I'm not sure quite what you are suggesting. The blog post looks right, but it's not saying anything that the language spec doesn't already say, albeit tersely. The language spec already says that the type must be identical. Do you think we should just add a couple more examples?

@ianlancetaylor thanks for your prompt reply! :)

The blog post looks right, but it's not saying anything that the language spec doesn't already say, albeit tersely. The language spec already says that the type must be identical.

I agree, the information is there but, as you said, it is probably packed in a way not easily understandable, especially for Go newbie like me (I'm also coming from a Java background and that doesn't help for sure). As additional context, before being bitten by the mentioned issue (i.e. impossible type assertion), I did read the type assertions section before but TBH I didn't realise at all that type assertions on pointer types were even a thing (again, probably it's just me).

Do you think we should just add a couple more examples?

My suggestion would be to add at least one example of type assertion involving a pointer type and, if possible, put in words the subtle difference between pointer vs non-pointer type assertion.

To be clear, I don't think there is any difference between pointer and non-pointer type assertion. An interface value holds a value of some type. That type can be a pointer type or a non-pointer type. Type assertions work exactly the same way.

But I think it would be reasonable to add an example to the language spec of a type assertion to a pointer type. Leaving for @griesemer to decide.

What @ianlancetaylor said. I think the blog is at least confusing when it says "keep in mind that a pointer to a type is not the same as the type itself". It seems to imply that this is unusual behavior, which it is not.

I suspect the confusion comes from not being intimately familiar with the notion of a pointer type, which is understandable when coming from "pointer-less" types such as Java or JavaScript (never mind that these languages use pointers everywhere under the covers but try to hide that, which leads to other confusions).

I've reread the spec section on type assertions and it seems pretty clear. It talks in general terms about a type T and refers to the notion of type identity. Thus, to understand this one needs to understand (read) the section on types and type identity. The latter section makes it clear that a pointer type *T and its base type T are different. So adding more prose seems not justified here; in general the spec is written pretty concisely and we avoid restating things that are being said elsewhere.

Leaving open in case somebody wants to suggest a good pointer example. Send any such CL to me. Thanks.