Doesn’t that mean that they are pretty trivial to accidentally (or purposefully) ignore? Sounds bad for maintainability? Plus what about issues like an error value able to be a slim pointer null, a fat pointer null (which won’t compare to null correctly), or an error itself? And what is the state of the first return value when the error has happened? Does it still contain useful information? Is it garbage? If it’s garbage then why does it exist at all? If it “depends on the call” then why isn’t that told by the types?

Just a tiny tiny sampling of reasons why every language need sum types. Just having prod types like structs is insufficient, need sum types as well, like variants.

Thanks for the blog, GoLang have support for errors in a really simple way. Go functions return errors as a second return value.