| 8 Nov 2024 |
 K900 | Yeah I think I have a fix | 13:13:29 |
| K900 | Verifying onw | 13:13:32 |
| K900 | * Verifying now | 13:13:33 |
| K900 | I am also on aarch64 so it takes forever to build | 13:14:05 |
| K900 | There's no way it'll cross-compile, right | 13:16:09 |
| K900 | Wait what the fuck | 13:19:04 |
| K900 | Why is it building two of them | 13:19:07 |
 @hexa:lossy.network | In reply to @mattleon:matrix.org
Another issue I discovered: since 2024.10.3, some change to nix's toYaml function caused "on" and "off" to be serialized to yaml as on and off, whereas they used to be 'on' and 'off'. It looks like home-assistant uses yaml 1.1, where these are literal bools, and nix is now outputting yaml 1.2, where quoteless on and off are interpreted as string literals. I tried double quoting them in nix, but then it's serialized to a double-quoted string in yaml, which doesn't parse properly in hass either remarshal update: https://github.com/remarshal-project/remarshal/issues/52 | 13:20:21 |
 mattleon | I'm quite surprised the matter-server module tests didn't pick up on the missing dylib... I'll need to investigate and see if there's a weakness in the tests | 13:33:00 |
| K900 | https://github.com/NixOS/nixpkgs/pull/354491 | 13:44:36 |
| K900 | 
Download 1000004117.png | 14:16:03 |
| K900 | Oh hey it actually worked | 14:16:06 |
| @hexa:lossy.network | so one thought about remarshal is that we could patch hass to use yamlcore and thus interpret things as yaml 1.2 | 14:19:22 |
| @hexa:lossy.network | https://perlpunk.github.io/yaml-test-schema/schemas.html | 14:19:23 |
| @hexa:lossy.network | the other would be to try and revert to an older remarshal version that still emits yaml 1.1 | 14:19:46 |
| mattleon | In reply to @hexa:lossy.network
so one thought about remarshal is that we could patch hass to use yamlcore and thus interpret things as yaml 1.2 Have you already made progress on this? If not, I'd be happy to start on it | 23:29:55 |
| @hexa:lossy.network | no, we probably need a treewide fix | 23:30:26 |
| @hexa:lossy.network | so remarshal_0_17 and provide that as pkgs.formats.yaml_1_1, and the latest as yaml_1_2 | 23:30:57 |
| mattleon | Ah, that makes sense. Much more maintainable than trying to carry a large hass patch | 23:31:44 |
| 9 Nov 2024 |
 mfed3 | Trying to follow some of the comments up above about Declarative Nix scripts resolving "on" to state = "on" to state = true in yaml | 04:46:10 |
| mfed3 | Noticing all my automations with on/off checks are broken with that issue. Can someone direct me to which pull request fixes that or is this an ongoing issue that needs a fix still | 04:47:08 |
| @hexa:lossy.network | todo | 04:47:25 |
| mfed3 | Roger. Is there a current workaround I can substitute ? | 04:47:40 |
| @hexa:lossy.network | rollback | 04:47:45 |
| mfed3 | Thanks, appreciate it | 04:48:13 |
| mfed3 | My current version is 2024.10.4, should I just roll back to 2024.10.3? Or is there a version you recommend and know is good | 05:01:08 |
| @hexa:lossy.network | can you provide a config that obviously breaks? | 05:02:24 |
| @hexa:lossy.network | ideally one that breaks hard | 05:04:09 |
| mfed3 | Yea let me find the simplest one that doesn't any extra factors other than the on off true false things going on, some of my automations have a lot of conditions going on | 05:06:42 |
| mfed3 |
{ config, pkgs, ... }:
{
services.home-assistant.config.automation = [
{
alias = "Iron Auto Off";
trigger = {
platform = "state";
entity_id = "switch.sonoff_s31_lite_zb_switch_2";
to = "on";
for = "00:15:00";
};
action = {
service = "switch.turn_off";
entity_id = "switch.sonoff_s31_lite_zb_switch_2";
};
}
];
} | 05:09:46 |
