| 16 Nov 2025 |
 hexa | but effectively windows machines maintained by the infra team | 17:55:52 |
 7c6f434c | To add Tier-4 support RFC is not needed, to add Tier-3 RFC is kind of not enough | 17:56:07 |
| hexa | and the status quo is that we don't even have people to deal with the darwin machines as would be required | 17:56:12 |
 Corngood | I was thinking of running my own hydra/builder, but I wanted to check in to make sure there's no existing plans for windows builds. It sounds like there isn't. | 17:56:43 |
 dramforever | does nix builds even work on windows yet | 17:57:13 |
| Corngood | we can cross build a system image tarball (including nix) from linux, and then boot it and use it to build itself natively on cygwin. I'm not sure about mingw/msvc. I don't think they are as far along | 17:58:32 |
| hexa | Janne: are you offering? 🙂 | 17:58:45 |
| dramforever | oh, yeah i was thinking more about msvc | 17:58:56 |
| dramforever | that one probably doesn't work | 17:59:02 |
| dramforever | cygwin probably does but i'm going to worry about the sandboxing | 17:59:32 |
 Janne | Macos is not my main area of expertise. But tbh, windows would probably be way easier since there is software like trmm that makes fleet management easy while not costing a billion euros and locking you into their cloud | 17:59:42 |
| Janne | I could take over Linux if you take over all of the macos infra though 🤔 | 18:00:19 |
| Corngood | I've been wondering about sandboxing in general on windows. I'm not sure if anyone has made any progress on that. | 18:00:21 |
| Corngood | At least you can have separate cygwin installs that have their own process namespaces and stuff. Certainly nothing secure though. | 18:01:01 |
| Janne | I bet @simon.hauser:helsinki-systems.de would love porting the PSI thingies to the windows equivalent | 18:02:42 |
 eveeifyeve | Do you want me to look into that? | 19:01:11 |
| 17 Nov 2025 |
 John Ericson | I am catching up on this conversation now Corngood | 17:43:39 |
| John Ericson | (and eveeifyeve ) | 17:43:51 |
| John Ericson | K900: you can weigh in of course, but we've had a lot of conversations with @emilylazy and Alyssa Ross about how Cygwin is the last disruptive way to try to get Windows support going | 17:45:42 |
| John Ericson | * K900: you can weigh in of course, but we've had a lot of conversations with @emilylazy and Alyssa Ross about how Cygwin is the last disruptive way to try to get Windows support going in Nixpkgs | 17:45:49 |
| John Ericson | (least disruptive to Nixpkgs) | 17:45:52 |
| John Ericson | I would like for MinGW and LLVM in MSVC-ABI-compat mode to obviate the need for Cygwin longer term for many tasks, but I don't want to block Windows support on "first we have to rewrite stdenv and all the bash" | 17:46:48 |
| John Ericson | Indeed, after first I was excited contemplating switching from bash, etc., but Emily and Alyssa convinced me that was irresponsible, and we have to take small steps | 17:47:27 |
| John Ericson | So the conversation we have there should at least acknowledge the Corngood has received lots of feedback so far that his approach is a good first incremental step, and he's come here to ask about tarballs now because he has made amazing progress far faster than I, at least, I thought was going to happen on the basis of basically 1 person's effort | 17:48:26 |
| John Ericson | * So the conversation we have there should at least acknowledge the Corngood has received lots of feedback so far that his approach is a good first incremental step, and he's come here to ask about tarballs now because he has made amazing progress far faster than I, at least, thought was going to happen on the basis of basically 1 person's effort | 17:49:28 |
| John Ericson | * So for the conversation we have there should at least acknowledge the Corngood has received lots of feedback so far that his approach is a good first incremental step, and he's come here to ask about tarballs now because he has made amazing progress far faster than I, at least, thought was going to happen on the basis of basically 1 person's effort | 17:49:46 |
| John Ericson | * So for the conversation that we have, we should at least acknowledge the Corngood has received lots of feedback so far that his approach is a good first incremental step, and he's come here to ask about tarballs now because he has made amazing progress far faster than I, at least, thought was going to happen on the basis of basically 1 person's effort | 17:50:00 |
| John Ericson | * Indeed, after first I was excited contemplating switching from bash, etc., but Emily and Alyssa convinced me that doing that right away was irresponsible, and we have to take small steps | 17:57:43 |
| Corngood | I also started a discussion on nix-community
https://github.com/orgs/nix-community/discussions/2025
What I'm working on seems quite similar to nix-darwin or nixbsd. Maybe part of it can live there? | 18:07:55 |
| John Ericson | Corngood: this part is a nixpkgs question, but with modular services I would love to see your stuff go upstream | 18:11:04 |
