| 28 Sep 2025 |
 Taeer Bar-Yam | John Ericson You were talking about getting rid of the std::enable_shared_with_this for SourceAccessor. We should talk at some point about
- Why you want to get rid of it
- If you have any ideas about how to get rid of it
- How you feel about
boost::intrusive_ptr
Because it's currently driving me up a wall
| 21:46:01 |
 John Ericson | @shine:proqqul.net: I thought it was only needed for path values. If you do something else for path values, then we can like have a separate source accessor address spaces with indices | 21:52:58 |
| John Ericson | And index regular shared pointers | 21:53:16 |
| John Ericson | No need to do shared from this | 21:53:22 |
| Taeer Bar-Yam | you're imagining an array of shared pointers? | 21:54:16 |
| Taeer Bar-Yam | oh i see. and hold an index in the Value | 21:54:25 |
| Taeer Bar-Yam | hmmm | 21:54:26 |
| Taeer Bar-Yam | what's your reason for wanting to get rid of enable_shared_with_this? | 21:55:11 |
| Taeer Bar-Yam | I'm not opposed, I'm just trying to understand the constraints on the design space | 21:56:46 |
| John Ericson | In reply to @shine:proqqul.net
what's your reason for wanting to get rid of enable_shared_with_this? We don't always put the accessors behind shared pointers | 23:18:08 |
| John Ericson | And I just hate enable shared from this lol | 23:18:17 |
| John Ericson | It feels deeply non modular | 23:18:41 |
| John Ericson | (I wonder if explicit this could make it better, but that's another topic) | 23:19:48 |
| 29 Sep 2025 |
|  samppu joined the room. | 13:47:01 |
 niksnut | Taeer Bar-Yam: https://github.com/DeterminateSystems/nix-src/commit/4c028281a7a7ee657b0f180000dce3b19d485b9b | 17:13:09 |
| 30 Sep 2025 |
 Sergei Zimmerman (xokdvium) | Ok Robert Hensing (roberth), I've debugged why builtins.readDir behaves so weirdly. Wrote some unit tests documenting the currently broken behavior https://github.com/NixOS/nix/pull/14123 | 00:20:27 |
|  mjolnir banned  @mightyiam:matrix.org (divisive behavior.). | 19:22:19 |
|  daylin joined the room. | 20:53:10 |
| 1 Oct 2025 |
 fzakaria | Finally! | 03:29:21 |
| fzakaria | removed all shellcheck exclusions...
https://github.com/NixOS/nix/pull/14129 | 03:29:26 |
 lassulus | why does nix repl behave different when I run nix repl . vs running :lf in nix repl? | 11:32:12 |
| lassulus | like inputs is not available in the first case, is there a rationale for that I'm not understanding? | 11:32:39 |
| Sergei Zimmerman (xokdvium) | In reply to @lassulus:lassul.us
why does nix repl behave different when I run nix repl . vs running :lf in nix repl? I recall there was an issue or a discussion about this. Can’t find it now though | 11:37:44 |
| lassulus | ok, maybe I will try looking in the issue tracker than, than you :) | 11:38:42 |
 Mic92 | In reply to @lassulus:lassul.us
like inputs is not available in the first case, is there a rationale for that I'm not understanding? Also find this confusing | 12:54:33 |
| Mic92 | matthewcroughan: any chance this can be git bisected? https://github.com/NixOS/nix/issues/14130 | 15:26:01 |
|  @aloisw:julia0815.de left the room. | 15:28:06 |
 matthewcroughan | It's not reproducible, it seems to happen more often on bcachefs | 15:30:04 |
| matthewcroughan | and I was on a way earlier version of nix | 15:30:21 |
|  magic_rb left the room. | 18:23:10 |
