| 9 Jul 2025 |
 raitobezarius | i'm not clear on the security consequences of all of this | 17:22:11 |
 jade_ | probably not many or none | 17:27:25 |
| raitobezarius | ok | 17:28:19 |
| raitobezarius | so for me, i will remove this topic from my mental charge | 17:28:24 |
|  lambadada joined the room. | 23:49:48 |
| 10 Jul 2025 |
| jade_ | anyone interested in reviewing some improvements to the notorious dependency propagation section in the nixpkgs manual? I added the necessary explanation for the math in there. https://github.com/NixOS/nixpkgs/pull/423954 | 05:48:41 |
 K900 | can we kill target instead | 05:50:13 |
 Linux Hackerman | In reply to @raitobezarius:matrix.org
Actually: 13013, 13387 and 13388 are not ported should I send them, or are they sufficiently on $somebody's radar? | 13:27:34 |
| raitobezarius | I think you can send them but expect rework to be necessary | 13:27:53 |
 niko ⚡️ | In reply to @k900:0upti.me
can we kill target instead But what about the Canadian cross then which isn’t even a thing anyway?! /s | 13:50:47 |
| Linux Hackerman | In reply to @raitobezarius:matrix.org
I think you can send them but expect rework to be necessary do we already have something for dealing with the renames? | 14:04:37 |
 emily | would actually be more possible if we killed target, FWIW | 14:08:59 |
| emily | LLVM by default on all platforms… one day it will happen… | 14:09:29 |
| raitobezarius | In reply to @linus:schreibt.jetzt
do we already have something for dealing with the renames? sorry, we don't | 14:10:28 |
| raitobezarius | if you know of a magic tool, i'm interested | 14:10:33 |
| Linux Hackerman | I wrote a script which is not very magical, but remains to be seen how effective it is | 14:44:59 |
| 11 Jul 2025 |
| jade_ | https://gerrit.lix.systems/c/lix/+/3632 confused at why I had to hit owners-override on here; I thought that we had any-owner as default for unowned files? | 03:37:37 |
| jade_ | https://gerrit.lix.systems/plugins/gitiles/lix/+/refs/meta/config/OWNERS oh! that looks like maybe not the cleverest idea as it will get inherited if we don't have a descendent config that looks like *. | 03:55:59 |
| jade_ | well, I guess we have a reason to ship https://gerrit.lix.systems/c/lix/+/3633 haha | 03:57:58 |
 piegames | As for owners, please add me to functional/lang and functional2/lang | 08:48:24 |
| raitobezarius | In reply to @jade_:matrix.org
https://gerrit.lix.systems/c/lix/+/3632 confused at why I had to hit owners-override on here; I thought that we had any-owner as default for unowned files? Worth root cause analyzing or just shipping owners indeed | 09:09:37 |
| raitobezarius | (esp given that you are default owner on the top of that!) | 09:09:49 |
| emily | FWIW, the build directory relocation thing on macOS is breaking a fair number of builds (mostly in tests) because Unix socket paths are limited to like 104 bytes or something. (I believe on Linux it's 108 but we don't expose the /nix/var/nix/builds path there) | 14:20:41 |
| emily | I don't really have a good suggestion other than… rename it to something shorter. /tmp vs. /nix/var/nix/builds is a pretty big difference | 14:21:07 |
| emily | but I don't know how palatable /nix/b would be :) | 14:21:19 |
| K900 | 🅱️ | 14:21:55 |
| emily | another alternative would be shortening or eliminating the nix-build- prefix, which is now presumably redundant | 15:07:46 |
| raitobezarius | hm | 18:13:52 |
| raitobezarius | how can this be a sustainable fix though? | 18:13:59 |
