**** BEGIN LOGGING AT Fri Jan 24 02:59:58 2020 Jan 24 05:55:14 guys, does the AB publishes any of the images? Jan 24 07:07:19 New news from stackoverflow: How can i remove linker problem faced while cross-compiling Jan 24 08:04:48 good morning Jan 24 08:11:45 morning! Jan 24 08:19:57 erbo: good lck with that SO guy. Jan 24 08:26:17 LetoThe2nd: yeah I was a bit unsure if I wanted to touch that issue :) Jan 24 08:27:22 erbo: i've tried to help him at least 5 times now. i am through. Jan 24 08:28:26 erbo: a.k.a. "i cannot even compile anything more complicated than a 'hello world' manually, i wanna do embedded" Jan 24 08:28:58 couple of days back he actually asked "what is a linker flag" Jan 24 08:29:45 and by now obviously understood "a linker flag is something you add to the compiler call to make things magically happen. hopefully" Jan 24 08:30:06 * LetoThe2nd gets really angry. sorry. need to vent for all the time i've wasted. Jan 24 08:32:44 Oh, looking at the SO history now I can see clean patterns of a help vampire :) Jan 24 08:33:40 nah, i think he's just a clueless idiot. Jan 24 08:34:13 he even removes all the old threads that clearly led to the conclusion that he just has no idea and should do his homework. Jan 24 08:42:32 LetoThe2nd: everyone starts from somewhere and take the time they need to acquire the same level of knowledge. I hate people asking questions before even documenting themselves a little beforehand as well but there are such people, just ignore them if you don't want to talk with them and let other people make their own opinions Jan 24 08:42:46 qschulz: yeah. i know. Jan 24 08:42:56 sorry. i just really needed to vent. Jan 24 08:44:54 LetoThe2nd: yeah I know. Try to not insult people that's it. Just critic the way they're doing things without making it personal or offensive to the person. Jan 24 08:45:16 LetoThe2nd: that's actually something surprisingly hard. For example in code reviews. Jan 24 08:45:42 LetoThe2nd: TGIF for you I guess :) Jan 24 08:45:43 qschulz: i just want my wasted time back. Jan 24 08:46:21 LetoThe2nd: ask for beers from the adepts of your sect to forget about it :) Jan 24 08:46:28 hrhr Jan 24 09:07:45 New news from stackoverflow: Yocto build error: Building image with debian package management Jan 24 09:33:41 I added ptest to distro features as per wiki, now how can I selectively add -ptest packages in image recipe? With IMAGE_INSTALL i just get nothing provides package-ptest error Jan 24 09:57:54 hi, can anyone recall why sqlite3 is using the bundled source format in yocto? Is it to avoid tcl, tk etc build time dependencies? applying CVE etch patches is much tricker though since other distros don't use the bundled format and upstream doesn't provide patches for bundled format either.. Jan 24 10:04:15 (got disconnected, not 100% sure what was my last msg :p) Jan 24 10:05:11 Anyways, the issue of bb.plain not showing in stdout (https://pastebin.com/hJKYpMBg) doesn't seem to be related to `bitbake-layers` as it happens when the layer is added fully manually Jan 24 10:06:55 Also the example seems to work fine when copying the `bitbake` directory from poky and creating a bare minimal conf/bb structure Jan 24 10:08:21 That means no `meta` no `meta-poky`... Jan 24 10:09:24 So I'm not currently clear if this (putting `bb.plain()` in a recipe and not seeing it during the build in stdout) is a bug or a feature Jan 24 10:11:41 Perhaps that's the core question to solve first. Should `bb.plain()` in a recipe task show during build in a vanilla setting poky? Jan 24 10:11:45 MeanEngi: as the recipe gets created by bitbake-layers that way, and we intentionally ship bitbake-layers plus advocate it as the preferred way to create a new layer, it should be considered a bug. so, please file your findings. the worst that can happen is a "won't fix because..." Jan 24 10:12:23 LetoThe2nd: Fair :) Will create the issue Jan 24 10:29:16 hi Jan 24 10:29:39 is it possible to change the interpreter for