| 20 Sep 2024 |
 ma27 | IIRC it used to be an issue when git patched the underlying issue and https://github.com/NixOS/nix/commit/0256e5578e97a11db66207e1f8e231db115c91f8 was the workaround? | 17:21:35 |
| ma27 | (I think I filed a follow-up because it didn't cover all cases, but you get the idea) | 17:21:54 |
 aloisw | But the git executable has the same behaviour. And I can reproduce with a clean repo on Lix (with different error message). | 17:22:44 |
 aleksana (force me to bed after 18:00 UTC) | In reply to @k900:0upti.me
The more common failure case is sudo nixos-rebuild --flake ~/my-cool-nixos Just proposed a transparent way to ask for root privilege without sudo or --use-remote-sudo, but haven't looked into it yet: https://github.com/NixOS/nixpkgs/issues/342657 (this should be discussed in this issue) | 17:36:58 |
 puck | In reply to @aleksana:mozilla.org
@[Robert Hensing (roberth)] @Eelco also one folk from Chinese group reported this which seems to be the same type of bug (but only triggered on hydra): https://github.com/NixOS/hydra/issues/1409 ..okay, that's stack exhaustion, i think | 17:42:04 |
| puck | yeah, hydra-eval-jobs makes the same mistake nix-eval-jobs did | 17:43:37 |
| puck | starting a worker on a non-primary thread | 17:43:46 |
| puck | and because it's on a non-primary thread (with a fork inbetween), i think the stack guard got lost | 17:44:29 |
| puck | it's hard to tell without looking at the process under gdb directly (core files don't have memory maps), and i'm not in the state to run the entire build rn | 17:46:40 |
 jade_ | In reply to @puck:puck.moe
it's hard to tell without looking at the process under gdb directly (core files don't have memory maps), and i'm not in the state to run the entire build rn they do, maint info regions iirc | 22:45:07 |
| jade_ | but info proc map may work as well in core files ime | 22:45:22 |
| puck | info proc map misses stack (just mapped from files), maint info regions has all sections but they're labeled load1 through load410 | 22:47:13 |
| puck | * info proc map misses stack (just mapped from files), maint info sections has all sections but they're labeled load1 through load410 | 22:48:07 |
| puck | rsp is set to 0x7fbaeee00fd0, and staring at maint info sections:
[53] 0x7fbaeee00000->0x7fbaeee01000 at 0x612c8000: load41 ALLOC READONLY
[54] 0x7fbaeee01000->0x7fbaef601000 at 0x612c8000: load42 ALLOC LOAD HAS_CONTENTS
which feels like a stack plus its guard page to me | 22:48:26 |
| puck | (especially because the frame starts at 0x7fbaeee01050, so juuuust inside that region) | 22:49:02 |
| 21 Sep 2024 |
|  Autumn joined the room. | 05:42:47 |
 mwoodpatrickmx | I'm confused where sha256 value specified in the unit package comes from.
https://github.com/NixOS/nixpkgs/blob/nixos-unstable/pkgs/servers/http/unit/default.nix#L77
When I use
nix-prefetch-git --rev 1.33.0 --url https://github.com/nginx/unit.git
| 16:33:58 |
| mwoodpatrickmx | * I'm confused where sha256 value specified in the unit package comes from.
https://github.com/NixOS/nixpkgs/blob/nixos-unstable/pkgs/servers/http/unit/default.nix#L77
When I use
nix-prefetch-git --rev 1.33.0 --url https://github.com/nginx/unit.git
I get a completely different hash.
Any pointers on the best way to creating and testing a new package from scratch would be greatly appreciated
| 16:35:32 |
 magic_rb | In reply to @mwoodpatrickmx:matrix.org
I'm confused where sha256 value specified in the unit package comes from.
https://github.com/NixOS/nixpkgs/blob/nixos-unstable/pkgs/servers/http/unit/default.nix#L77
When I use
nix-prefetch-git --rev 1.33.0 --url https://github.com/nginx/unit.git
I get a completely different hash.
Any pointers on the best way to creating and testing a new package from scratch would be greatly appreciated
leave the hash empty, "" in the nix expression, eval it, nix will complain about a mismatch, fix the mismatch | 16:53:51 |
| mwoodpatrickmx | In reply to @magic_rb:matrix.redalder.org
leave the hash empty, "" in the nix expression, eval it, nix will complain about a mismatch, fix the mismatch That's definitely good to know but I'm curious as to why they are different, Is this approach documented somewhere? Any pointers on best practices for Nix Package development would be greatly appreciated | 16:59:21 |
| magic_rb | idk honestly, ive learned to not question certain things, lifes too short | 16:59:55 |
 K900 | This is not the right room for this | 17:00:20 |
| K900 | You want #Nix / NixOS or #contri | 17:00:26 |
| K900 | You want #Nix / NixOS or #Nixpkgs / NixOS contributions | 17:00:28 |
| magic_rb | (right, forgot to move this, sorry) | 17:00:38 |
| 22 Sep 2024 |
|  Sam Lehman joined the room. | 13:28:03 |
| 23 Sep 2024 |
|  elikoga joined the room. | 15:29:26 |
|  connor (he/him) (UTC-7) changed their display name from connor (he/him) (UTC-5) to connor (he/him) (UTC-7). | 17:57:57 |
| 24 Sep 2024 |
 Mic92 | It looks like I shaved of 10minutes or though from the ubuntu CI build. | 07:20:32 |
| Mic92 | * It looks like I shaved of 10 minutes or though from the ubuntu CI build. | 07:25:49 |
