yowl
yowl
Would I be able to create the `Game` struct in the `start` call (called once per W4 lifetime), then clobber that over some static byte[] arrary. Then in at the...
Ah, because I thought it was `=` not `=>` . And if you take that sharplab link and switch to the IL it has a `newobj`. Maybe RyuJit removes that,...
> It won't be a heap allocation How does that happen, when I look at the IR, I see the helper call for `RhpNewArray` ``` ------------ BB02 [000..01C) (return), preds={BB01}...
Aha! Right, better download Roslyn to see how it decides. Thanks! I guess ` [System.Private.CoreLib]` is my problem
Thanks, that saved me some debugging. Looks like this ctor is important ``` unsafe public ReadOnlySpan(void* pointer, int length) { _pointer = new ByReference(ref Unsafe.As(ref *(byte*)pointer)); _length = length; }...
Hmm, I guess that's obvious looking at the IL, oh well.
> I'm now leaning towards saying "anything from [0,255] is permitted" That works for me 👍
I want to add that invoking c++ constructors through a new exported function would be a problem for .Net CoreCLR as any exported function would go through it's reverse P/Invoke...
I notice that llvm geps where the offset is a constant produce a local in the wat which is unneeded and optimised away with O1 . It would seem possible...
Sounds good. Is `components` a bit generic, some other options `wit-bindgen-generated`, `wit-bindgen`, `wit-components`, `wit-generated`.