:52:47 <LetoThe2nd> thats why
Feb 13 12:52:57 <LetoThe2nd> anyways. have you checked the other hint?
Feb 13 13:01:24 <tmp123> ok, so I still don't understand what is happening
Feb 13 13:05:08 <tmp123> I have rebooted PC and not its building
Feb 13 13:05:17 <tmp123> I don't understand what was the problem?
Feb 13 13:07:18 <tmp123> and now* its building
Feb 13 13:13:25 <JaMa> pebkac?
Feb 13 13:25:00 <xtron> trying to add dependency using the following line, do_compile[depends] += "${@'${INITRAMFS_IMAGE}:do_image_complete' if d.getVar('INITRAMFS_IMAGE') else ''}"
Feb 13 13:26:06 <xtron> but it doesn't work when the variable INITRAMFS_IMAGE is not set,
Feb 13 13:26:33 <xtron> Error: .... do_image_complete' if d.getVar('INITRAMFS_IMAGE') else ':'}' does not contain exactly one ':' character
Feb 13 13:26:50 <xtron> Error: Task 'depends' should be specified in the form 'packagename:task'
Feb 13 13:29:30 <perdmann> hi, is there an available recipe which compiles ubifs tools? (like ubiattach or mkfs.ubifs)?
Feb 13 13:29:57 <ernstp> xtron: d.getVar('INITRAMFS_IMAGE') + ':do_image_complete' if d.getVar('INITRAMFS_IMAGE') else ''
Feb 13 13:30:46 <ernstp> perdmann: yeah it's in mtd-utils
Feb 13 13:32:19 <LetoThe2nd> perdmann: mtd-utils
Feb 13 13:32:34 <LetoThe2nd> perdmann: it provides an additional package mtd-utils-ubifs
Feb 13 13:40:28 <xtron> ernstp, isn't the error about dependency on a NULL task? so why this fix the problem, NULL task is still there in INITRAMFS_IMAGE is not set
Feb 13 13:41:00 <ernstp> if INITRAMFS_IMAGE isn't set up you should end up in the else side with only ''
Feb 13 13:41:45 <xtron> ernstp, yes it will become -> do_compile[depends] += ''
Feb 13 13:42:12 <xtron> e.g: do_compile[depends] += "${@d.getVar('INITRAMFS_IMAGE') + ':do_image_complete' if d.getVar('INITRAMFS_IMAGE') else ''}"
Feb 13 13:42:18 <ernstp> xtron: but that works fine. try to add that just like that. so I think something else is happening
Feb 13 13:42:33 <ernstp> xtron: that should work fine afaik at least  :-)
Feb 13 13:42:55 <xtron> ernstp, hmm
Feb 13 13:43:32 <ernstp> you should stick to pure python in there anyway
Feb 13 13:43:43 <ernstp> ${INITRAMFS_IMAGE} isn't python
Feb 13 13:44:54 <xtron> ernstp, may be, it works that way ${} on some other places, anywayl
Feb 13 13:46:51 <ernstp> xtron: it works in "bb" code and shellscripts, and I guess it works on python code also, but it's probably a bad idea...
Feb 13 13:47:51 <xtron> ernstp, thanks
Feb 13 13:48:36 <perdmann> LetoThe2nd Hm i added mtd-utils but i have no ubifs programs
Feb 13 13:51:56 <LetoThe2nd> perdmann: read again.
Feb 13 13:53:18 <ernstp> perdmann: the output of the mtd-utils recipe is split into multiple packages: oe-pkgdata-util list-pkgs -p mtd-utils
Feb 13 14:09:31 <perdmann> Thanks it works now. I searched for all mtd recipes… Until now i thought that recipes == packages
Feb 13 14:09:51 <perdmann> i searched with bitbake-layers show-recipes
Feb 13 14:15:53 <LetoThe2nd> perdmann: :)
Feb 13 14:24:08 <fullstop> I'm looking for some advice on how to structure something.  I have a process which initializes itself by running a bunch of files in, say, /etc/mybin.d/*.conf
Feb 13 14:24:17 <fullstop> These files are different for each platform
Feb 13 14:24:28 <fullstop> and some may not even exist on some platforms.
Feb 13 14:25:35 <fullstop> Where should these files live?
