Is there a reasonable situation where someone might want to format a function pointer as a base-10 number? It would be a shame to deny that because of a rare misunderstanding. (Have you seen this come up more than once?)

"Reasonable" is very subjective. Right now functions behave just like pointers, which includes %d showing them in base-10 and %v as hex. That is probably what an programmer with a C background would expect, too.

I do not necessarily mean fmt behavior should change; I'm merely documenting the problems I see more than one person having. This misunderstanding may still be rare, but I've seen more than one person succumb to it; the part that especially tricks them is when they expected to see a number, and can't tell what they're seeing is a completely different kind of a number.

Perhaps there is an alternate route, e.g. via go vet, golint, etc -- though the new gophers that suffer from this don't know of these tools either! So perhaps the "right fix" is play.golang.org integration of go vet/golint?

I don't think the fmt behavior could change, even if it was desired. A vet check for this seems natural. Perhaps this CL should be about that. (wdyt @robpike?)

There's issue #7597 about adding vet to the playground, and even some CLs to make it happen. I am due to review those soon.

Possibly worth catching in vet but I am unconvinced. There's little point in ever printing a function value; you can't even compare function values. But it's a pretty rare case and the "newbie" argument is never compelling. A better solution might just be to teach newbies to use %T when they are confused.

Vet's job is not to catch every possible mistake.

This just came up here: https://go-review.googlesource.com/#/c/14031/1/src/cmd/compile/internal/gc/fmt.go

Yes. I think it's such an easy thing to add to vet's existing print checker that it should be done.

CL https://golang.org/cl/14122 mentions this issue.