a _append function? Jan 24 10:31:05 crazy_imp: hum, like what? Jan 24 10:32:55 problem is, my do_rootfs_append fails with syntax error - i highly suspect the cause is: ./meta/classes/image.bbclass:197:fakeroot python do_rootfs () Jan 24 10:33:11 so the do_rootfs() is initially a python function Jan 24 10:33:20 and my (old) do_rootfs_append is bash Jan 24 10:33:34 and if bash code gets append to the python code.... Jan 24 10:34:38 crazy_imp: hmh Jan 24 10:39:02 crazy_imp: rather than _append you probably want to use ROOTFS_POSTPROCESS_COMMAND or IMAGE_POSTPROCESS_COMMAND depending on when you want it to run Jan 24 11:07:43 RP: sent a revised dosfstools patch, can you drop the one in next please Jan 24 11:26:02 rburton: the packagegroup one? Jan 24 11:31:34 RP: ues Jan 24 11:34:03 rburton: gone Jan 24 12:24:52 how do I include an additional kernel module? The sources for the module exist but .ko is not on image Jan 24 12:25:02 (total kernel noob) Jan 24 12:25:28 stuom1: is the module being built (e.g., kernel config enables it) Jan 24 12:26:02 probably not, I guess that is what I would like to do Jan 24 12:27:30 is there some yocto-magic for it or should I just go and learn how that is done in general in linux Jan 24 12:30:03 depends a bit on your usecase Jan 24 12:30:19 i personally prefer to nail my defconfig down, but other options exist. Jan 24 12:33:17 I found this https://wiki.yoctoproject.org/wiki/TipsAndTricks/QuickAndDirtyKernelConfig Jan 24 12:33:25 maybe i try the quick and dirty way first ;) Jan 24 12:34:18 Is this at all the right place to start? Jan 24 12:36:01 LetoThe2nd: Made the issue: https://bugzilla.yoctoproject.org/show_bug.cgi?id=13749 Is it common practice to make a "test case" also? Jan 24 12:36:02 Bug 13749: normal, Undecided, ---, richard.purdie, NEW , Example recipe created with bitbake-layers create-layer doesn't print to stdout Jan 24 12:39:01 MeanEngi: the easier it is to reproduce for somebody looking, the better. Thank you, looks absolutely fine for me. Jan 24 12:43:31 LetoThe2nd: Thanks :) Jan 24 13:24:48 "Value requested for CONFIG_AUFS_FS not in final .config" what causes that? Jan 24 13:27:27 Ad0: I think we need a tad bit more info :) Which recipe? Did you do anything special with it? (bbappend, overriding files, etc.)? Jan 24 13:29:01 yes. I have my own entire defconfig . recipes-kernel[dir] -> linux[dir] -> linux-raspberrypi_%.bbappend, linux-raspberrypi[dir] -> defconfig Jan 24 13:29:32 so I know my defconfig makes it's way into there Jan 24 13:30:17 kernel is 4.19.71 so it should support AUFS5. Jan 24 13:31:25 Ad0: I don't know exactly honestly. How did you create your defconfig? Jan 24 13:31:40 I dumped it from a device and enabled my own stuff Jan 24 13:31:55 using the /proc method Jan 24 13:33:23 Using the proc method? Jan 24 13:33:42 I do not understand what you did :) Jan 24 13:33:45 cat /proc/config.gz | gunzip > running.config Jan 24 13:33:52 ok, then? Jan 24 13:34:07 I just wanted to mention that using EXCLUDE_FROM_SHLIBS was exactly what I needed in my case to maintain my own ABI information for a few select libraries. Jan 24 13:34:09 then renaming that to defconfig and enabling AUFS Jan 24 13:35:04 is there some intelligent config merging / mish mashing? Jan 24 13:35:36 should I perhaps use config fragments or do something else? I want to tweak the entire config myself and totally replace it, should be possible Jan 24 13:36:35 how did you do "enabling AUFS"? Jan 24 13:37:51 CONFIG_AUFS_FS=y Jan 24 13:37:57 yeah no Jan 24 13:38:05 CONFIG_AUFS_BRANCH_MAX_127=y Jan 24 13:39:02 it's not really a yocto problem I think - https://github.com/torvalds/linux/blob/master/scripts/kconfig/merge_config.sh Jan 24 13:40:04 I am running Toaster inside Build Appliance. after "bitbake myrecipe" it parses recipes and stops at 85-95% and remain unfinished. What may be wrong? Jan 24 13:42:13 Is there are any new documentation how to run Toaster? Hob is outdated. Jan 24 13:42:28 Is there test