Jeroen van Rijn
Jeroen van Rijn
Great work. It deserves a knighthood.
AMD/Intel can compute `abs` in several ways. https://stackoverflow.com/questions/32408665/fastest-way-to-compute-absolute-value-using-sse ARM has a SIMD operation for it for floats, apparently. https://developer.arm.com/documentation/ddi0596/2021-12/SIMD-FP-Instructions/ABS--Absolute-value--vector--
Still crashes the compiler: ``` W:\Odin-other\Odin-test\foozle>odin run units.odin -file -show-debug-messages [DEBUG] build_paths[0]: W:/Odin-other/Odin-test/foozle/units.odin [DEBUG] build_paths[1]: [DEBUG] build_paths[2]: [DEBUG] build_paths[3]: C:/Program Files (x86)/Windows Kits/10/bin/10.0.20348.0/x64/ [DEBUG] build_paths[4]: C:/Program Files (x86)/Windows Kits/10/Lib/10.0.20348.0/um/x64/ [DEBUG]...
Because you're trying to fill `[2]f32` with either 1 or -1, and then you have a second ternary for a second return value the proc doesn't have, this was never...
Shouldn't have even made it to the backend, because the first param isn't an `^int`.
Still present: `Internal Compiler Error: Type_Info for 'untyped nil' could not be found`.
Any particular motivation, @laytan? 1) I don't know whether it makes sense for the `bytes` package to have case conversion. The package doesn't assume any particular encoding and the `strings`...
> Most of core:bytes is a 1:1 to core:strings. Maybe we should remove all the duplicated procedures? There is a case to be made to have both, and they could...
If you want, feel free to add those tests to `tests\core\...` as well.
Have you tried packing the varargs together into one argument? `push_op(main, .mov, {.RAX, .RCX})`