CC @timothy-king.

1. There is always some tension between more informative and being pithy and precise in vet warnings. IMO the current message is already a bit long, but no part of it is redundant. The end is directing users to a helpful place to self service format %s has arg <redacted code> of wrong type bool see also https://pkg.go.dev/fmt#hdr-Printing. We have URLs fields coming soon so perhaps this verbosity will soon no longer required as we can redirect to a landing page that then has the equivalent of see also https://pkg.go.dev/fmt#hdr-Printing. So vet may have more space to play with soon.

2. So the message is already pretty helpful. We are mostly discussing whether the message could be more helpful.

3. I am somewhat hesitant to think we know user intent here. The checker does know something is wrong so the current message is fine from that regard. Adding a recommendation ("should") kinda requires having an opinion about user intent. And there are also multiple choices valid. Why not recommend %v? We can soften the language do be (Did you mean %t?).

4. I am not sure we want to add a recommendation just for bool. What do we think about other cases? Is it all of the types that have a default format for %v? Do we really want to recommend %d over %x or %v for int?

Thoughts from others?

Thoughts from others?

At a glance, is incompatible type any better than wrong type?

Occasionally it's relevant if some wrong type implements fmt.Formatter, or has a GoString, String, or Error methods, if there's an inclination to provide a guess about what verbs are compatible.

I agree with @timothy-king. The message tells you the offending expression, type, and format verb. Usually there are multiple possible fixes among which the tool has no basis to choose---was the problem with the format or the argument?---yet the appropriate fix is obvious to the user.

Great, thank you for taking a look at this!