**** BEGIN LOGGING AT Thu Nov 05 02:59:57 2020 Nov 05 07:34:32 good morning Nov 05 07:47:29 yo dudX Nov 05 07:47:42 hi LetoThe2nd Nov 05 10:06:28 join #poky Nov 05 10:06:40 sorry :) Nov 05 10:07:09 Hey guys and gals, I wanted to ask for some help, is this a good place? Nov 05 10:08:04 I am trying to install vboxguest additions for a NI Linux RT system which is based on Poky (I think), so I followed the guide found in here: https://git.yoctoproject.org/cgit.cgi/poky/plain/meta/recipes-core/images/build-appliance-image/README_VirtualBox_Guest_Additions.txt Nov 05 10:08:33 Unfortunately there are some missing files (probably more than one) and I cannot finish the ```make scripts``` step Nov 05 10:09:59 For example `classmap.h` is missing. And it is not present on the system, I searched for some similar problems on Fedora distribution but there are no present solutions, they seem to struggle with similar problems as well. Nov 05 10:28:14 nlhnt: well the readme you linked specifically refers to the build-appliance image, so it probably does not work on any given poky-based build Nov 05 10:28:17 see: https://git.yoctoproject.org/cgit.cgi/poky/plain/meta/recipes-core/images/build-appliance-image_15.0.0.bb Nov 05 10:28:53 this includes kernel-dev and kernel-devsrc, which are probably both not present in your image Nov 05 10:31:40 letothe2nd: Ok, will do, thank you! :) Nov 05 10:57:35 What does the SH in SHLIBS stand for? I knew but it slipped my mind. Nov 05 10:58:00 um, "shared"? Nov 05 11:12:35 LetoThe2nd: Ah, yes.. Now I remember ;) Thx Nov 05 11:13:21 I remember it was something pretty obvious when I first learned about it Nov 05 11:13:30 Still I forgot Nov 05 11:15:50 :) Nov 05 12:39:43 fray: hey, are you around to discuss meta-xilinx-standalone ? :) Nov 05 15:05:10 epyc 7ksomething for build server? good or bad idea? Nov 05 15:05:30 Hi, I'm having trouble with cross-linking a trivial c program with the yocto SDK: I get the error message "cannot find crti.o: No such file or directory" despite including it in my IDE Nov 05 15:07:32 guest252: are you properly passing the sysroot, repectively using the environment setup script? does it work manually (e.g. without the ide?) Nov 05 15:09:41 aarch64-poky-linux-gcc -L/opt/poky/3.0.4/sysroots/aarch64-poky-linux/lib -L/opt/poky/3.0.4/sysroots/aarch64-poky-linux/usr/lib -L/opt/poky/3.0.4/sysroots/aarch64-poky-linux/usr/lib/aarch64-poky-linux/9.2.0 -o "test" ./src/test.o Nov 05 15:10:09 these are the sysroot directories Nov 05 15:10:25 I'm not using the environment setup script Nov 05 15:12:12 guest252: then thats the problem. use the script, its there for a reason. Nov 05 15:12:26 and AFAIK, passing random -L flags is not the same as passing --sysroot Nov 05 15:15:41 guest252: you don't *need* to use the environment script, but you *do* need to pass --sysroot Nov 05 15:16:11 because having 1 sysroot in a SDK is basically a nice edge case Nov 05 15:16:29 multilib, multiarch, multitune, all have multiple sysroots Nov 05 15:17:09 Okay, that makes sens rburton, thank you Nov 05 15:17:15 rburton: so my AFAIK strikes! Nov 05 15:17:46 yeah, gcc needs to find its internal libraries and stuff Nov 05 15:19:00 * LetoThe2nd feels mightily proud. Nov 05 15:24:20 Hm, it's probably easier if I actually use the environment script, the alternative is copy&paste anyway, so does anyone know a tutorial to include this script in Eclipse? Nov 05 15:25:30 * LetoThe2nd knows how to get rid of eclipse Nov 05 15:26:43 when I used a SDK inside vscode the trick was to tell it that the compile command was . environment.. && $CC Nov 05 15:27:14 Ah, that's actually a great idea, thank you Nov 05 15:27:48 rburton: sounds legit, or robs latest blueprint for vsc Nov 05 15:28:30 yeah that is next-level Nov 05 15:36:09 I was wondering if this is the style of code you are using inside bitbake python: https://pastebin.com/6z2438FY instead of `return fn in self.clean` ? Nov 05 15:36:59 Oups that pastebin has expired. Nov 05 15:38:30 https://pastebin.com/6tqdUcis Nov 05 15:40:45 roussinm: it's not a question of general style, just individual choices about readability. generally we favor idiomatic python, but we try not to sacrifice readability just for conciseness. it's case by case. Nov 05 15:42:34 * zeddii nods Nov 05 15:43:13 I've read too much perl/script/python code where the author is just trying to show how clever they are, and make things as small/concise/single line as possible. Nov 05 15:43:16 kergoth: what about usage of functions like `all` and `any` ? Nov 05 15:43:36 we've