toaster version in the web? Jan 24 13:45:17 why Toster WebInterface may show "ToasterSetting object" instead or variable name in Django admin page/ Toaster settings? Jan 24 13:46:27 dv|2: poky ships the corresponding, current toaster version Jan 24 13:47:29 LetoThe2nd, is it usable inside Build Appliance image? Jan 24 13:50:17 Ad0: you get the sources (or use devshell or devtool or whatever) if your kernel, then put your running.config at the root and rename it .config. Then you run menuconfig, from there you select the options you want, you save when you exit. Then you run savedefconfig. The defconfig is at the root. You take this one. That's how one creates a defconfig Jan 24 13:51:42 Ad0: you can use fragments yes, that's encouraged for kernel-yocto recipes (linux-raspebrry is one). It's honestly a little bit more complex but not too much. You take the .config after your kernel recipe has built. With that one, you run savedefconfig. You backup the defconfig. Then you run menuconfig, from there you select the options you want, you save when you exit. Then you run savedefconfig. Jan 24 13:52:07 from there, you do a diff of the backed-up defconfig and the new one. You put it in a .scc file and add it to your SRC_URI Jan 24 13:52:16 thanks Jan 24 13:52:28 maybe that's the way to go Jan 24 13:52:56 You're the second or third people to ask the same question in a few days. We're missing some docs or tools for sure then. Jan 24 13:53:06 but I still want to know the reason why a mismatch is created, a mismatch between what? Jan 24 13:53:39 https://www.openembedded.org/wiki/Kernel_Building so doing this one ? Jan 24 13:53:55 bitbake -c menuconfig virtual/kernel Jan 24 13:54:43 the biggest problem for me is where the files are modified and how I am sure they are saved etc Jan 24 13:54:48 OK, that's one wiki we could expand. There's much more to say and I'm surprised there is a tool to concatenate fragments but not create them Jan 24 13:55:24 Ad0: anything done by bitbake is temporary let's say. If you don't modify something in the Yocto sources or recipes, it's wrong basically. Jan 24 13:55:32 when I do bitbake -c menuconfig virtual/kernel, what defconfig does it read from as a base? Jan 24 13:55:35 menuconfig is nice for debugging purposes but it won't save anything Jan 24 13:55:45 ok Jan 24 13:55:51 it takes whatever the recipe said it should take Jan 24 13:56:11 RP: I'm thinking patch v3 from last night should solve the logrotate issue. It's a relatively simple fix, but if the test is only meant to test that logrotate works then I think it's a sufficient fix Jan 24 13:56:13 it's per recipe, so you would need a bit of digging in the recipe to know exactly what;s going on Jan 24 13:56:23 most likely some default defconfig plus fragments for your case Jan 24 13:56:28 yeah Jan 24 13:56:48 maybe raspberrypi3 doesn't like me replacing the entire defconfig and creates a mismatch because it doesn't expect it or something Jan 24 13:57:38 * qschulz shrugs Jan 24 14:00:47 wait wtf... there's a diffconfig tasks /o\ Jan 24 14:01:06 maybe you could try menuconfig, select your things and right after a diffconfig Jan 24 14:02:17 that should be way better. Hopefully that works. I'm so pissed I didn't know :/ Jan 24 14:03:53 Ad0: ^ Jan 24 14:04:09 thanks :) Jan 24 14:04:22 yeah well I want to base it off an existing device's config tho Jan 24 14:05:10 https://github.com/agherzan/meta-raspberrypi/blob/warrior/recipes-kernel/linux/linux-raspberrypi.inc Jan 24 14:07:44 a mismatch is because the option probably depends on something that is not enabled, so yeah I would use menuconfig I think. Jan 24 14:08:04 or trawl through the defconfig to see what I have to enable by hand Jan 24 14:08:22 Yes, try with menuconfig and then diffconfig. This should output a config fragments Jan 24 14:08:34 No. You do not edit a defconfig by hand Jan 24 14:08:36 same for .config Jan 24 14:11:15 ok Jan 24 14:13:51 thing is that a lot of documentation define the variable