 | Robotnix | 244 Members |
| Build Android (AOSP) using Nix | https://github.com/danielfullmer/robotnix | 75 Servers |
| Robotnix | 244 Members |
| Build Android (AOSP) using Nix | https://github.com/danielfullmer/robotnix | 75 Servers |
| Sender | Message | Time |
|---|---|---|
| 24 May 2024 | ||
 atemu12 | I might be getting ajs124 ' old Pixel 4a soon though at which point I could | 20:33:58 |
 jaen | Don't worry, with my abysmal track record you may not end up needing it (at least not for my sake) xD In the mean time I have created a small PR fixing Python script tests – https://github.com/nix-community/robotnix/pull/236. I didn't test building anything with those changes yet. | 20:48:20 |
| jaen | Here's how the formatting changes would look, if you're curious: https://github.com/jaen/robotnix/pull/1/files. I'll probably make a PR for this after the Python fixes are tested and merged, so I can also make a check for that. | 20:50:42 |
| atemu12 | The auto-generated files should perhaps be excluded? Otherwise, that does look pretty good to me already | 20:55:27 |
| jaen | I personally think it's okay if they get formatted — nix fmt with this setup runs pretty fast and GH checks/precommit hooks can always ensure you didn't forget to format anything. I'm also not quite sure which files are generated — apart from chromium vendoring, which is obvious — and but other than that I'd have to go through each file and see, which I'd rather avoid if I can. Unless it's a big deal and/or there's a list of them somewhere, then I guess I could do that then. | 21:28:00 |
| atemu12 | Not a big deal and those files are subject to removal anyways | 21:32:23 |
| jaen | Ok, so if I understand correctly, you want me to nuke tests for the LineageOS, right? What about those for mk_repo_file, should those stay or go? | 21:41:20 |
| atemu12 | Those can probably stay | 21:54:57 |
| atemu12 | The LOS updater tests are quite useless as they haven't been ran a single time in the past few years and therefore likely broken | 21:55:30 |
| jaen | Okay, pushed one more commit — patching derivation won't work, because it's just a shim apparently and it downloads it's code from the interwebs anyway. I did leave overlaying pkgs to a matching version, though (not sure if it matters, but why not). Other than that removed python2 from a couple of places that seemed safe and switched flake-compat to the nix-community version. Let me know if you can test this works for you and/or you have any other remarks. I tried to build some lineageos myself and at least it fetched the code and started building the ROM. It failed when trying to build Chromium, but I guess that's fare enough for it to not be related — would appreciate a double–check, though. | 23:04:49 |
| 25 May 2024 | ||
 eyJhb | Maybe i should try to bump the lineageos deps from my mirror... | 06:55:47 |
| eyJhb | In reply to @atemu12:matrix.orgI have a Pixel 8 now, but it's still in the box, and very likely to be sold. Doesn't seems right to keep it around for not using it as a daily driver 😅 Seems like aiding ewaste | 06:56:31 |
| atemu12 | In reply to @jaen:matrix.orgDisable chromium for now | 10:02:25 |
| atemu12 | I'll be removing it | 10:02:35 |
| jaen | Hmmm, doesn't Vanadium need that, though? As far as I can see there are no binaries available. So unsure if it's removable if we want to keep GrapheneOS? Thought that's a separate topic I guess, for now I just want to see if any ROM will build on my PR, currently running this to see if it builds: Don't have an appropriate phone to test it on, though. If there's anything else I should test or you can test it at least for the things you use and should know work — let me know. | 10:54:40 |
| jaen | Also not sure what we want to do about repo — like I mentioned in the PR, applying a patch to nixpkg's derication won't help, because the resulting package is basically just the main repo script and it downloads all the rest from the interwebs on use. So at least for this PR we'd have to live with it downloading stuff, but we probably need to figure a more official place to put it, instead of on my GitHub? Not sure if nix-community would be appropriate? | 10:57:13 |
| jaen | Here's the reformat PR FWIW — https://github.com/nix-community/robotnix/pull/237. It's based on top of the Python fixes, so they will have to go in first. I'll make one more PR to re-enable running those checks in CI (and maybe pre-commit hooks — I'm ambivalent on them personally, but some find them useful) and then I'll maybe start looking at GrapheneOS finally, though who knows how well will that go, it's not quick to iterate on without a 32 core Threadripper xD | 14:12:01 |
| atemu12 | No idea w.r.t. Vanadium, I have never used GOS | 15:29:23 |
| atemu12 | My view on things is that if nobody has been around to maintain it for years, it's better to just remove it | 15:30:36 |
| atemu12 | Neither Vanadium nor GOS have been touched by anyone the past few years | 15:31:03 |
| atemu12 | As for repo, is it really not possible to run it without it attempting to download some other version? I'd prefer to patch it to not download any updates whatsoever | 15:32:54 |
| jaen | Well, on the other hand I never used LineageOS ; d What I'm just saying that unless you want to drop GrapheneOS wholesale, then it might need to stay. At least it's how it looks to me. I guess I don't care much either way, since I can try fixing it to the side anyway, or something. | 18:13:50 |
| jaen | In reply to @atemu12:matrix.orgI would've preferred to not have done that particular yak shave, but oh well — https://github.com/nix-community/robotnix/pull/236/commits/4a347287aa2812d50ada2c87c7284e59617c9a41. This seems to work and have local sources as far as I can tell. Kind of ugly with the git shenanigans, but didn't see a better way so far. | 18:15:06 |
| 26 May 2024 | ||
| jaen | Ok, got three basic checks (running
| 13:47:23 |
| atemu12 | I'm not a fan of pre-commit hooks. When I commit, I should commit precisely what I saw in the staging area. They're also a huge security liability which is why I've disabled them globally and I think that it's insane that git enables them by default. | 14:06:06 |
| atemu12 | CI should be red on wrong format, that's good enough. | 14:06:24 |
| atemu12 | I have no idea about flakes. | 14:06:58 |
| jaen | Yeah, like I said I'm mostly ambivalent on them, so if you don't want them, then I don't feel like I need to add them. If you don't have an opinion re: subflake dependencies, then I'll try adding it and see how it works out, I guess? | 14:21:02 |
| atemu12 | Sure, knock yourself out | 14:21:33 |
| atemu12 | As long as I can nix-build, I'm happy. | 14:21:51 |