used them on a number of occasions and continue to do so Nov 05 15:43:47 zeddii: precisely, they often sacrifice readability and make maintenance harder in the doing Nov 05 15:44:06 we aren't going to be clever just because the stdlib has a function. if it makes sense and it's idiomatic and readable, we'll use it Nov 05 15:44:47 certainly patches to make the codebase more idiomatic would be considered for sure Nov 05 15:46:22 We all have threshold of readability I guess, I would prefer using `return MYCONDITION` instead of spliting false/true. I'm used to the second syntax because I use a clang-tidy (C++) check that requires to "simplify" statement. Nov 05 15:46:46 *first syntax. Nov 05 15:47:18 IMO it depends on the complexity and readability of the condition Nov 05 15:48:37 I love people who use obfuscators.. 'Don't steal ma code, I'm an eleete haxor'.. dude, it's an interpreted language, the code is open.. Nov 05 15:48:53 you can always extract to a function or variable to put a name on the condition which should simplify. Nov 05 15:48:54 if not not foo: do this. Nov 05 15:49:07 zeddii: oh god. don't. Nov 05 15:49:17 signed-off-by: his-eliteness@elite.org Nov 05 15:49:18 if not not not foo: Nov 05 15:49:30 :D Nov 05 15:49:41 if not not not foo or not not foo: Nov 05 15:50:30 (actually I was going through some code a few weeks ago, where the code was something like foo = not bar \n (about a dozen lines later) if foo or bar: .... Nov 05 15:50:35 so always do this? Nov 05 15:51:04 zeddii: you're not really eleet until you get your PoCs back from customer presentations with the commend "worked nicely, but they asked us what foo and bar meant" Nov 05 15:52:02 :D Nov 05 15:52:14 for the record: yes, this has happend to me. Nov 05 15:52:49 it definitely level sets for the rest of the conversation, doesn't it ? Nov 05 15:53:45 ayup Nov 05 15:55:02 kergoth: https://clang.llvm.org/extra/clang-tidy/checks/readability-simplify-boolean-expr.html that's the check I'm using in our code base. I don't have much experience in Python, does python has similar tooling? Auto-refactoring, Auto-modernization? Nov 05 15:55:11 One thing I do, and have been called on it numerous times.. "for k,v in pairs(table)" (lua code).. "Please expand what k and v do".. for a lot of cases, that makes sense, cause the key and value do have meaning... but I wrote a function that simply copied a structure.. got the same comment and say "stands for key and value, since it can copy any table..." Nov 05 15:55:43 so it was clear the reviewer just went 'k,v' but enver looked at the comment to the function that says 'copy a table' Nov 05 15:57:04 roussinm: https://github.com/python-rope/rope? Nov 05 15:58:03 rburton: Yes I use rope inside VSCode. That's the only one I use with PyLance server. Nov 05 16:07:49 Pylance has been very usefull to find unused imports and unused variables, but I don't think I understand how import works in bitbake because when I experiment by removing unused imports in a module, another module fails because it's missing that import. It feels like transitive import dependency? Maybe that's the python way? Nov 05 16:09:21 halstead: is patchwork known to be broken or is it just really slow? https://patchwork.openembedded.org/project/oe/patches/?submitter=&state=*&q=&archive=both&delegate= shows last patch from 2020-10-30 and there were many patches on ML after that Nov 05 16:14:10 JaMa, I keep repairing patch processing and then it fails again. I'm not sure why this time. Nov 05 16:15:51 kergoth: the maximum python version supported in BitBake atm is 3.6, correct? Nov 05 16:16:30 Email to OE patchwork-next is bouncing again just like on Oct 17th. I need to change this set up instead of repairing. Nov 05 16:16:34 roussinm: https://github.com/openembedded/bitbake/blob/master/lib/bb/__init__.py#L15 Nov 05 16:16:40 roussinm: sadly we can't use f-strings yet :) Nov 05 16:17:24 kergoth: :( Do you have a ETA on when it's going to be possible? Because even Debian has Python 3.7, debian is always behind usally. lol Nov 05 16:18:04 would have to ask RP, but oe+bitbake is expeted to run on a wide variety of hosts, so it always tends toward a certain level of compatibility Nov 05 16:18:12 would be nice, though Nov 05 16:18:46 it would improve readability on a lot of strings ;) Nov 05 16:35:27 JaMa, It's starting to catch up now. Nov 05 16:44:52 halstead: thanks! any update on -commits MLs? Nov 05 16:54:36 Question about wic: when calling wic from the command line, it uses the NATIVE_SYSROOT from the wic-tools