name, but it's hard to see in menuconfig what option which only has a description corresponds to the variable name Jan 24 14:14:25 Use the "/" Luke :) Jan 24 14:14:35 hehe Jan 24 14:14:44 there is a search in menuconfig, use the name of the config without CONFIG_ in front Jan 24 14:15:17 thanks a lot Jan 24 14:15:22 that would solve a bunch of issues Jan 24 14:16:37 let us know what's the outcome :) Jan 24 14:16:52 (i.e. if it works or not. My English seems broken again :) ) Jan 24 14:25:20 AUFS is nowhere to be found Jan 24 14:25:29 any pi3 and/or alsa people arond, workign on FOSDEM demo :) Jan 24 14:28:42 Ad0: how do you know this is supported for this kernel? Jan 24 14:29:25 I have to check further Jan 24 14:29:42 Is there any working Toaster demo? Jan 24 14:32:42 Ad0: I mean where did you get this specific Kconfig options? Jan 24 14:33:19 https://raw.githubusercontent.com/moby/moby/master/contrib/check-config.sh Jan 24 14:33:37 maybe AUFS support was ditched Jan 24 14:34:08 weird that it works fine in one ARM vexpress kernel but not in raspberrypi's . Jan 24 14:34:31 I think AUFS is dead and is replaced by overlayfs2 Jan 24 14:34:32 Hello, I stuck, I created closed package (let's call it A) with commercial license using bin_package. Now I want to use it as dependency of final package (let's call it B) adding DEPENDS += "A". Recipe is parsed correctly, package A is found but not installed in B recipe-sysroot? Jan 24 14:34:37 Any idea why? Jan 24 14:35:16 Crofton|road: I expect the table is now full with Beth's disco-in-a-box and Alex's demo hardware Jan 24 14:35:48 tolszak: How do you know if package A is found? Jan 24 14:35:49 Well, any hint on sound, glad I got this going again, should finish it :) Jan 24 14:36:17 tolszak: also, not all sources are installed in a sysroot so it might depend where your file is installed in packageB to be able to see it in sysroots Jan 24 14:36:27 qschulz: If I change it to DEPENDS += "Afoo" bitbake fails because it can Jan 24 14:36:36 it can't find Afoo package Jan 24 14:36:38 Crofton|road: I like my boards silent haha, not done any work with sound on the Pi 3 or other hw for several years Jan 24 14:36:54 And with DEPENDS += "A" fails on compilation Jan 24 14:36:55 gqrx tries to bring up some alsa stuff and dies Jan 24 14:37:36 jsut need to get past the error, and possible something needs updating in bsp Jan 24 14:38:07 tolszak: what's the error at compile time? Have you checked in the sysroot that the file needed (header or lib) is present *somewhere*? Jan 24 14:39:07 https://www.irccloud.com/pastebin/tMRXN2aK/ Jan 24 14:40:18 Ad0: from what I could gather, AUFS is a set of patches. It's not in mainline Jan 24 14:40:31 ok thanks. Jan 24 14:40:31 qschulz: That's the point, I can see those files in "A" build dir: tmw/work/.../A/... Jan 24 14:40:40 I don't need it I found out but now I am more educated on menu config Jan 24 14:40:46 Ad0: http://aufs.sourceforge.net/ Jan 24 14:40:46 and you found about diff :) Jan 24 14:40:54 qschulz: but it is not present in tmp/work/.../B/.../recipe-sysroot Jan 24 14:40:57 Ad0: yup, one always learn :) Jan 24 14:41:38 tolszak: what's the path to the missing file in packageA? Jan 24 14:42:45 qschulz: /usr/locale/include Jan 24 14:43:04 qschulz: /usr/local/include/A/header.h Jan 24 14:43:14 qschulz: Does it make a difference? Jan 24 14:44:10 qschulz: and libs in /usr/local/lib/A/libname.so Jan 24 14:44:29 yeah, I don't think /usr/local is part of the sources put into a sysroot Jan 24 14:44:54 qschulz: Is it documented somewhere? Jan 24 14:46:06 I wonder if I can enforce and mark A to be considered as proper build dependency Jan 24 14:46:28 tolszak: I don't know, try quickly to put the library from packageA in /usr/lib and see if that helps Jan 24 14:47:33 qschulz: FYI I started to learn yocto a week ago and I'm close to finalizing my own distro, so far I love it and I was a bit scared reading in docs about steep learning curve Jan 24 14:47:43 Will try