Um… what? These changes would make use; valid, which it is certainly not. A use statement must be followed by an identifier.

The grammar already existing appears to me to be correct.

I take back what I said about the grammar existing being correct—I’ve been reading other grammar rules too much recently (the IETF RFC ones in particular) and was thinking […] meant optional. There is a problem with the current grammar, but the new one is wrong also. I think that just tacking a question mark after the ] on the existing grammar would be correct? OK, OK, that responsibility got shifted to the path_glob rule… not doing well today, am I?

My line of thought is that a single identifier is also a path. path_glob can match multiple paths.

// Valid declarations
use ident = simple::path;
use simple::path;
use {many, idents}; // doesn't start with an ident!
// Invalid
use simple::path = another::path;
use *;
use;