| 2 Jan 2025 |
countoren | and I am able to access wifi, serial from the service without any other config | 20:34:58 |
countoren | using crossPkgs, pkgsStatic | 20:36:03 |
oak 🏳️🌈♥️ | Properly packaging a native low-level Linux-daemon for Android is a process which involves multiple kinds of steps, for example you need to provide the necessary SELinux policy in addition to other things | 20:36:26 |
countoren | yes we had to do that too | 20:37:13 |
oak 🏳️🌈♥️ | Builds done with pkgsStatic won't use the Android's bionic libc, but instead something that is provided by nixpkgs, statically built into the binary | 20:37:52 |
atemu12 | nix-daemon when? | 20:38:23 |
countoren | yep | 20:38:51 |
oak 🏳️🌈♥️ | For quick hacking that is probably fine, but if you want to properly package software for Android System then it should be built as part of the Android tree | 20:39:10 |
oak 🏳️🌈♥️ | For example if you are actually aiming to productize something officially | 20:39:21 |
atemu12 | I mean, we'd just use the nixpkgs ones and add the selinux rule to allow the Nix store just like on Fedora | 20:39:45 |
atemu12 | Or perhaps add a permission for apps to use it | 20:40:02 |
countoren | it will be the actual differences | 20:40:27 |
countoren | * what will be the actual differences | 20:40:37 |
countoren | * what will be the actual differences? | 20:40:47 |
atemu12 | I think we're squarely in the "hacking stuff" field rather than anything close to "productise" or "officially" | 20:41:00 |
atemu12 | I think they mean to build whatever software you want to run as a daemon against the AOSP sources such that it's like the other daemons in Android | 20:41:56 |
atemu12 | When in Rome, do as the Romans do | 20:42:05 |
oak 🏳️🌈♥️ | Build done inside the Android tree will use the same compilers and other same tooling that is used to build the rest of the system | 20:42:33 |
countoren | * what will be the actual differences? like any special restrictions ? | 20:42:36 |
countoren | there is one that i know of | 20:42:52 |
countoren | I had harder time to use logcat in the app | 20:43:21 |
countoren | ended up using spdlog instead | 20:43:32 |
countoren | * I had harder time to use logcat in the service | 20:43:57 |
countoren | if building from the tree you should be able using dynamic libs of the device without conflicts. but we are living in the nix world so we usually have nice set of options to replace android libs with. | 20:49:26 |
countoren | I guess it depends what you need to do. you might get to a "dead end" | 20:50:02 |
| 7 Jan 2025 |
| pentane (DECT CYPT/2978) joined the room. | 10:49:24 |
| 9 Jan 2025 |
atemu12 | oak: I just talked to Jonas and he said that it's likely okay for robotnix to use nix-community builders for CI but we'd need to see how it turns out in practice to know for sure | 13:09:52 |
atemu12 | So if you want to go ahead with adding CI, please feel free :) | 13:10:07 |
oak 🏳️🌈♥️ | Need to check out what would be the CI tools they provide, like do they provide buildbot or what? | 22:03:44 |
oak 🏳️🌈♥️ | Answering my own question: https://nix-community.org/continuous-integration/ | 22:04:42 |