@zygoloid, please decide whether that is editorial or not. Implementations seem to agree on the intent of the wording.

I'm actually confused by [dcl.array]/3 (unrelated to #2012). Consider int (*p)[] = {0};. It has an array declarator ((*p)[]), which is followed by an initializer (= {0}). Does it mean that the array bound should be deduced?

@cpplearner: The point here is that we're not actually initializing an array object. I agree that could be clearer in the wording.

I'm not comfortable with making this change editorially, and would like CWG to explicitly consider it. (And CWG should also consider whether new-expressions should get the same treatment.)

I just found CWG 1525 which seems to suggest that [dcl.init.aggr] should apply in an explicit type cast.

OK, that's good enough for me.