| 27 Jun 2025 |
 kdn | yeah, I'm wondering why is it waiting for a network connection at all and not get denied immediately | 14:30:29 |
| kdn | those 120/180s timeouts all have runtime.netpoll+210 reference in the stack, not sure how to identify a specific function exactly in this case | 14:31:09 |
| kdn | could there be something in Nix configuratio that could affect this behavior? | 14:37:15 |
| kdn | * could there be something in Nix configuration that could affect this behavior? | 14:37:21 |
| kdn | maybe firewall rules (no idea if those apply at all in case of Nix builders)? | 14:37:36 |
| kdn | FYI: I have asked about it on Gophers, maybe somebody will help there https://gophers.slack.com/archives/C0VP8EF3R/p1751039539399319 | 16:06:42 |
| kdn | funny enough, my customized install-iso also builds Netbird pretty quickly (under a minute) | 17:35:21 |
| kdn | a Hetzner machine with quite similar config builds slowly already | 17:41:08 |
 K900 | Well feels like you have a starting point to bisect from | 17:41:50 |
| kdn | actually looking at btop seems like the hetzner build is CPU bould | 17:42:08 |
| kdn | could it be something with ZFS slowing things down? | 17:43:14 |
| kdn | could nix builder have some problem with ~500 bind mounts due to persistence setup? | 17:57:38 |
| kdn | nevermind, I don't have persistence on my laptop at all | 17:57:48 |
| kdn | * nevermind, I don't have persistence on my laptop at all, I have it on my rpi4 that builds the stuff quickly enough too | 17:58:34 |
| 1 Jul 2025 |
|  Puna joined the room. | 09:08:37 |
| Puna | don't really know much about Go and its different compilers, so i figured I'd ask here: how many things would explode if i replaced go with gccgo in Nixpkgs? is Go stuff generally compatible with the latter, maybe only with missing Nix glue here and there for it to be a direct go = gccgo job?
i have a platform that go doesn't really support anymore since 1.9 (or at least not in a manner where affordable hardware can use it), and upstream has no interest in fixing that support, so I'm planning to mark go as unsupported on it and am looking for alternatives. this replacement would only affect that particular platform. | 09:21:44 |
| K900 | gccgo is not practically usable AFAIK | 09:22:18 |
| Puna | in that the project itself is borked / not up-to-spec and incompatible with out-in-the-wild Go code, or is our packaging of it bad? | 09:23:59 |
| K900 | I think upstream | 09:24:31 |
| K900 | The GCC 12 and 13 releases include a complete implementation of the Go 1.18 standard library. However, GCC does not yet include support for generics.
| 09:25:14 |
| K900 | https://go.dev/doc/install/gccgo | 09:25:17 |
| K900 | So yeah it's stuck on an ancient Go version | 09:25:21 |
| Puna | hmm… and i assume usage of generics is common enough for that to matter?
any important packages i'd miss out on by just marking go unsupported and accepting that Go is just a no-go (lol)? | 09:30:42 |
| K900 | Honestly, lots | 09:32:48 |
| K900 | What platform is that even? | 09:32:51 |
| Puna | big-endian ppc64 | 09:33:00 |
| Puna | support for POWER < 8 got dropped, and those systems are the only ones that are affordable, and where using big-endian really made sense | 09:34:11 |
| K900 | Yeeeeeeeeeeah | 09:34:51 |
 Diamond (it/she) | on a basic system i dont think theres much youd miss out, more than that it depends on what youll use it for | 21:00:09 |
| Diamond (it/she) | like a lot of web stuff are in Go for example | 21:00:17 |
