**** BEGIN LOGGING AT Mon Feb 18 02:59:56 2019 Feb 18 08:41:34 RP: I think I found an issue concerning .pyc files in both the 3.5.x and the new 3.7.2 recipes: The file modification timestamp in the .pyc files lags behind the file filestamp of the actual .py files and thus precompiling them is of no use. Feb 18 08:54:37 tristanram: not good :(. Any idea where the discrepancy comes from? Feb 18 08:56:53 RP: Not yet but I only noticed it because I most use readonly rfs. Do you know there the .pyc files are created during the build? For python2.7 it was in distutils.bbclass, but for python3, I have not yet found it. Feb 18 09:04:15 RP: Also, it happens on a build from scratch of the entire image and an incremental build as well. Upon invoking 'python3 -vv -c ""' on target, I get many messages such as "bytecode is stale for 'site'". Feb 18 09:05:17 RP: As expected, python3 then tries to create the .pyc again but on my system it cannot write it: could not create '/usr/lib/python3.7/__pycache__/site.cpython-37.pyc': OSError(30, 'Read-only file system') Feb 18 09:07:04 RP: Python3 works fine, but i think this causes a high CPU load at every launch of the interpreter :( Feb 18 09:07:48 RP: For reading the timestamp from pyc files i use: https://qiita.com/amedama/items/698a7c4dbdd34b03b427 Feb 18 09:08:57 RP: With python 3.7.2, the header has one additional 4-byte value before the 4-byte mtime value: https://www.python.org/dev/peps/pep-0552/ Feb 18 09:10:14 RP: So you will have to adjust the script to read 1 more 4 byte value between magic_str and mtime_str (in case you wanna test it yourself) Feb 18 09:19:04 tristanram: you might want to talk to kanavin on #yocto Feb 18 09:19:24 tristanram: unfortunately my knowledge is out of date with this Feb 18 09:21:26 tristanram: I'd love to have tests on the readonly rootfs for this btw Feb 18 09:21:36 tristanram: we seem to break it too often Feb 18 09:22:13 RP: Ok, i will message him. Feb 18 09:23:26 RP: It seems pretty easy to test since only two numbers would have to be compared for every .pyc file at some late stage in the build. Feb 18 11:22:35 hm. im bitbaking, and had an error earlier where a build failed due to missing bdftopcf. i installed xfont-utils and that made the error go away, but i forgot to document the error i got, which i want to do. Feb 18 11:22:57 i font remember which package/recipe it was in, but i figure that if i clean i will just rerun everything Feb 18 11:23:08 but when trying to run bitbake -c clean im told "Nothing to do. Use 'bitbake world' to build everything, or run 'bitbake --help' for usage information." Feb 18 11:23:32 same with -c listtasks , i get the same output. why is this? is there really no clean task to run? Feb 18 11:24:12 i dont even get the task "ov-image" which i know exists because thats what i run to start building (bitbake ov-image). shouldnt at least that task be listed? Feb 18 11:30:11 ok i found occurrences of bdftopcf in a bunch of recipes/packages under the tmp folder now, so i have a bunch of package names. i'd like to clean one or two of them to reproduce the error. but how do i do that if clean isnt a task thats available? Feb 18 11:30:40 for example terminus-font is one that apparently depends on bdftopcf being installed to build Feb 18 11:39:07 * rawtaz threw it all away and started over Feb 18 16:22:31 any suiggestions for fixing this message? Feb 18 16:22:33 Gtk-Message: 15:06:26.518: Failed to load module "pk-gtk-module" Feb 18 16:22:33 Gtk-Message: 15:06:26.519: Failed to load module "canberra-gtk-module" Feb 18 16:22:43 after starting a python app Feb 18 16:48:03 tristanram: having an automated test for it would be great, just takes someone a little effort to write it, then we stop it regressing **** ENDING LOGGING AT Tue Feb 19 02:59:57 2019