Dylan Baker

The more I think about it the more I like option 3, since it has the added benefit of simplifying searching for "core" cps files as well as simplifying searching...

> I'm not sure what is being clarified. Are you looking for something like, "For example, while you may think of your package as my:cat, this is not a valid...

I'm sorry, I wrote up a reply yesterday, and apparently never posted it! I thought about Option 1, and don't think it actually helps, consider the following: ``` cps/zlib/zlib.cps #...

I believe that the additional windows files the `.lib` and `.pdb` must have the same name as the `.dll` to be automatically loaded, I think that Windows `.exe` files can...

Hmmm... IIUC, then we would just need to install the .lib and .pdb with the same name but .pdb, though maybe we have someone with more Windows expertise? skipping Windows...

Rustc is just awful to deal with, and this is no exception. It invokes different link.exe directly, but on linux it invokes **gcc** as the linker. It's really, really hard...

I don't see how it can't be required if you're using a Windows implementation `.lib` with a runtime `.dll`, Meson would be unhappy if you only have the `.dll` as...

If I remember correctly the only time we explicitly link a module (except in some deprecated backwards compat cases) is for Android because it requires it, otherwise I think we'd...

> > but I realised one thing: we probably don't need to add `['-lFortranRuntime', '-lFortranDecimal']` (now `['-lfortran-rt']`) at all, because flang should take of this itself. > > IIRC the...

@bonzini I can press merge if you think this is the right approach (since you've already looked at it more than I have)