| 7 Dec 2024 |
 6pak | we have to inject DllImportResolver in some way anyway | 23:29:18 |
| 6pak | so I thought of making a <NixInitClassWhatever> with a ModuleInitializer method in it | 23:29:35 |
| 6pak | might aswell add a static string[] Needed property to it | 23:29:45 |
 GGG | yeah, we can do that, but I'd like a way to manage stuff without needing to decompile code | 23:29:52 |
| GGG | I guess having a static readonly array could work | 23:30:02 |
| GGG | but that wouldn't be very efficient since we'd need to linearly iterate through it | 23:30:17 |
| 6pak | eh | 23:30:45 |
| 6pak | you need to iterate over every method to find DllImport attributes anyway | 23:30:53 |
| GGG | though that should only matter for big stuff like game engines | 23:30:56 |
| GGG | yeah, but that's compile time, not runtime | 23:31:05 |
| GGG | I'm worried about possible performance impacts to applications running in nixos | 23:31:17 |
| GGG | and those would be very hard to debug since we'd be "invisibly" (to the end user) adding stuff to the code | 23:31:39 |
| 6pak | I mean isn't that how dynamic linker works in native land | 23:31:46 |
| GGG | * and those would be very hard to debug since we'd be "invisibly" (to the end user and original package maintainers) adding stuff to the code | 23:31:49 |
| GGG | well, it loads stuff up on boot | 23:32:00 |
| GGG | nothing prevents someone from calling NativeLibrary.Load multiple times | 23:32:12 |
| GGG | as dumb as that would be | 23:32:15 |
| 6pak | it gets cached afaik | 23:32:21 |
| GGG | though I guess then the blame is on the dev | 23:32:23 |
| 6pak | but it might get cached by the file path returned from the resolver | 23:32:43 |
| 6pak | * but it might get cached on the file path returned from the resolver | 23:32:47 |
| GGG | I think this is just me worrying about an issue that doesn't exist honestly | 23:33:01 |
| GGG | wouldn't worry about it for now | 23:33:11 |
| GGG | you'd need a really massive number of libraries to actually get a performance hit from a linear search | 23:33:33 |
| 6pak | I bet the ffi stub for dlopen will take more time than iterating 100 strings in the array | 23:33:46 |
| GGG | yeah, it's probably a non-issue | 23:34:02 |
| 6pak | I'd be down to write it | 23:35:53 |
| 6pak | so the most important question: whats the name | 23:36:06 |
| GGG | patchcil | 23:43:02 |
| GGG | is what I was gonna call it | 23:43:06 |