Feb 13 14:26:23 <LetoThe2nd> fullstop: /usr/share/whatever
Feb 13 14:27:00 <fullstop> LetoThe2nd: Sorry, should they live in the recipe directory, an overlay, as patches?
Feb 13 14:27:18 <LetoThe2nd> fullstop: depends.
Feb 13 14:27:33 <fullstop> I'm trying to do this so that devtool works
Feb 13 14:28:53 <fullstop> and if I put them in files/sama5d2/ devtool wants to delete them or it gets very confusing as to what to do with the different branches it creates.
Feb 13 14:29:18 <fullstop> and I can see that it's a can of worms once other people here start making changes.
Feb 13 14:32:54 <opennandra> hello, is dropbear started by default during systemd service? I saw dropbear@.service but I cannot see any symlink in /etc/systemd/system/...
Feb 13 14:33:16 <LetoThe2nd> opennandra: dropbear usually is .socket activated.
Feb 13 14:33:45 <opennandra> LetoThe2nd: yes I saw socket also so it means if I want to ssh to devie then it's started?
Feb 13 14:33:52 <opennandra> and also key is generated?
Feb 13 14:34:02 <LetoThe2nd> opennandra: it should all work out of the box
Feb 13 14:34:19 <opennandra> LetoThe2nd: I'm using readonly rootfs and I'm trying to move generation of rsa key to /data partition
Feb 13 14:34:54 <opennandra> but after first boot /data is not populated with rsa generated key so I was curious if jey generation is done during first ssh attempt
Feb 13 14:34:59 <opennandra> *key
Feb 13 14:35:45 <LetoThe2nd> opennandra: we're RO too and it works
Feb 13 14:36:06 <LetoThe2nd> no, the keys should be pre-generated during image creation IIRC
Feb 13 14:36:21 <opennandra> LetoThe2nd:  I don't want to have same key for all devices
Feb 13 14:37:22 <opennandra> LetoThe2nd: so I've override dropbearkey.service and also dropbear@.service to point to my /data directory
Feb 13 14:37:33 <opennandra> just was curious why it's not started automatically
Feb 13 14:42:33 <gaston54> Hello, is there anybody here who knows how to install Boost 1.70.0 library version in Yocto Linux image ?
Feb 13 14:49:24 <erbo> gaston54: there are boost recipes in openembedded-core
Feb 13 14:49:42 <erbo> which version they are will depend on which release you are on
Feb 13 14:50:58 <erbo> Current version is 1.72, not sure if there's some release that uses exactly 1.70.0
Feb 13 14:51:08 <gaston54> erbo Yes in fact that is my problem. First I can not find boost 1.70.0 version at all . Second I need it for my SUMO yocto version
Feb 13 14:52:18 <LetoThe2nd> gaston54: it has all been said already. there was never a 1.70 recipe in the layers, we went from 1.69 to 1.71 directly
Feb 13 14:52:36 <LetoThe2nd> gaston54: and there is nothing official for sumo, certainly
Feb 13 14:52:49 <erbo> gaston54: Then I'd suggest trying to create you own 1.70.0 version based on either 1.69 or 1.71
Feb 13 14:53:29 <erbo> But perhaps it's ok with a newer version than 1.70?
Feb 13 14:55:19 <gaston54> LetoThe2nd That's it buddy , I have just gave another try and I think that there is no way to use 1.70 version for SUMO branch. Thank you
Feb 13 14:56:24 <gaston54> erbo Should I update all the kernel of SUMO ? Or can I use the 1.72 without upgrading the kernel ?
Feb 13 14:57:18 <erbo> gaston54: I would just use the newer boost recipes by copying them into some local layer
Feb 13 14:57:39 <LetoThe2nd> gaston54: boost and the kernel have nothing in common.
Feb 13 14:57:47 <erbo> need to leave now though for kindergarden pickups, good luck :)
Feb 13 14:58:25 <gaston54> erbo Okey thank you
Feb 13 14:58:54 <gaston54> LetoThe2nd I mean the yocto Branch
Feb 13 14:59:59 <LetoThe2nd> gaston54: like i said, its not exactly one of the least problemateic recipes, but also not one of the highest. go, try and find out.
