14 Oct 2024 |
Ross A. Baker | In reply to @ross:rossabaker.com Pinning nixpkgs for both nix-darwin and lix-module works. I'm trying to bisect nixpkgs, but my endpoint security makes that an awful slog. I've traced the problem to this commit with a minimal flake to reproduce it. It's probably a nixpkgs issue rather than Lix? I am baffled why an IDE upgrade would cause manpage rendering to fail. | 14:34:00 |
qbit | In reply to @ross:rossabaker.com I've traced the problem to this commit with a minimal flake to reproduce it. It's probably a nixpkgs issue rather than Lix? I am baffled why an IDE upgrade would cause manpage rendering to fail. wow, maybe bruno is doing some .. test? or .. something? | 15:40:22 |
Ross A. Baker | I couldn't find any bruno references within nixpkgs that suggests how it would affect manpages. But a colleague can reproduce it. | 15:43:21 |
K900 | nix why-depends it? | 16:28:57 |
Ross A. Baker | I'm not sure what the second arg should be, but nix why-depends .#darwinConfigurations.simple.system nixpkgs#bruno comes up empty. Also, the only bruno in my store was instantiated by that command. | 17:35:54 |
K900 | Try addinf --derivation | 17:37:28 |
Lily Foster | In reply to @ross:rossabaker.com I'm not sure what the second arg should be, but nix why-depends .#darwinConfigurations.simple.system nixpkgs#bruno comes up empty. Also, the only bruno in my store was instantiated by that command. .#darwinConfigurations.simple.pkgs.bruno would better ensure comparison against the same nixpkgs instance | 17:38:52 |
Ross A. Baker | Ah, thanks. That's the incantation I thought I needed but didn't know... | 17:39:34 |
Lily Foster | but it sounds like you still need --derivation if i'm understanding (the admittedly amount i read of) scrollback correctly | 17:40:02 |
Lily Foster | * but it sounds like you still need --derivation if i'm understanding (the admittedly small amount i read of) scrollback correctly | 17:40:11 |
Ross A. Baker | Same result:
$ nix why-depends --derivation .#darwinConfigurations.simple.system .#darwinConfigurations.simple.pkgs.bruno
'git+file:///Users/RABaker/src/lix-build-failure#darwinConfigurations.simple.system' does not depend on 'git+file:///Users/RABaker/src/lix-build-failure#darwinConfigurations.simple.pkgs.bruno'
and bruno is only instantiated by this command, and can be nix-store deleted afterward.
| 17:44:29 |
Ross A. Baker | Would this see build-time dependencies? Even then, it should have been instantiated to try to build Lix, I'd think? | 17:45:40 |
raitobezarius | did someone else reproduce already? | 17:47:52 |
Ross A. Baker | A colleague did. And qbit had a similar error, but seemed to get past it moving something to HEAD. HEAD of what, I'm not sure. Getting on HEAD of lix and lix-modules last night didn't help me. | 17:49:42 |
Ross A. Baker | I wonder if I can strip this down to a derivation that tries to render a manpage. That's where the Lix build is failing, after successfully compiling most of the C++. | 17:51:09 |
Ross A. Baker | I think it's failing while invoking lowdown in render-manpage.sh. | 17:54:23 |
qbit | In reply to @ross:rossabaker.com A colleague did. And qbit had a similar error, but seemed to get past it moving something to HEAD. HEAD of what, I'm not sure. Getting on HEAD of lix and lix-modules last night didn't help me. I was able to build HEAD of lix yesterday - the lix in nixpkgs wouldn't build | 17:54:45 |
qbit | or.. what ever day it was <.< | 17:55:06 |
Ross A. Baker | Can you see which commit of nixpkgs? I'm trying lix main again from my repro flake and the bruno nixpkgs commit that broke lix-2.91.0. | 17:59:00 |
Ross A. Baker | Lix main fails the same way with this flake. | 18:05:43 |
Charles | In reply to @ross:rossabaker.com I'm not sure what the second arg should be, but nix why-depends .#darwinConfigurations.simple.system nixpkgs#bruno comes up empty. Also, the only bruno in my store was instantiated by that command. pass --inputs-from . to make sure the nixpkgs references the same version in your flake instead of whatever your system/user's flake registry happens to be | 18:51:17 |
Molly Miller | does anyone know if it's possible to use nix-eval-jobs (of either the cppnix or lix flavours) with hydra? i had a conversation earlier this evening which got me wondering (and i don't have a hydra instance sitting in front of me right now this instant which i can experiment with) | 20:03:59 |
K900 | The Lix fork of Hydra uses nix-eval-jobs | 20:04:29 |
Molly Miller | TIL there's a lix fork of hydra | 20:23:33 |
Molly Miller | thanks for the pointer! | 20:23:40 |
@jade_:matrix.org | In reply to @ross:rossabaker.com Lix main fails the same way with this flake. scratching my head, but maybe the thing that is separating peoples' experiences is whether they have sandbox enabled in nix.conf? | 20:25:53 |
@jade_:matrix.org | plus or minus OS version? | 20:26:00 |
@jade_:matrix.org | and then the nested sandbox possibly enabled by lowdown is what is bork? | 20:26:10 |
Ross A. Baker | I /just/ finished minimizing it to a lowdown invocation. | 20:28:24 |
@jade_:matrix.org | https://github.com/kristapsdz/lowdown/blob/master/main.c#L84-L86 this looks like the code that is bork | 20:28:28 |