that and let know Jan 24 14:50:24 qschulz: Yeah it works Jan 24 14:50:37 the /usr/local/ was the issue Jan 24 14:51:07 the issue is that I never remember where this thing is done in code so I can't check quickly Jan 24 14:51:25 which sources are taken in a DEPENDS => /me shrugs Jan 24 14:53:17 https://www.yoctoproject.org/docs/current/mega-manual/mega-manual.html#ref-tasks-populate_sysroot Jan 24 14:55:12 may I set Preffered_version of the package in img.bb file? Jan 24 14:56:41 qschulz: !!!! SYSROOT_DIRS need to be set in A recipe Jan 24 14:56:54 qschulz: Many, many thanks! Jan 24 14:57:04 tolszak: yes. The question now is why do you think libs in /usr/local is correct? Jan 24 14:57:15 I meant why do you put it here Jan 24 14:57:43 qschulz: It isn't, it is just 3rd party library that my software depends on, and probably have hardcoded rpath Jan 24 14:57:45 Because I don't think the libs will be found without you adding LD_LIBRARY_PATH to your environment Jan 24 14:58:30 qschulz: there is custom detection of A paths in my B configuration file Jan 24 14:58:47 qschulz: Damn, it compiled! Jan 24 14:59:35 tolszak: then install it in /usr/lib (${D}${libdir} to be exact), less pain in the future :) Jan 24 14:59:39 tolszak: great :) Jan 24 15:00:19 qschulz: if libs provided as binaries have hardcoded rpath to /usr/include/lib .. then I don't want to struggle with it Jan 24 15:01:34 qschulz: But still, I'm astounded how consistent yocto is Jan 24 15:02:09 ugh... we patchelf those sometimes. Pre-built binaries are the worst Jan 24 15:04:57 We try hard to make it that way :) Jan 24 15:07:40 qschulz: I think, for me it will be safest to just not do it, those are also used on ubuntu etc, so I'm ok with just adding those paths in configurations files Jan 24 18:01:57 Hi. I have a question re: dnf on the target with an image built including package-management in the IMAGE_FEATURES. I'm using latest thud branch. Problem is that "dnf list installed" on the host comes back empty. I'm wondering if the database that dnf uses is split off into another package that I'm not including in the image? This used to work on pyro. Jan 24 18:02:19 dnf list installed on the target, I mean. Jan 24 18:03:26 @armver Jan 24 18:03:45 @arkver: Do you mean running dnf on the host or the target? Jan 24 18:04:22 @RobertBerger: the target, sorry for confusion Jan 24 18:04:27 Hello does anyone know the reason why qualcomm insists on using old versions of yocto in their BSP distributions? I'm working on something for the APQ8053 and it seems like the latest BSP available still uses Yocto Jethro with terribly broken multilib support. Jan 24 18:05:02 @arkver: Did you create a package feed on the host and tell the target where to find it? Jan 24 18:05:38 yes, that's working OK and dnf can see updated packages OK, but won't update anything since nothing is installed Jan 24 18:05:41 (it thinks) Jan 24 18:06:35 @arkver: I am a bit in the dark with dnf, since I usually use ipk/opk. Jan 24 18:07:48 @arkver: Would it be easy to switch pkg manager for a test and see if it works with another pkg manager? Jan 24 18:09:11 ok. With those, is including package-management (with the correct PACKAGE_CLASSES) enough? Jan 24 18:09:32 Here is what I do: Jan 24 18:10:46 1) IMAGE_FEATURES_append = " package-management" Jan 24 18:11:09 2) build core-image-minimal Jan 24 18:11:51 3) bitbake package-index Jan 24 18:14:08 4) expose the package feed via a web server e.g. in /var/www/html/imx6q-phytec-mira-rdk-nand/core-image-minimal-package-repo/3.0.1 a symlink like this: core-image-minimal-package-ipk -> /workdir/build/multi-v7-ml-debug-training-pkgs/tmp/deploy/ipk Jan 24 18:15:51 5) on target add to /etc/opkg/base-feeds.conf your feeds Jan 24 18:16:15 ok, I'm doing that, sort of. My image recipe has IMAGE_FEATURES += " package-management". I wonder if it really needs to be in the build's local.conf Jan 24 18:17:08 I'm using PACKAGE_FEED_[URIS|BASE_PATHS|ARCHS] which means that the conf