Is this a case that go vet could provide an error for?

If a deferred function writes to a variable defined in the surrounding functions scope that has already returned, it seems to me like that is likely an error.

The spec says

deferred functions are executed after any result parameters are set by that return statement but before the function returns to its caller.

which is a more accurate representation of what's happening.

this might not pass the accuracy bar for cmd/vet, ex #8220

I think we could reasonably tweak the blog entry to make it slightly more accurate (cc @adg since he wrote the blog post).

Happy to review any amendments to the blog post.

An individual vet check for this pattern seems to be an overkill for the reason mentioned in #8220. A more general check is to examine whether the writes within defer are used or not. In this example, b = 2 is not used (by the return) so this is suspicious. If b is a named return value then it is used by "return b". Of course the checker shall know about the correct control flow, e.g. "deferred functions are executed after any result parameters are set by that return statement but before the function returns to its caller".

func f() int {
    b := 1
    defer func() {
        b = 2
    }()
   return b
}

This may be added as an extension of the "unusedwrite" checker.