Feb 13 15:02:46 <gaston54> LetoThe2nd I am new in Yocto world that's why I still little bit confused. I will give a try thank you
Feb 13 15:04:05 <LetoThe2nd> gaston54: i take that you have already watched the relevant episodes of https://www.youtube.com/playlist?list=PLD4M5FoHz-TxMfBFrDKfIS_GLY25Qsfyj
Feb 13 15:07:33 <gaston54> LetoThe2nd I have not watched it yet. Those episodes will help me more understand Yocto. Thank you
Feb 13 15:08:34 <LetoThe2nd> gaston54: enjoy!
Feb 13 15:38:12 <yocti> New news from stackoverflow: Yocto Project: Ubuntu 16.04 LTS keeps crashing when running: bitbake fsl-image-gui <https://stackoverflow.com/questions/60211165/yocto-project-ubuntu-16-04-lts-keeps-crashing-when-running-bitbake-fsl-image-g>
Feb 13 16:08:17 <yocti> New news from stackoverflow: Where the Jenkinsfile should be stored when working with different repos managed by the repo tool? <https://stackoverflow.com/questions/60211683/where-the-jenkinsfile-should-be-stored-when-working-with-different-repos-managed>
Feb 13 17:25:08 <khem> RP: the busybox patch seems you got prior one ? I have pushed the one to kraj/pu http://git.yoctoproject.org/clean/cgit.cgi/poky-contrib/commit/?h=kraj/pu&id=96bbf1a18c5743e26aacdddf6bcbcb2351a83e12
Feb 13 17:25:19 <khem> it has been building fine ..
Feb 13 17:53:05 <RP> khem: I took what was on the mailing list
Feb 13 18:00:18 <khem> RP: OK, usually git send-email does not molest patches but who knows
Feb 13 18:13:01 <khem> https://fosdem.org/2020/schedule/event/nextgencontributors/ for zeddi and denix who were heckling me yesterday :)
Feb 13 18:20:02 <Marex> hey, is there something in OE that would do some sort of jobserver ?
Feb 13 18:20:38 <Marex> I mean, if I have e.g. 64 core machine, and I run bitbake , it either overutilizes the machine at the beginning and cannot utilize it at the end of the build or it's underutilized all the time
Feb 13 18:21:02 <Marex> e.g. when building stuff with qtwebengine, the paralelization is good at the beginning of the build, but at the end it's only the webengine that's being built
Feb 13 18:23:23 <khem> heh, I suffer same problem.
Feb 13 18:23:48 <khem> so I used to use a high -j value and then combine it with -l 1.5
Feb 13 18:24:01 <khem> so my system is always adequately loaded
Feb 13 18:24:23 <khem> I stopped doing it, after replacing my ssd, and motherboard fan
Feb 13 18:25:50 <Marex> khem: well, that's an option all right
Feb 13 18:49:25 <Ad0> what is the best practice to populate / install files in a volatile-binded directory? when using NetworkManager in a readonly rootfs, I have bound it to the /data partition. should I do install on that destination?
Feb 13 18:50:02 <khem> Ad0: they should get copies there on first boot
Feb 13 18:51:19 <Ad0> you mean run a script depending on that bind being done?
Feb 13 19:19:58 <fullstop> when using SYSTEMD_SERVICE_${PN} should I see something in the postinst script?
Feb 13 19:25:12 <fullstop> Other recipes do, so mine is a bit different.  I'm adding a service for something in busybox which is done in a busybox_%.bbappend -- the files show up but postinst doesn't do the systemd magic.
Feb 13 19:25:44 <fullstop> Is it valid to "inherit systemd" in a bbappend?
Feb 13 19:26:10 <Ad0> I don't see why not since it just adds to the original bb and inherit is more like include
Feb 13 19:26:16 <fullstop> hmm..
Feb 13 19:26:42 <Ad0> I made my own recipe instead of tacking on to something else
Feb 13 19:27:31 <fullstop> The service is for runit, part of busybox.
Feb 13 19:27:36 <Ad0> ok
Feb 13 19:28:06 <fullstop> I could separate them without much of an issue, I suppose.