file is included in the image automatically. Jan 24 18:17:18 5) the base feeds.conf looks like that: src/gz all http://192.168.42.105/imx6q-phytec-mira-rdk-nand/core-image-minimal-package-repo/3.0.1/core-image-minimal-package-ipk/all Jan 24 18:17:18 src/gz armv7a-neon http://192.168.42.105/imx6q-phytec-mira-rdk-nand/core-image-minimal-package-repo/3.0.1/core-image-minimal-package-ipk/armv7a-neon Jan 24 18:17:18 src/gz multi-v7-ml http://192.168.42.105/imx6q-phytec-mira-rdk-nand/core-image-minimal-package-repo/3.0.1/core-image-minimal-package-ipk/multi_v7_ml Jan 24 18:17:45 @arkver Jan 24 18:18:08 @arkver if you have a package manager in your image chances are it took it ;) Jan 24 18:18:23 And yeah, it has suitable lines for the package server. That part is OK. It's the initial list of installed packages that should imho have been generated during do_rootfs that's missing. Jan 24 18:20:26 @Maybe you need to do some kind of update/upgrade? Jan 24 18:21:30 @arkver: I usually do on the target opkg update and then the party starts Jan 24 18:22:07 @arkver: You could also try to build a new package, which is not in the rootfs and try to install that Jan 24 18:25:01 @RobertBerger: DNF syncs the with the package repo using: dnf distro-sync --refresh but fails to notice an updated version of a package that's already installed since it doesn't think it's installed. Will try adding a new one. I'll dig deeper to see what might have changed since pyro (which worked). Jan 24 18:26:00 @arkver: I remember something weird with rpms, let me check a bit Jan 24 18:28:33 @arkver: I guess you have a /etc/yum.repos.d file or so? Jan 24 18:28:44 I do indeed Jan 24 18:28:56 and it looks ok Jan 24 18:29:40 dnf --nogpgcheck makecache Jan 24 18:30:37 @arkver: I guess you don't sign your packages Jan 24 18:32:01 I sign the packages, but not the feed. gpgcheck isn't mentioned in the repo config file. makecache didn't magically restore the list of installed packages (sadly) Jan 24 18:33:20 dnf clean all Jan 24 18:33:36 dnf update -v Jan 24 18:35:30 that pulls down the metadata from the repo again (though there are some missing bits? deltainfo, updateinfo) Jan 24 18:35:37 list installed is still empty Jan 24 18:35:59 this will be something missing from the rootfs. dnf should know what's installed regardless of whether it can connect to a server. Jan 24 18:36:59 @arkver: I am afraid I am not much of a help here, since I rarely use dnf ;) Jan 24 18:37:54 Yeah. I recall I had some "fun" getting it to work right before (also for an imx6q incidentally). Thanks for the help anyway Robert. Jan 24 18:38:22 @arkver: Anyhow, I don't see anything obviously wrong in what you are doing. Maybe it's really a bug in this specific version of dnf. Jan 24 18:39:30 @arkver: I would just try with another package manager. If that works, you can dig into dnf. If that does not work, maybe something more fundamental is broken. Jan 24 18:39:59 Indeed, will give that a whirl too. Jan 24 19:01:34 JPEW: the last time i tried using bmaptool (following the instructions in the manual) it didn't work. i'm happy to see the manual has been updated with new instructions, and it now works :-D Jan 24 19:02:25 tgamblin: i haven't heard back yet, sorry, i'll ping him again on monday if i haven't heard from him by then (college course) Jan 24 19:02:31 Hi, I get this error when using devtool. "ERROR: Something went wrong with source extraction - the devtool-source class was not active or did not function correctly:" Jan 24 19:02:54 tlwoerner: Good... was that with a Rock Pi 4 then? Jan 24 19:03:51 JPEW: yes, i was just testing if calling the tool from a build environment (oe-run-native) worked Jan 24 19:04:16 "[Errno 2] No such file or directory: /poky/build/tmp/work///0.1+gitAUTOINC+2015F7073er15/devtooltmp-zc312g2f/initial_rev" Jan 24 19:04:50 I had deleted all the files inside workspace folder by mistake. Could that be causing this problems Jan 24 19:04:58 tlwoerner: no worries! Jan 24 19:05:05 