| 5 Sep 2021 |
 pie_ | nitpick, https://docs.robotnix.org/ still mentions freenode | 17:07:56 |
| pie_ | danielrf: btw, isnt --arg applied everywhere/every function invocation? im always a little weirded out when someone recommends using it | 17:44:07 |
| pie_ | meanwhile, turns out the person offering me the beefy machine for builds cant offer it anymore, so i again need to find someone that can give me build time | 18:06:32 |
| pie_ | * nitpick, https://docs.robotnix.org/ and https://docs.robotnix.org/modules/flavors.html still mentions freenode | 18:14:16 |
 samueldr | --arg passes the --arg value only to functions it directly encounters AFAIUI; e.g. if the expression you evaluate is a list of function, it will pass the arguments to the functions from that list, but not further.;
This mirrors the nix-build behaviour where building a list of derivations build all of them; if you have a list of expressions needing to be parameterized to produce derivations to build, I guess it makes sense to provide args to all of them | 18:34:09 |
| samueldr | (I'm going from the description in the man page, haven't actually seen --arg being given to anything else than a top-level function, and I've been using --arg extensively in Mobile NixOS) | 18:35:53 |
 danielrf | In reply to @jcie74:matrix.org
nitpick, https://docs.robotnix.org/ and https://docs.robotnix.org/modules/flavors.html still mentions freenode Thanks, just pushed a fix that should propagate to docs.robotnix.org soon | 18:36:38 |
| danielrf | My understanding is about --arg matches what samueldr said. The original idea was to have configuration as a top-level argument, so that it matched what NixOS does in nixpkgs/nixos/default.nix. | 18:37:50 |
| danielrf | The newer nix command line interface when using flakes does not support --arg on the commandline, you currently have to make a flake.nix and call robotnix.lib.robotnixSystem | 18:39:46 |
| danielrf | But if the configuration is short enough (and I'm just testing something out), I like the convenience of specifying the options directly on the command line, as opposed to making a .nix file every time. | 18:42:02 |
| samueldr | (off-topic) huh, I didn't know that about flakes... a step backward in UX imho | 18:42:42 |
| danielrf | kranzes: Hey, that's great! If you're interested in figuring out exactly what is breaking it in the lineageos configuration, you can start commenting out the sections that don't seem needed and testing. Bth, my best guess now actually is the webview configuration overrides we do in the lineageos flavor | 18:46:50 |
| danielrf | * kranzes: Hey, that's great! If you're interested in figuring out exactly what is breaking it in the lineageos configuration, you can start commenting out the sections that don't seem needed and testing. Tbh, my best guess now actually is the webview configuration overrides we do in the lineageos flavor | 18:46:57 |
| danielrf | samueldr: I have mixed feelings about it as well. Things that should be easy take a bit of boilerplate to make work. Reproducibility/purity benefits outweigh the costs for me though. | 18:48:16 |
 kranzes | That would take me ages | 18:48:18 |
| kranzes | every build is 3 hours | 18:48:26 |
| kranzes | and i don't even know where to start | 18:48:42 |
| danielrf | up to you if you want to continue tracking it down. The long turnaround time is an unfortunate thing we deal with on android. | 18:50:21 |
| danielrf | But I'd probably start with commenting out the webview stuff like so: | 18:50:38 |
| danielrf | diff --git a/flavors/lineageos/default.nix b/flavors/lineageos/default.nix
index 1216ca8..5407fc8 100644
--- a/flavors/lineageos/default.nix
+++ b/flavors/lineageos/default.nix
@@ -106,9 +106,9 @@ in mkIf (config.flavor == "lineageos")
})
];
- # LineageOS will sometimes force-push to this repo, and the older revisions are garbage collected.
- # So we'll just build chromium webview ourselves.
- "external/chromium-webview".enable = false;
+# # LineageOS will sometimes force-push to this repo, and the older revisions are garbage collected.
+# # So we'll just build chromium webview ourselves.
+# "external/chromium-webview".enable = false;
}
] ++ optionals (deviceMetadata ? "${config.device}") [
# Device-specific source dirs
@@ -134,16 +134,16 @@ in mkIf (config.flavor == "lineageos")
source.manifest.url = mkDefault "https://github.com/LineageOS/android.git";
source.manifest.rev = mkDefault "refs/heads/${LineageOSRelease}";
- # Enable robotnix-built chromium / webview
- apps.chromium.enable = mkDefault true;
- webview.chromium.availableByDefault = mkDefault true;
- webview.chromium.enable = mkDefault true;
-
- # This is the prebuilt webview apk from LineageOS. Adding this here is only
- # for convenience if the end-user wants to set `webview.prebuilt.enable = true;`.
- webview.prebuilt.apk = config.source.dirs."external/chromium-webview".src + "/prebuilt/${config.arch}/webview.apk";
- webview.prebuilt.availableByDefault = mkDefault true;
- removedProductPackages = [ "webview" ];
+# # Enable robotnix-built chromium / webview
+# apps.chromium.enable = mkDefault true;
+# webview.chromium.availableByDefault = mkDefault true;
+# webview.chromium.enable = mkDefault true;
+#
+# # This is the prebuilt webview apk from LineageOS. Adding this here is only
+# # for convenience if the end-user wants to set `webview.prebuilt.enable = true;`.
+# webview.prebuilt.apk = config.source.dirs."external/chromium-webview".src + "/prebuilt/${config.arch}/webview.apk";
+# webview.prebuilt.availableByDefault = mkDefault true;
+# removedProductPackages = [ "webview" ];
# Needed by included kernel build for some devices (pioneer at least)
envPackages = [ pkgs.openssl.dev ] ++ optionals (config.androidVersion == 11) [ pkgs.gcc.cc pkgs.glibc.dev ];
| 18:50:48 |
| danielrf | * up to you if you want to continue tracking it down. The long build/test cycle time is an unfortunate thing we deal with on android. | 18:51:33 |
| pie_ | samueldr: danielrf https://gist.github.com/cleverca22/240c21dab528666c85a95c4edfdeae58 | 19:13:40 |
| pie_ | re: --arg | 19:13:43 |
| pie_ | i couldnt remember where to find the relevant part of the source though | 19:13:53 |
| samueldr | ah, I think that's because they float up to the top-level | 19:15:00 |
| samueldr | or something like that | 19:15:26 |
| samueldr | though I guess it's less of an issue than could be thought, as I've literally never seen that being an issue | 19:15:45 |
| pie_ | i think what happens is that things passed via such get added to a global autoargs variable | 19:15:51 |
| pie_ | yeah im not sure why it never appears to be a problem | 19:16:10 |
| pie_ | maybe they are only set when an argument isnt otherwise passed or something? | 19:16:28 |
