| 16 Nov 2025 |
Sergei Zimmerman (xokdvium) | Mic92 took a stab at syncing with detsys master and ripping out things at https://github.com/NixOS/experimental-nix-installer/pull/51. | 18:59:04 |
| 17 Nov 2025 |
Taeer Bar-Yam | auto arena = es.mem.exprs.alloc;
decltype(attrs) foo = {std::move(attrs), arena};
attrs = {};
new (&attrs) decltype(attrs){std::move(foo), arena};
| 19:55:39 |
| SCR4MBL3 joined the room. | 21:07:22 |
John Ericson | Sergei Zimmerman (xokdvium): I forgot that I do need your file descriptor source accessor work also because that is how I want the mini recursive nix to add to store | 23:24:52 |
John Ericson | (send a file descriptor over the socket, add whatever that refers to to the store) | 23:25:20 |
John Ericson | if you want to open a draft PR, that would be great! | 23:59:52 |
| 18 Nov 2025 |
Sergei Zimmerman (xokdvium) | I'm a bit stumped as to how best to handle FDs that are not for directories. There isn't a great portable way to reopen an FD on systems other than linux... | 00:01:33 |
John Ericson | can you open /dev/fd/N? | 00:02:01 |
Sergei Zimmerman (xokdvium) | Only on linux | 00:02:20 |
Sergei Zimmerman (xokdvium) | As in you need procfs, not /dev/fd | 00:02:31 |
John Ericson | I think /dev/fd might work? | 00:02:51 |
John Ericson | don't have a mac at hand, however | 00:03:13 |
Sergei Zimmerman (xokdvium) | The problem with sending fds with source accessors would really be with handling things at the top-level. No way to express a single symlink or not very easy to have a single file as well | 00:16:32 |
Sergei Zimmerman (xokdvium) | Dunno what your idea was for the varlink interface. Maybe passing an FD for a directory and having a subdir in there as the top-level NAR member would be better? Otherwise you can't have all the type of NARs without also making them part of the data model | 00:21:22 |
| @brainworms2002:matrix.org removed their profile picture. | 06:26:24 |
| @brainworms2002:matrix.org removed their display name Sunny || ☀️🪱. | 06:26:27 |
| @brainworms2002:matrix.org left the room. | 06:26:31 |
| 19 Nov 2025 |
Taeer Bar-Yam | in nix_api_value.h, the EvalState argument of nix_init_path_string is not documented in the comment above. Is this a mistake? Are we intending to document every argument, and if so is that something we can check in CI?
https://github.com/NixOS/nix/blob/f4989b118bfba31f2e13504f2b950d07bf9539bb/src/libexpr-c/nix_api_value.h#L525-L532
/** @brief Set a path
* @ingroup value_create
* @param[out] context Optional, stores error information
* @param[out] value Nix value to modify
* @param[in] str the path string, copied
* @return error code, NIX_OK on success.
*/
nix_err nix_init_path_string(nix_c_context * context, EvalState * s, nix_value * value, const char * str);
| 01:40:37 |
| 7karni joined the room. | 10:52:19 |
| Leo joined the room. | 12:40:18 |
| 19 May 2021 |
| @eelco:nixos.org changed the history visibility to "world_readable" from "shared". | 15:40:14 |
| @eelco:nixos.org changed the room name to "Nix Hackers" from "Nix Development". | 15:45:04 |
| @eelco:nixos.org changed the room topic to "For people hacking on the Nix package manager itself" from "Development of the Nix package manager". | 15:45:04 |
| cransom joined the room. | 15:45:08 |
| sumner joined the room. | 15:45:19 |
| danielle joined the room. | 15:47:55 |
| aaron joined the room. | 15:54:00 |
| cransom changed their display name from casey © to cransom. | 15:56:27 |
| @eelco:nixos.org invited grahamc. | 16:02:36 |
| grahamc joined the room. | 16:02:51 |