| 16 Dec 2025 |
 niklaskorz | looks ok to me 😅 | 12:33:42 |
 WeetHet | Needing primaryUser is the thing I have an issue with | 12:34:00 |
| niklaskorz | ah, well you can also provide a store file there I guess | 12:34:20 |
| niklaskorz | as there's no secret information in known_hosts | 12:34:35 |
| niklaskorz | and about the identityfile, the linux-builder sets it up in /etc/nix/ | 12:34:53 |
| niklaskorz | ❯ ls -la /etc/nix
.rw------- 411 root 11 Jan 16:06 builder_ed25519
.rw-r--r-- 99 root 11 Jan 16:06 builder_ed25519.pub
| 12:35:10 |
| niklaskorz | you could also move the known_hosts there I guess | 12:35:15 |
| WeetHet | It has an advantage of being able to modify the authorized keys on the machine | 12:35:26 |
| WeetHet | I also can do it but then it's less automatic | 12:35:38 |
| WeetHet | Ehh, for now I'm gonna live with a primaryUser | 12:35:51 |
| WeetHet | I'm gonna come up with something else when it gets removed from nix-darwin | 12:36:06 |
 Randy Eckenrode | Knowing that I need to primarily build packages from source, I’m not going to package any more dependencies other than what I have. These are used by the toolchain and build with CMake. | 12:39:21 |
| Randy Eckenrode | For some reason though, statically building SwiftASN1 breaks building Swift Crypto. It makes absolutely no sense. | 12:39:57 |
| Randy Eckenrode | I really wish SwiftPM had a way to provide pre-built artifacts for these, but AFAIK it does not. The intent is you build everything from source. | 12:41:02 |
| Randy Eckenrode | I may go back to the drawing board and provide a binary output that actually does the build, so things that use SwiftPM can still build right. | 12:41:44 |
| Randy Eckenrode | I do want to hack up Swift PM to make it use our toolchain’s binaries for the experimental Swift Syntax binaries flag, but I still need to separate them out. | 12:42:27 |
| Randy Eckenrode | I’m currently grinding through getting SwiftPM building. I want the result to look like what it would if build by SwiftPM. The CMake files do not seem meant for production use …. | 12:43:25 |
|  vic set their display name to oeiuwq. | 15:10:07 |
| vic set a profile picture. | 21:32:12 |
| vic changed their display name from oeiuwq to vic. | 21:32:21 |
| 17 Dec 2025 |
| Randy Eckenrode | Does ${!outputInclude} not actually work? | 02:50:37 |
| Randy Eckenrode | $ echo $outputInclude
out
$ echo ${!outputInclude}
/Users/reckenrode/Developer/nixpkgs/outputs/out
$ echo ${!outputs[@]}
include lib out
$ echo ${outputs[include]}
/Users/reckenrode/Developer/nixpkgs/outputs/include
| 02:50:58 |
| Randy Eckenrode | dev is reserved for evil hacks. mkSwiftPackage adds a passthru.dev that vendors the patched source for use with SwiftPM. | 02:51:32 |
|  @azuwis:matrix.org left the room. | 07:06:15 |
 Ihar Hrachyshka | I posted a PR forcing darwin in electron* meta.platforms on Linux: https://github.com/NixOS/nixpkgs/pull/471643 so that hydra etc. knows it supports darwin. It's not very pretty (uses overrideAttrs) but I'm not ready to fix darwin source build of electron. any major reservations against this approach? | 18:08:32 |