How can I reset whole devtool to restart Jan 24 19:05:19 and regenerate stuff inside workspace folder Jan 24 19:06:08 @RobertBerger: I reflashed the system from the image and dnf list installed now shows the correct list. I assume the database had simply been trashed somehow previously. Sorry for the false alarm, and thanks again. Jan 24 19:36:00 JPEW: (ps i'm actually testing bmap on the tinker board since that patch comes before the rock pi 4 stuff, and i'm testing sequentially) Jan 24 19:51:18 tlwoerner: ok Jan 24 20:12:13 @arkver: good to hear that it was just a glitch in the matrix ;) Jan 24 20:58:12 Hi! Im working on a distro for a customer to use on a custom board. They want systemd which was no problem but they also want /usr/sbin/service and /usr/sbin/invoke-rc.d for a bunch of old stuff. Is there an easy way to add them? Jan 24 21:01:56 hello! poll for anyone that uses Yocto for embedded firmware - how long do you support your firmware releases? Jan 24 21:08:33 I've been supporting firmware built with buildroot for nearly a decade now. Jan 24 21:08:49 yocto is still new to me, though Jan 24 21:09:06 We have typical product lifetimes of 7-10 years Jan 24 21:10:47 fullstop JPEW thanks! Jan 24 21:11:03 which industry? Jan 24 21:11:20 Consumer Electronics Jan 24 21:11:28 ty Jan 24 21:11:32 automotive telemetry / fleet management Jan 24 21:11:37 ty Jan 24 21:12:52 JPEW in those 7-10 years do you maintain a snapshot of Yocto and back-port fixes all that time? Jan 24 21:13:02 or do you roll from one release of yocto to the next? Jan 24 21:13:44 radsquirrel: It depends. We have a wide number of product to support; some of them are fixed on a specific version, some roll forward to newer versions Jan 24 21:14:29 radsquirrel: Rolling forward is what we prefer; we only stick on older versions when we have to (which is more than we'd like ;) Jan 24 21:14:37 right Jan 24 21:15:07 I also have to support a given version for upwards of 10 years...was just curious if there are a bunch of people out there maintaining old yocto snapshots Jan 24 21:15:26 there doesn't seem to be much activity in the older releases. Jan 24 21:15:55 radsquirrel: I'd guess most people do the maintenance on a fork Jan 24 21:16:06 ok. bummer. Jan 24 21:16:28 radsquirrel: It it Jan 24 21:16:36 radsquirrel: It is :) Jan 24 21:20:22 I've backported a ton of stuff to a very old kernel, unfortunately. Jan 24 21:21:40 I was hoping everyone would tell me I'm crazy to do that. Jan 24 21:22:10 thx again for answering the poll! Jan 24 21:22:14 You assume that we are all sane. ;-) Jan 24 22:05:22 radsquirrel: I'll say you're crazy. I'd also like people to start sharing work... Jan 24 22:05:36 radsquirrel: I'm trying to make it easier for the core project to test older things Jan 24 22:06:16 RP: thanks! (for the former and the latter) Jan 24 22:17:16 RP: I wanted to add /usr/bin/core_perl to PATH in eSDK during install it need to find HOSTTOOLS which are in /usr/bin/core_perl Jan 24 22:17:56 but it does not respect the shell envs. So I think eSDK installer has its own settings for PATH if I knew I could stick it there Jan 24 22:21:20 eSDK installer core is the regular SDK installer Jan 24 22:21:30 so it's part of the populate_sdk stuff, embedded in all of that Jan 24 22:23:42 thats fine, I am looking for place where its setting it Jan 24 22:23:56 it doesnt matter where it does that Jan 24 22:24:33 it can look for things in /usr/local/bin e.g. and /usr/bin Jan 24 22:24:38 where is that logic Jan 24 23:32:09 OpenSorceress: nice nickname Jan 24 23:32:24 ;) Jan 25 00:39:16 Is there a way to tell toaster which python to use, i.e. to be able to a virtual environment? Installing globla packages (system or user) is ...so 2019. Or 2017 even. Jan 25 00:42:23 I can do it the old way, activating the virtual environment before calling any of toaster and bitbake. This way the python executable points to my venv and not the system **** ENDING LOGGING AT Sat Jan 25 02:59:59 2020