recipe, right? But when wic is called from a run.do_image_wic script will it use the NATIVE_SYSROOT from the image (so recipe-sysroot-image)? Nov 05 16:55:38 fray: oh hello Nov 05 16:55:42 fray: you still around ? Nov 05 16:55:46 marc1: yes Nov 05 16:56:35 fray: I have a question about some interpreted language (python) scripts (meta-xilinx-standalone) :-) Nov 05 18:15:11 is there a bgg with PN being only [0-9-]+ and ALLOW_EMPTY_${PN} = "1" breaking? (thud, haven't checked newer; package_rpm.bbclass seems pretty stable; pretty sure this didn't happen in rocko) Nov 05 18:15:21 s/bgg/bug/ Nov 05 18:53:37 do we have a tool that would compare the pn-buildlist with the recipes really used for an image ? Nov 05 18:54:39 I'd think a first step would be to include the output of "bitbake -g" in buildhistory, and write this as a buildhistory-analysing script ? Nov 05 18:56:00 * yann does not like to spend time waiting for the build of useless stuff, especially when bisecting Nov 05 18:56:02 Are you just trying to get a list of the recipes that were used as build tools? Nov 05 18:56:14 nothing is built that isnt in the dependency graph. if it's there, something depended on it Nov 05 18:56:19 so i'm not sure what you're looking for Nov 05 18:56:32 you can't just blindly remove dependencies, unless they're pulled in through PACKAGECONFIG and you can opt out of those fatures Nov 05 18:56:33 features Nov 05 18:57:28 eg. lmsensors default to building sensord, which pulls rrdtools, which pulls perl - I just noticed that by chance, glancing idly at the bitbake progress and gaping at the sight of rrdtool Nov 05 18:58:36 that's easily fixed by a PACKAGECONFIG tweak, but I'm sure there are many such occurences, and we should have a way to see that easily Nov 05 18:59:15 and yes there are cases where we just did not add the PACKAGECONFIG items yet, I'm sure :) Nov 05 19:01:37 looking at task-depends.dot I can see I have some more work if I do want to avoid building perl, which does not every reaches my rootfs, it's sad not to be warned :) Nov 05 19:03:53 oftentimes packageconfigs add runtime dependencies as well, which *would* end up in the image, and yet is still an optional dep you might not want Nov 05 19:04:14 and warning about everything you build that doesn tgo in the image will show you hundreds of native and cross tools. i guess if you filter it to target. Nov 05 19:18:37 RP: it seems the python3 recipe staged in master-next needs some rework http://errors.yoctoproject.org/Errors/Build/112196/ Nov 05 19:19:38 Guest9816: yes, sorry. Let me drop that lot Nov 05 19:19:47 it was an experiment which didn't work Nov 05 19:20:39 khem is in disguise these days Nov 05 19:20:40 armpit: the license errors I mentioned ^^^ Nov 05 19:54:22 zeddii: I use matrix for IRC these days and it forgets nicks Nov 05 19:59:07 RP: I have some patches for master to fix my world builds on ml Nov 05 20:16:56 Hi! I run basic core minimal image x86 runqemu and want to SSH into it. I installed ssh-server-dropbear but ssh root@192.168.7.2 gives "connection refused" Nov 05 20:18:52 EXTRA_IMAGE_FEATURES ?= "debug-tweaks ssh-server-dropbear" Nov 05 20:19:17 ping 192.168.7.2 works fine Nov 05 20:23:53 is it actually running? Nov 05 20:29:43 how can I check it? Nov 05 20:30:20 Starting Dropbear SSH server: dropbear. Nov 05 20:30:21 I see this on qemu start Nov 05 20:31:14 ip a -> inet 10.0.2.15/24 which looks strange, ip should be 192.168.7.2 Nov 05 20:33:41 oh sorry qemu Nov 05 20:33:48 no, 10.0.2.15 is the right ip Nov 05 20:33:56 192.168 presumably is your machine's ip? Nov 05 20:34:52 - You can access the host computer at 192.168.7.1 within the image. Nov 05 20:34:52 - Your qemu system will be accessible as 192.168.7.2. Nov 05 20:35:04 this is from poky/scripts/runqemu.README Nov 05 20:35:07 well, in that case, the qemu networking setup is a bit duff Nov 05 20:36:21 can I check somehow that sshd service is running? Nov 05 20:39:19 ps ax Nov 05 20:39:29 but the IPs not matching might be the problem Nov 05 20:39:49 runqemu assumes that you've got the qemu networking config files installed in the image Nov 05 20:39:58 if you don't then networking will be different Nov 05 20:40:40 201 root 2924 S /usr/sbin/dropbear -r /etc/dropbear/dropbear_rsa_host_key -p 22 -B Nov 05 20:43:03 I did not provide any networking config for the image, should I? Nov 05 20:45:28 Can I provide something useful here ? runqemu qemux86 qemuparams= Nov 05 22:29:33 DanmerZ: Can you ssh to localhost from inside qemu? **** ENDING LOGGING AT Fri Nov 06 02:59:57 2020