| 24 Nov 2024 |
 oak | * However, upgrade to LineageOS 21 on my dailydriver instantnoodlep went just fine, and it fixed the issue about annoying long boot time because some sensors couldn't be found | 02:22:45 |
 atemu12 | Cool! | 09:31:04 |
| atemu12 | Yeah we really ought to have a webview | 09:31:20 |
| atemu12 | https://github.com/nix-community/robotnix/issues/253 | 09:32:00 |
| atemu12 | Oh, I duplicated https://github.com/nix-community/robotnix/issues/246 | 09:32:27 |
| atemu12 | Well, it overlaps | 09:33:01 |
| 25 Nov 2024 |
| oak | I wonder how much work it would be to just fix the webview building and upgrade it to newer version at the same time | 00:00:42 |
| oak | Few weeks ago when I started looking at things, the first thing I tried to fix was the webview thing but then I just disabled it because I was told it is broken :D but then the errors I saw were related to something like when you upgrade nixpkgs and have too new versions of tools, for example the Python build scripts used features that were already removed from Python 3.11 | 00:15:36 |
| oak | * Few weeks ago when I started looking at things, the first thing I tried to fix was the webview thing but then I just disabled it because I was told it is broken :D but the errors I saw were related to something like when you upgrade nixpkgs and have too new versions of tools, for example the Python build scripts used features that were already removed from Python 3.11 | 00:16:03 |
| oak | Okay, it looks like the build process and build scripts of the webview are even more complex than the Android itself :D | 00:38:43 |
| oak | * Okay, it looks like the build process and build scripts of the webview are even more complex than the stuff for Android itself :D | 00:38:58 |
| oak | This seems to be another rabbit hole | 00:52:32 |
| oak | Apparently something has changed inside the Chromium build scripts because the mk-vendor-file.py script works for the old version but not for the newest version | 00:53:49 |
| atemu12 | I really don't know, I've never ever touched that stuff | 01:25:13 |
| oak | Pain, they have added their own "vpython3" distirbution to the depot_tools of Chromium, I tried updating depot_tools since it was from 2020 | 02:54:36 |
| oak | Maybe it is possible to get it working if I autoPatchelf the binaries or something | 02:55:23 |
| oak | Or try to create similar Python environment with Nix and then patchShebangs | 02:57:30 |
| atemu12 | I thought chromium was built inside of an FHS container? | 02:58:52 |
| atemu12 | Like the android build? | 02:59:00 |
| oak | Can I use normal dynamic linking if I create FHS shell? | 02:59:33 |
| atemu12 | Yup | 02:59:52 |
| atemu12 | That's what it's for | 02:59:57 |
| oak | That's a thing to try tomorrow | 03:00:21 |
| oak | This mk-vendor-file.py calls some depot_tools/gclient.py flatten --pin-all-deps tool, and it seems to use some bundled "vpython3" tool, which is a virtualenv + some version of Python3 I think | 03:01:55 |
| oak | And they extensively do just call the process vpython3 everywhere | 03:02:41 |
| oak | They expect to find executable vpython3 in PATH | 03:03:10 |
| oak | They had repo tool, but then they had to write different tool for Chromium project | 03:04:05 |
 ryantrinkle | What's the smoothest phone and android to use with robotnix these days? | 16:58:52 |
| ryantrinkle | like, least likely to run into issues with building, updating, etc. | 16:59:16 |
| ryantrinkle | i'm on graphene now, but haven't updated in a long time | 16:59:35 |
