| 19 Jan 2026 |
 just1602 | Are you running the latest commit, because I think the skip were added already? | 20:16:55 |
 raitobezarius | it should be a skipif I think yeah | 20:21:08 |
 Arian | Is there some way to have a pkgs.closureInfo that adds a drv as if you had done a substitution? Now if you ad a drv path it adds the entire build-time closure | 20:57:24 |
| Arian | Basically; can I in nix create a disk image that has a nix store that looks like it did a nix-instantiate && nix-store--realize where it substituted everything | 20:57:53 |
| raitobezarius | In reply to @arianvp:matrix.org
Is there some way to have a pkgs.closureInfo that adds a drv as if you had done a substitution? Now if you ad a drv path it adds the entire build-time closure Why not unsafe discard? | 22:32:25 |
| 20 Jan 2026 |
 DolceTriade | Hello, I'm hitting an assert in Lix and I'm not entirely sure how to debug further...
nix-daemon: lix/libstore/build/derivation-goal.cc:326: kj::Promise<Result<Goal::WorkResult>> nix::DerivationGoal::outputsSubstitutionTried(): Assertion `false' failed.
I'm running nix develop --impure on our codebase and I see output like this:
fetching git input 'git+https://github.com/zhaofengli/nix-base32.git'
fetching git input 'git+https://github.com/zhaofengli/nix-base32.git'
copying path '/nix/store/x3kqgb3wac95i6l74mi5drvrp7l58bra-nodejs-20.19.1-source' from 'https://cache.nixos.org'...
copying path '/nix/store/1z0543imx3gwb9l4wa8sbzhpxjxxh3s3-git-2.47.2-debug' from 'https://cache.nixos.org'...
error: Nix daemon disconnected unexpectedly (maybe it crashed?)
I've tried removing any custom binary caches and only have cache.nixos.org
| 18:21:34 |
| DolceTriade | * Hello, I'm hitting an assert in Lix and I'm not entirely sure how to debug further...
nix-daemon: lix/libstore/build/derivation-goal.cc:326: kj::Promise<Result<Goal::WorkResult>> nix::DerivationGoal::outputsSubstitutionTried(): Assertion `false' failed.
I'm running nix develop --impure on our codebase and I see output like this:
fetching git input 'git+https://github.com/zhaofengli/nix-base32.git'
fetching git input 'git+https://github.com/zhaofengli/nix-base32.git'
copying path '/nix/store/x3kqgb3wac95i6l74mi5drvrp7l58bra-nodejs-20.19.1-source' from 'https://cache.nixos.org'...
copying path '/nix/store/1z0543imx3gwb9l4wa8sbzhpxjxxh3s3-git-2.47.2-debug' from 'https://cache.nixos.org'...
error: Nix daemon disconnected unexpectedly (maybe it crashed?)
I've tried removing any custom binary caches and only have cache.nixos.org
nix --version
nix (Lix, like Nix) 2.94.0
| 18:22:48 |
| raitobezarius | do you have a flake reference to reproduce it? | 18:27:37 |
| DolceTriade | No, I haven't tried to create a minimal repro yet because I didn't understand why this is happening so if I were to try, it would be a random guess, so i was hoping for more insight on where to direct my attention, but if that's the first step, I can try. | 18:29:31 |
 K900 | Core dump would be nice | 18:29:56 |
| K900 | Or a backtrace at least | 18:30:01 |
| DolceTriade | Yeah, I was surprised I didn't see a coredump in systemd | 18:30:55 |
| DolceTriade | I guess since the nix-daemon is crashing, I probably have to edit the systemd file to set the ulimits to allow for core generation? | 18:41:07 |
| DolceTriade | I guess I managed to get it to load by commenting out parts of my flake and adding them back line by line until it built 🤷 | 19:13:59 |
| DolceTriade | The parts that were failing were part of a custom binary cache | 19:14:29 |
| raitobezarius | so a misbehaving binary cache perhaps? | 19:18:36 |
| raitobezarius | still, we shouldn't crash because of that | 19:18:52 |
| DolceTriade | Yes, that seems likely. | 19:19:55 |
| raitobezarius | if we can get access to it and have a reproducer in a bug issue, we can look into that | 19:20:41 |
| raitobezarius | we would debug it via gdb and look into the state machine | 19:20:49 |
| DolceTriade | We're using attic as our binary cache and I guess we discovered that one of binary cache uploads failed. Dunno how to reproduce this. I can try to build a reproducer. | 19:22:20 |
| DolceTriade | * We're using attic as our binary cache and I guess we discovered that one of binary cache uploads failed. Dunno how to reproduce this. I can try to build a reproducer that simulates a similar situation | 19:22:36 |
| DolceTriade | * We're using attic as our binary cache and I guess we discovered that one of binary cache uploads failed. Dunno how to reproduce this. I can try to build a reproducer that simulates a similar situation. We have to pin to specific random versions for legacy reasons. :( | 19:22:54 |
| 21 Jan 2026 |
 Winter | running into the same silly “writeTextFiles from my system build are taking longer than i expect to build,” where “longer than i expect” = like a second per drv… 🤨
can reliably ish repro at least… but still very weird
| 02:06:58 |
| Winter | * running into the same silly “writeTextFiles from my system build are taking longer than i expect to build” again, where “longer than i expect” = like a second per drv… 🤨
can reliably ish repro at least… but still very weird
| 02:07:14 |
| Winter | am i going insane, or are NixOS system closures Just Like That | 02:07:25 |
| raitobezarius | In reply to @winter:catgirl.cloud
running into the same silly “writeTextFiles from my system build are taking longer than i expect to build” again, where “longer than i expect” = like a second per drv… 🤨
can reliably ish repro at least… but still very weird
Sandbox setup time grows with respect to the number of derivations that you are building overall | 06:32:46 |
| raitobezarius | As you have more derivations, forking takes more time even if the derivation is trivial to build | 06:33:20 |
| raitobezarius | It's on our radar to fix this | 06:33:28 |
| raitobezarius | Because it's impossible to build 1M of trivial drvs due to that | 06:33:38 |
