| 18 Feb 2026 |
|  Erin [ DEPRECATED ] left the room. | 12:02:57 |
|  lucy changed their display name from evoea0 to lucy. | 17:01:00 |
|  -> @n:rab.lol changed their display name from niko ⚡️ to -> @n:rab.lol. | 23:28:24 |
| 19 Feb 2026 |
 ian-h-chamberlain | Aha! I've figured something out, this is not a Lix issue but indeed a cursed macOS shebang quirk. You can see it more obviously with one like this:
#!/usr/bin/env -S echo nixpkgs#foo nixpkgs#bar
The man pages for env claim (seemingly incorrectly) that # characters only matter at the beginning of a split argument, and can be escaped with \#: https://ss64.com/mac/env.html
This StackExchange question has some more details about a similar, related issue: https://unix.stackexchange.com/a/774145
and it seems macOS simply copied man pages from BSD even though the behavior is different :( so the best bet might be something like the "polyglot" script they suggest for JS there, if you can figure out how to adapt it for your use case.
| 04:45:06 |
|  niko joined the room. | 08:44:12 |
|  catsarecute set a profile picture. | 19:00:59 |
| -> @n:rab.lol left the room. | 22:53:14 |
| 20 Feb 2026 |
 toonn | Isn't a nix-shell shebang always just `#! /usr/bin/env nix-shell`? So no hashes to mess up and the second "shebang" line is parsed by Nix or Lix as the case may be, not by a shell as a shebang. | 10:13:30 |
 hexa | yep | 11:19:53 |
| ian-h-chamberlain | maybe but in this case they were trying to use nix shell with flake refs (which as I understand has had shebang parsing removed from Lix), not nix-shell | 19:31:11 |
| hexa | nix shell has no shebang support from what I know | 19:32:06 |
| toonn | And either way those flake refs would have to be passed on the second, not really a shebang, line no? | 19:34:17 |
| ian-h-chamberlain | yea. maybe more accurate description is "not added to Lix", cppnix has https://nix.dev/manual/nix/2.33/command-ref/new-cli/nix.html?highlight=shebang#shebang-interpreter | 19:35:22 |
| ian-h-chamberlain | but on non-darwin it seems you can fake a subset of this by using OS built-in shebangs and env -S parsing, which is the example I gave earlier | 19:39:00 |
| ian-h-chamberlain | * | 19:39:27 |
| toonn | But shebang parsing has nothing to do with Nix when it's the shell parsing it. | 19:42:37 |
| toonn | Nvm, that's kinda what you said anyway. | 19:45:31 |
| toonn | POSIX actually doesn't require more than one or two arguments (not sure whether env`s -S is an explicit exception or not) to be supported in shebangs. So it's probably a bad idea to try and do it all in a shebang anyway. The second shebang line is a pretty decent workaround IMO. | 19:50:35 |
|  shoe joined the room. | 21:56:32 |
| shoe | hi everyone | 21:57:28 |
| shoe | i'm brand new to nix/lix (but not to *nix) and decided to jump in headfirst after downgrading my macbook and discovering that brew doesn't provide binaries anymore | 21:58:24 |
| shoe | can't be spending 25 minutes compiling cmake | 21:58:46 |
 raitobezarius | brew doesn't provide binaries for aarch64-darwin anymore? | 21:59:06 |
| shoe | intel, on macos monterey | 21:59:23 |
| raitobezarius | fyi, nixpkgs cache will also start dropping x86_64-darwin soonish | 21:59:36 |
| raitobezarius | cc emily who knows better the timeline than me | 21:59:44 |
| shoe | long as its around for at least another year i'm good :p i was going to buy a new laptop but i decided to wait a little longer | 22:00:33 |
 leona | After 26.05 branchoff, I.e. the full 26.05 support cycle still supports it, but not 26.11 | 22:00:35 |
| raitobezarius | so this basically means that in Oct 2026, Intel macOS is no more in the cache | 22:00:56 |
| raitobezarius | for stable releases | 22:01:02 |
