| 15 Feb 2025 |
 Whovian9369 | (Ah this may not be the best channel for the question, sorry!) | 02:30:57 |
 aloisw | In reply to @llakala:matrix.org
but nix bugs have resulted in the fileset library from being banned from nixpkgs because they have to constantly work around these issues In fairness, this bug has not been fixed in Lix either as far as I know. | 05:58:40 |
 K900 | The fun thing about that particular bug is that it's impossible to fix without changing existing semantics | 06:15:53 |
| K900 | You could even argue there is no bug, just extremely unexpected behavior | 06:16:14 |
| K900 | Because the behavior is consistent | 06:16:50 |
| K900 | It's just completely insane | 06:16:55 |
| aloisw | In reply to @k900:0upti.me
The fun thing about that particular bug is that it's impossible to fix without changing existing semantics I don't think "code behaves differently when evaluated in a "chroot" store" are existing semantics that are worth preserving. | 06:17:33 |
 Charles | clearly the solution is new fixed builtin functions with ' suffixed to their names | 06:17:53 |
| K900 | In reply to @aloisw:julia0815.de
I don't think "code behaves differently when evaluated in a "chroot" store" are existing semantics that are worth preserving. I mean | 06:18:21 |
| Charles | or perhaps you could pull from php naming conventions because it's done this kind of thing a lot | 06:18:23 |
| K900 | I'm not saying it shouldn't be changed | 06:18:38 |
| K900 | I'm just saying the situation is more complicated than just "bug not fixed" | 06:18:57 |
| K900 | In practice I think we should go ahead and just break it because it's absolutely impossible for the old behavior to actually be useful for anyone ever | 06:19:22 |
| Charles | i was gonna ask what your take would be | 06:19:40 |
| K900 | But it is still a compatibility break and should be handled with care | 06:21:31 |
| aloisw | In reply to @k900:0upti.me
I'm just saying the situation is more complicated than just "bug not fixed" I don't think it is actually, fixing the bug would just make a behaviour that is already present (namely in the non-"chroot" case) consistent. | 06:21:55 |
| Charles | is there a good testing framework for nix expressions? doing assertions on positive cases is obvious, but asserting that eval should fail and how it should fail is less so | 06:54:34 |
| Charles | there are docs now btw | 08:58:56 |
| Charles | * there are docs now btw, in the readme mostly | 08:59:10 |
| Charles | * there are docs now btw, in the readme mostly | 08:59:43 |
| Charles | Project idea: nix minifier | 09:04:00 |
|  BenjB83 joined the room. | 10:18:02 |
| BenjB83 changed their display name from BenjamÃn Buske to BenjB83. | 10:43:15 |
 piegames | In reply to @charles:computer.surgery
Project idea: nix minifier I think nixfmt can already do that | 10:56:03 |
 Maximilian Marx | for maximal confusion, call it minix | 11:03:49 |
| piegames | Or it at least has the infrastructure to trivially implement, because it has other IR transformations already | 10:56:45 |
| piegames | In reply to @kfears:matrix.org
Lix has also been backporting stuff from newer versions like 2.24, like pipe-operator, so as far as normal or even advanced usage (going as far as maintaining Nixpkgs) is concerned, you should be good to go Bad example, pipe operator was implemented first in Lix and then independently in CppNix (they have different semantics) | 10:58:36 |
 KFears (no longer human) | In reply to @piegames:flausch.social
Bad example, pipe operator was implemented first in Lix and then independently in CppNix (they have different semantics) Wut | 11:19:33 |
| KFears (no longer human) | Tell me more? | 11:20:31 |
| piegames | Different precedence for the backpipe operator | 11:45:26 |