Feb 13 19:28:26 <Ad0> I saw something with systemd (I am quite fresh in yocto myself)
Feb 13 19:28:31 <Ad0> there's an own enable statement
Feb 13 19:28:50 <fullstop> SYSTEMD_AUTO_ENABLE, defaults to "enable"
Feb 13 19:29:00 <Ad0> right
Feb 13 19:29:39 <Ad0> SYSTEMD_PACKAGES = "${PN}"
Feb 13 19:29:40 <fullstop> I'm cleaning busybox and building again.. I changed SYSTEMD_AUTO_ENABLE to SYSTEMD_AUTO_ENABLE_${PN} since my service is runit.service and not busybox.service
Feb 13 19:29:56 <Ad0> DEPENDS_append_class-target = " systemd"
Feb 13 19:30:01 <Ad0> RDEPENDS_${PN}_class-target = "systemd"
Feb 13 19:30:09 <Ad0> not sure what that was from
Feb 13 19:30:14 <Ad0> I just add shit I saw into there
Feb 13 19:30:47 <fullstop> Well, that did not work.  Maybe I'll just add a separate recipe instead of fighting it.
Feb 13 19:31:15 <Ad0> I mean it should be possible to do it in a bbappend
Feb 13 19:31:48 <Ad0> the systemd stuff I've seen are in separate .bb though not bbappend
Feb 13 19:32:16 <Ad0> https://github.com/intel/meta-intel-iot-security/blob/master/meta-security-smack/recipes-core/systemd/systemd_%25.bbappend
Feb 13 19:35:00 <Ad0> https://github.com/ostroproject/meta-ostro/blob/master/meta-ostro/recipes-swupd/swupd-client/swupd-client_%25.bbappend
Feb 13 19:35:34 <fullstop> yes, those look very similar to what I was doing
Feb 13 19:35:38 <Ad0> fullstop, are there other bbappends?
Feb 13 19:35:48 <Ad0> that does systemd stuff. note SYSTEMD_SERVICE_${PN} += and not =
Feb 13 19:36:09 <fullstop> Ad0: no other appends that I know of
Feb 13 19:36:14 <fullstop> and I tried +=
Feb 13 19:36:15 <Ad0> ok
Feb 13 19:36:30 <Ad0> you can do bitbake -e recipe and grep it to look at the SYSTEMD_SERVICE_${PN} variables
Feb 13 19:36:41 <Ad0> see what's going on
Feb 13 19:37:25 <Ad0> how do you install the actual file?
Feb 13 19:37:56 <Ad0> do_install_append () and dest  ${D}/${systemd_system_unitdir} right?
Feb 13 19:38:21 <fullstop> https://pastebin.com/raw/9Xq6xYwV
Feb 13 19:40:12 <Ad0> hm no FILES_${PN} += " ?
Feb 13 19:40:56 <fullstop> one sec, it was there
Feb 13 19:41:15 <Ad0> I have ${systemd_system_unitdir}/* in that var
Feb 13 19:41:18 <fullstop> FILES_${PN} += " ${bindir}/runsvdir-start ${systemd_system_unitdir}/runit.service"
Feb 13 19:41:35 <Ad0> it all seems by the book
Feb 13 19:41:42 <fullstop> I put the runit stuff in its own recipe and postinst has the systemd magic in it.
Feb 13 19:41:54 <fullstop> Perhaps this is better; I can update this stuff without reinstalling busybox.
Feb 13 19:42:19 <Ad0> it makes debugging easier too because of that
Feb 13 19:42:28 <Ad0> I hate kernel bbappends with a passion heh
Feb 13 19:42:43 <Ad0> I change one small .dts file and it all has to recompile
Feb 13 19:42:56 <fullstop> I separated my dts for that reason.
Feb 13 19:43:13 <fullstop> It makes it so much easier to iterate through dts changes on a new board
Feb 13 19:43:26 <Ad0> yeah
Feb 13 19:43:35 <Ad0> another technique I have is to have a devshell open
Feb 13 19:43:42 <Ad0> for virtual/kernel
Feb 13 19:52:47 <xyzzy42> What is the right way to change an existing package to use my git server instead of package's official site?  I tried redefining SRC_URI in a bbappend, but some of the packages have additional files added in SRC_URI beyond the git server and I lose those.
Feb 13 19:53:49 <xyzzy42> Obviously I could edit the files in the meta layer the package is from, but that seems wrong.  It should be possible to do this in a meta layer I make.
Feb 13 19:55:22 <xyzzy42> I could do some kind of python regex replace to change the just the git server in the source(s) in the SRC_URI.  That seems ugly, and fragile, since it breaks if source meta layer changes their git server URI.
Feb 13 19:57:07 <fullstop> xyzzy42: I clone the repo and rename "origin" to "upstream", and push my new "origin" to the git repo here.
