Robotnix | 248 Members | |
| Build Android (AOSP) using Nix | https://github.com/danielfullmer/robotnix | 76 Servers |
| Sender | Message | Time |
|---|---|---|
| 30 Mar 2024 | ||
| Tried to work on getting the nixpkgs version up-to-date with 23.11, but that's a challenge.. Stuck on trynig to get clang things to work with lineageos.. Something something ncurses not being found. | 23:08:24 | |
| But doing the same trick as in grapheneos requres a bigger change than expected. | 23:08:43 | |
| Also, not suer how ot avaid NIXPKGS_ALLOW_INSECURE, as some places "require" python2 for building. Mostly older Android versions. | 23:09:42 | |
| * Also, not suer how to avoid NIXPKGS_ALLOW_INSECURE, as some places "require" python2 for building. Mostly older Android versions. | 23:18:40 | |
| 31 Mar 2024 | ||
| eyJhb: If they become a bother, feel free to drop them | 01:31:22 | |
| Though if you can, I'd prefer simply only including them optionally depending on the android version | 01:31:46 | |
| The first thing I'm trying to do, is to get it to build by specifying the allow insecure envvar. Then I can check which ones I can bump to python2 | 08:17:49 | |
| * The first thing I'm trying to do, is to get it to build by specifying the allow insecure envvar. Then I can check which ones I can bump to python3 | 08:17:54 | |
| The current issue is the clang thing | 08:18:04 | |
| But I bumped the entire flake.lock, I don't think that's ideal. @Atemu is there a good reason for unstable and 21.11 pin? Couldn't it just be 21.11 or something? | 08:44:04 | |
| Or just be unstable. Not sure which is better. | 08:44:12 | |
| Nono other than that I haven't been able to make them work | 10:40:55 | |
| Fair :D | 10:55:55 | |
| It seems like lineage needs the same treatment as grapheneos after bumping to 23.11, as shown here https://github.com/nix-community/robotnix/blob/c54c511b3aa3e827e7ae320d604646a0ecc7c96c/flavors/grapheneos/kernel.nix#L25-L36 , I've just initially tested it by adding
To the | 12:37:36 | |
| Feel broken, mostly because I'm not sure what I am even doing. Atemu is lineageos using a prebuilt kernel, and the vanilla and grapheneos is not? Ie. they are built using robotnix? | 12:54:13 | |
| 13:03:13 | |
| Fun, now it just fails further about libncurses.so.5 complaints. Not sure what changed in nixpkgs, that warrants these errors all of the sudden | 13:03:36 | |
In reply to @eyjhb:eyjhb.dkI think so yes | 15:07:10 | |
| Ah no, nvm I remember: It has kernel sources that are built during the android build | 15:07:36 | |
| For Vanilla/GOS we use "prebuilt" kernels that we build in Nix | 15:07:59 | |
| I can't make any sense of the latest error I'm getting. I'm now just trying to bisect nixpkgs, to figure out why libncurses.so.5 suddenly can't be found. | 15:08:01 | |
| But I have never really touched Vanilla | 15:08:10 | |
In reply to @atemu12:matrix.orgMakes sense :) | 15:08:28 | |
| I'm going slightly insane over why ncurses5 suddenly can't be found when bumping nixpkgs. It doesn't make much sense to me. | 15:08:54 | |
| I vaguely remember ncurses being a bit cursed (heh) | 15:09:42 | |
| Indeed. | 15:09:49 | |
| https://github.com/NixOS/nixpkgs/issues/89769 | 15:10:03 | |
| GrapheneOS does SOMETHING funky with it, which I'm doing as well, then it passes the first part. But then fails with another libncurses.so.5 for the bcc_strip_attr | 15:10:28 | |
In reply to @atemu12:matrix.orgThis issue is from 2020, we're using a 2021 nixpkgs version . But might be the issue? | 15:11:18 | |
| https://github.com/NixOS/nixpkgs/issues/89769#issuecomment-1837374922 | 15:12:55 | |