Type errors with inference need stacks

I’ve been thinking about Flow errors recently. They’re often inscrutable, and this is a problem that comes up in any typechecking engine with cross-function inference. Flow’s error format is a little strange and takes newcomers a while to get used to. Once you get used to it it fades into the background. Once you know the general layout, I repeatedly run into errors that are very confusing and don’t point to the real source of the error whenever there is any indirection (which there always is i...

Submit