Feb 13 19:57:18 <xyzzy42> The meta-layer this is in could have been written something like GIT_SERVER?="git://whatever, SRC_URI = "${GIT_SERVER}", so I could change the just part of what it adds to SRC_URI
Feb 13 19:58:04 <fullstop> This lets me merge upstream changes while staying local for builds
Feb 13 19:58:32 <xyzzy42> fullstop: How would I get the recipe for the package to reference my local git repo?
Feb 13 20:04:29 <fullstop> I see your problem now
Feb 13 20:05:04 <fullstop> Can you explain "but some of the packages have additional files added in SRC_URI beyond the git server and I lose those." a little more?
Feb 13 21:06:42 <champagneg> Hi! We creating a pkg_postinst_ontarget script is there a way to include only what's specified for _ontarget? The resulting script seems to contain both postinst and postinst_ontarget snippets
Feb 13 21:20:26 <xyzzy42> fullstop: It's a bootloader, barebox.  There are multiple packages, 1st stage, 2nd stage, utils.  Some of them add more files, stored in the meta layer, to the SRC_URI.  config file, patches, stuff like that.
Feb 13 21:21:57 <xyzzy42> It seems like, "I want to use my git repo / branch," would be common enough that one could do it in a meta layer.
Feb 13 21:32:47 <xyzzy42> If I look at how the kernel packages, where custom git repos are more common, deal with this, it looks like there's two main things.  One is design the recipes largely around generic inc files, so one can easily make a new pacakge bb file without duplicating much.  The other is to use SRC_URI_append to add source local recipe source files.
Feb 13 21:48:35 <kergoth> xyzzy42: i'd bbappend the recipe and use PREMIRRORS to redirect it to the other server.
Feb 13 21:48:53 <kergoth> if it's local on the same machine, use externalsrc instead, possibly via devtool modify
Feb 13 21:56:50 <xyzzy42> kergoth: Could I set PREMIRRORS_barebox_append to just set the mirror for this one package?
Feb 13 21:57:24 <kergoth> if you mean the recipe, since package doesn't make sense in this context, if you use a bbappend it by definition only applies to the recipe
Feb 13 21:57:37 <kergoth> you'd only need to use a conditional append if you're doing it from local.conf or something instead
Feb 13 21:57:52 <kergoth> if you want to use local.conf, then you have to use override syntax, not _barebox
Feb 13 21:57:58 <kergoth> PREMIRRORS_append_pn-barebox
Feb 13 22:00:16 <xyzzy42> kergoth: I see, yes I should have said recipe.  I'll give that a try, it might work better that what I've done, which is now working after I changed the barebox meta to use SRC_URI_append when it adds additional local source code.
Feb 13 22:03:05 <xyzzy42> Though it's not really a mirror, it's a fork.  But it doesn't seem like that matters too much behavior wise.
Feb 13 22:04:22 <xyzzy42> Hmm, still need to regex match the original SRC_URI.  I.e., they change from git to https, my regex doesn't match anymore.
Feb 13 22:22:24 <kergoth> xyzzy42: if it's http://, then it's not a git repo anymore, it's a tarball, and your premirrors won't be of much use anyway. but it's a group of regexes, you can make it match whatever you want
Feb 13 23:38:07 <xyzzy42> I've come to the conclusion I should create a new recipe for a new version of something with an existing recipe in another layer.  I'm able to use include files to avoid cut&paste from the existing recipe.  But there is a patch file in that layer I'd like to use, and it can't be found, I assume because FILESPATHS and FILESEXTRAPATHS only reference the new recipe in my layer.
Feb 13 23:39:55 <xyzzy42> I'm not sure how to add the original layer's recipe's path to FILESEXTRAPATHS without hard-coding the absolute path or the exact relative path between the two layers.
Feb 13 23:44:39 <RP> khem: the patches I tested were the v2, I just replied to the wrong thread fwiw
Feb 14 01:17:07 <mischief> what is the right way to add a dependency on lvm2-udevrules to cryptsetupp from meta-openembedded? the cryptsetup tools seem to need the lvm2 udev rules for devicemapper, otherwise they hang waiting for udev notifications.
Feb 14 01:53:14 <khem> RP: I am not sure whats going on, since I use same computer to send patches
**** ENDING LOGGING AT Fri Feb 14 02:59:59 2020