| 27 Aug 2021 |
 andi- | Even if the tooling isn't doing it how can you approve something that you have no clue about what it is used for? I wouldn't ever +1 a haskell-updates PR or some obscure data science package that I don't know anything about. | 14:08:52 |
 Mic92 | * It is just very expensive to figure out direct dependencies. | 14:08:59 |
| andi- | Saying the diff looks fine is one thing but judging if it should be merged.... | 14:09:24 |
| andi- | That is why I'd rather not have so many automated updated. | 14:09:39 |
| Mic92 | Yeah but even I know things about a package, there is still room for mistakes | 14:11:51 |
 adisbladis | And in this case it wasn't even automated | 14:13:25 |
| adisbladis | Fwiw I like the bot updates, but they require even more reviewing by maintainers | 14:16:29 |
| adisbladis | * Fwiw I like the bot updates, but they require even more reviewing/testing by maintainers | 14:16:47 |
| Mic92 | Maybe there could be a policy to require more reviews on a mass-rebuild? | 14:20:11 |
| Mic92 | If there was a cheap way of getting a graph of all packages, than nixpkgs-review could also exploit that and build the most popular packages in this list that are close to the changed packages. | 14:21:22 |
| Mic92 | I guess eventually nixpkgs-review needs it's own evaluator to improve speed and memory usage. | 14:21:58 |
| andi- | In most cases it would probably be sufficient to test building the small nixos test set. Now that isn't tiny but it also isn't huge and I'd expect anyone that works on that stuff to not be afraid of that. If you want to change something very low-level in the chain of dependencies you've to invest the time to test it and not just bump a hash and that a single package. | 14:24:05 |
| andi- | Staging or Hydra is not our playground to test experiments. | 14:24:43 |
| Mic92 | Maybe just communicate that in a friendly tone. | 14:28:17 |
| adisbladis | In reply to @mic92:nixos.dev
I guess eventually nixpkgs-review needs it's own evaluator to improve speed and memory usage. ;) | 14:44:22 |
| adisbladis | Mic92: I did quite some nice progress on https://github.com/nix-community/nix-eval-jobs | 14:45:11 |
| Mic92 | adis: I am interested on how you integrated this with jenkins | 14:45:21 |
| adisbladis | Remind me next week, I'll try to extract those bits. | 14:47:27 |
| adisbladis | It isn't doing anything too fancy | 14:47:37 |
| 28 Aug 2021 |
|  lvkm joined the room. | 06:29:49 |
| lvkm changed their display name from Istvan Ruzman to lvkm. | 06:30:25 |
 colemickens 🏳️🌈 | Whoa?! | 23:12:19 |
| colemickens 🏳️🌈 | This should be mentioned on the Discourse thread! This is what I was talking about. How exciting! | 23:12:47 |
| 30 Aug 2021 |
|  anubhavkini left the room. | 06:28:39 |
 Emil Karlson | systemd drops me to emergency shell nothing failed, how can I see why it's not going default? | 11:28:51 |
 Arian | What version? | 11:29:14 |
| Emil Karlson | whatever is in nixos-21.05 built today | 11:31:09 |
| Emil Karlson | the serial console seems to be broken | 11:31:19 |
| Emil Karlson | I think this was traditionally like 2 sessions open on same tty somehow fighting | 11:32:35 |
| Emil Karlson | wonder if linux-5.14 is just somehow too new? | 11:33:39 |
