| 9 Jan 2026 |
 Katalin 🔪 | mhm | 16:19:50 |
 Ivy | additionally this does nothing https://github.com/nix-community/home-manager/blob/0e4217b2c4827e71e2e612accccb01981c16afda/modules/services/gpg-agent.nix#L451-L453 | 16:21:03 |
| Ivy | as the names are far not what gpg actually wants | 16:21:20 |
| Ivy | nor does it know how to get them | 16:21:27 |
| Ivy | the only way to get them is through launch_activate_socket | 16:21:42 |
| Ivy | they could be used as the names but then there would have to be major translation to the real names | 16:22:20 |
| Ivy | which have to be "ssh", "extra", "browser" and always finally "std" | 16:22:37 |
| Ivy | this commit which added that literally seems to be untested https://github.com/nix-community/home-manager/commit/ef506124579ff6280a43a9596bb2a5049872bf8e as it will not work | 16:24:04 |
| Ivy | additionally, patching this is hard as it shouldnt actually be used the gpgConf we need to wrap gpg-agent | 16:26:19 |
| Ivy | i personally have it working but it required a lot of changes | 16:27:49 |
| Ivy | including this https://github.com/auscyber/dotfiles/blob/e69c5ae454167f21dbaca7eace8e50e69d5d3454/overlays/literal.nix#L33C1-L39C4 https://github.com/auscyber/dotfiles/blob/master/packages/gpg/default.nix | 16:28:35 |
| Ivy | * additionally, patching this is hard as it shouldnt actually be used the gpgPkg we need to wrap gpg-agent | 16:32:41 |
 emily | you don't have to use launchd socket activation though? | 16:50:08 |
| emily | you can have GnuPG manage the sockets itself IIRC | 16:50:11 |
| emily | (but yeah, socket activation is nicer. I wrote a version of that C program in Rust long ago) | 16:50:28 |
| 10 Jan 2026 |
 Noah | is there an easy way to add applications to startup items on macos? | 11:37:45 |
| 11 Jan 2026 |
|  ivan joined the room. | 01:56:32 |
|  ghpzin changed their display name from ghpzin (moved to @ghpzin:envs.net) to ghpzin. | 15:04:53 |
|  @ghpzin:envs.net left the room. | 16:18:24 |
| 12 Jan 2026 |
|  James Cook joined the room. | 01:15:01 |
 Ihar Hrachyshka | Randy Eckenrode: llvmPackages_20.compiler-rt is broken on 26 for x86_64-darwin.
We'll probably need this change: https://github.com/apple-oss-distributions/ld64/commit/1a4389663d65d6630e4b3e31ace2a86b6183b452#diff-8c97f4c6bb5cfd9f4a6f2d0cb4e7fc2caaf08692d0a11ec16ad28dff725435ecR4888
Hydra: https://cache.nixos.org/log/gj40yabysxzjap3fgkw4lq0z0qyi28v5-compiler-rt-libc-20.1.8.drv
I am running a build to confirm the hunk fixes it but it will take some time.
Are there plans to bump ld64 or backport the hunk?
| 01:58:46 |
| Ivy | you do tho, because its supervised mode in home-manager, which requires it to be given the sockets its going to use | 02:42:30 |
| Ivy | the logs ask for fds and it crashes out without the fds as supervised mode is explicitly "i am not going to make the fds, you do that some way and then give me them" | 02:43:56 |
 samasaur | can you not simply not use the existing home-manager module? | 03:05:51 |
| emily | HM modules being kinda broken is nothing new :) | 03:10:40 |
| samasaur | either to let GPG start the agent on demand (which is what I do) or to write your own module that doesn't use --supervised (if you need to configure it) | 03:10:45 |
| emily | only x86? | 03:11:11 |
| samasaur | not to downplay the fact that the blessed path is borked | 03:11:15 |
| Ivy | im just making a pr to fix the current behaviour | 03:23:44 |
| Ivy | how do placeholders work? | 03:52:34 |
