| 22 Oct 2025 |
 hexa | I'm going to sacrifice nix.dev ig | 15:52:50 |
 Jeremy Fleischman (jfly) | Have we considered moving websites to be under www.<whatever>? | 17:15:32 |
| hexa | something for marketing to decide | 17:17:40 |
| hexa | this breaks all kinds of search engine entries | 17:17:48 |
| hexa | aiui | 17:17:54 |
| hexa | but right now they're investigating alternatives to netlify where we can just point to ip addresses | 17:18:15 |
 Vladimír Čunát | It does? I'd assume that you still keep the non-www variant as a slow-ish http redirect. | 17:19:15 |
| hexa | yeah, but the url changes, you get duplicated content in the cache | 17:19:39 |
| hexa | vague memories here | 17:19:42 |
| Vladimír Čunát | Yes, you want the URL to change. | 17:19:56 |
| Vladimír Čunát | (and avoid using the non-www variant) | 17:20:10 |
| hexa | yes, but with the understanding what that means for our web presence | 17:20:18 |
| Vladimír Čunát | * (and avoid using the non-www variant in links, etc.) | 17:20:21 |
| hexa | which is not infra's job | 17:20:22 |
| hexa | feel free to suggest that to marketing though | 17:20:55 |
| Vladimír Čunát | Is there a problem with the current setup, though? | 17:23:33 |
| Vladimír Čunát | I haven't heard it being slow - or not much of any feedback really 😅 | 17:24:02 |
| hexa | just the ALIAS record imo | 17:24:19 |
| hexa | which means they resolve it from various pops around the world and that can fail | 17:24:32 |
| Vladimír Čunát | I mean now, without the ALIAS. Or is it back already? | 17:24:43 |
| hexa | in this case it was their pop in milano that couldn't resolve nixos-homepage.netlify.com apparently | 17:24:47 |
| hexa | good question, I don't have a full picture of what we terminate of the nixos.org vhost | 17:25:24 |
| hexa | * good question, I don't have a full picture of what we terminate over the nixos.org vhost beyond the website | 17:25:38 |
| hexa | and yeah, right now its some aws loadbalancer, which is likely anycasted | 17:26:02 |
| hexa | so not a bad solution | 17:26:14 |
| Vladimír Čunát | I'd assume that the performance (geo* stuff) might be worse than with ALIAS/CNAME, but no idea how much and whether we'd even care for the difference. | 17:27:21 |
| Vladimír Čunát | (for web) | 17:27:43 |
| 23 Oct 2025 |
| hexa | vcunat: fyi https://github.com/NixOS/hydra/issues/1535#issuecomment-3438017869 | 16:43:07 |
|  Mr Argoz joined the room. | 21:03:55 |
| 24 Oct 2025 |
|  KnownRabbit joined the room. | 03:45:07 |
