| 3 Dec 2025 |
 Mic92 | In reply to @roberthensing:matrix.org
so putting meta in derivations without impacting hashing was actually kinda low hanging fruit
https://github.com/NixOS/nix/pull/14686 I think i want better roadmaps that are actually followed up on for new experimental features. Nix starts to look like haskell with it's preamble of language extensions | 06:43:34 |
|  The Photonsphere joined the room. | 12:58:37 |
|  @wplanck:mozilla.org left the room. | 13:14:11 |
 matthewcroughan | Yeah | 14:33:46 |
 connor (burnt/out) (UTC-8) | If only we could enable them per-file ;) | 16:08:56 |
 Taeer Bar-Yam | How can I be getting the error message
error: output '/nix/store/6s177q59d74xhxkj8p54bga156d9g84b-test' is not allowed to refer to the following paths:
/nix/store/6s177q59d74xhxkj8p54bga156d9g84b-test
Without hitting the checkOutputs() function at all? That text doesn't appear anywhere else in the code!
Things I did:
- adding a debug print statement
std::cout << "HERE\n"; (didn't get printed)
- running with a breakpoint in gdb (didn't get hit)
- commenting out the relevant
throw (threw the error anyways)
- running on a fresh derivation in case i was hitting some cache (same results)
- putting obvious errors in
checkOutputs() (compile error. so I am recompiling this function)
- editing
src/nix/main.cc in obvious ways to make sure I was running the binary that was getting built (obvious changes showed up, so I'm successfully building and running the right binary)
| 17:46:50 |
| Taeer Bar-Yam | * How can I be getting the error message
error: output '/nix/store/6s177q59d74xhxkj8p54bga156d9g84b-test' is not allowed to refer to the following paths:
/nix/store/6s177q59d74xhxkj8p54bga156d9g84b-test
Without hitting the checkOutputs() function at all? That text doesn't appear anywhere else in the code!
Things I did:
- adding a debug print statement
std::cout << "HERE\n"; (didn't get printed)
- running with a breakpoint in gdb (didn't get hit)
- commenting out the relevant
throw (threw the error anyways)
- changing the text of the relevant
throw (printed the old text)
- running on a fresh derivation in case i was hitting some cache (same results)
- putting obvious errors in
checkOutputs() (compile error. so I am recompiling this function)
- editing
src/nix/main.cc in obvious ways to make sure I was running the binary that was getting built (obvious changes showed up, so I'm successfully building and running the right binary)
| 17:48:03 |
 Sergei Zimmerman (xokdvium) | Dumb question. Did you funnel this to the newly built nix's local store or a separate daemon? | 17:48:15 |
| Taeer Bar-Yam | OH is that code running on the daemon?? of course! | 17:48:37 |
| Taeer Bar-Yam | i knew i was going to kick myself when i figured it out | 17:48:42 |
| Taeer Bar-Yam | or when someone told me, rather | 17:48:46 |
| Taeer Bar-Yam | i've been spending so much time in the parsing code i forgot about the daemon | 17:49:07 |
| Taeer Bar-Yam | how do i force it to use local nix for the daemon again? | 17:49:31 |
| Sergei Zimmerman (xokdvium) | Depends. You can just use a chroot store with --store /path/to/chroot/store | 17:50:39 |
| Taeer Bar-Yam | thank you!! | 17:51:25 |
 Robert Hensing (roberth) | Sure. This one is a small change, and can be stabilized to become a normal requiredSystemFeatures feature. I would suggest to do that after 1 year. SBOM folks will use this stuff, so unlike say the TOML timestamps we'll get some testing in that time. | 22:05:22 |
| Robert Hensing (roberth) | As for requiredSystemFeatures becoming a Haskell-esque preamble, that's mitigated by having Nixpkgs do it, and you can bundle a bunch of them up into a single flag like Haskell2010 did. We're nowhere near that though. ca-derivations will remain a choice anyway (input addressing is useful), so structured attrs is the only other such flag so far. | 22:08:02 |
| Robert Hensing (roberth) | structuredAttrs and the discard references thing should have been requiredSystemFeatures, but that's not useful anymore at this point | 22:10:01 |
| 4 Dec 2025 |
|  Ido Samuelson joined the room. | 01:47:49 |
|  isabel changed their profile picture. | 16:41:31 |
|  Tanja (she/her) - ☎️ 4201 changed their display name from Tanja (she/her) to Tanja (she/her) - ☎️ 4201. | 18:08:28 |
| 5 Dec 2025 |
|  SamaelMatus joined the room. | 07:38:50 |
| 6 Dec 2025 |
| Mic92 | Sergei Zimmerman (xokdvium): https://github.com/NixOS/nix/pull/14696#issuecomment-3619665616 not sure yet if it's related to this change, but I just got an error that it couldn't find a "uid" in my dotfiles for hercules-ci-effects, unfortunally I didn't copy the error and now it no longer reproduces. | 06:57:26 |
| Mic92 | Let me also make sure that I rebase on latest master | 06:58:00 |
| Sergei Zimmerman (xokdvium) | In reply to @joerg:thalheim.io
Sergei Zimmerman (xokdvium): https://github.com/NixOS/nix/pull/14696#issuecomment-3619665616 not sure yet if it's related to this change, but I just got an error that it couldn't find a "uid" in my dotfiles for hercules-ci-effects, unfortunally I didn't copy the error and now it no longer reproduces. That only affects the tarball cache though. | 06:58:16 |
| Sergei Zimmerman (xokdvium) | And your dotfiles is just a regular repo and doesn’t get opened with packfilesOnly = true | 06:59:09 |
| Mic92 | Sergei Zimmerman (xokdvium): that was reported for the hercules-ci-effects repo specifically, I had this patch applied manually on top: https://github.com/Mic92/nix-1/commit/2c0f976026c0c0acb21e5bae4b205eacd94a8a6c | 07:00:00 |
| Mic92 | hercules-ci-effects.url = "github:hercules-ci/hercules-ci-effects";
| 07:00:20 |
| Mic92 | and yes it's the tarball cache in this case | 07:00:31 |
| Sergei Zimmerman (xokdvium) | Weird, I’m dogfooding that for some time and didn’t see any issues. Maybe the tarball cache gave out for unrelated reasons? | 07:03:26 |
