On Fri, 14 Feb 2020, Jason Merrill wrote: > On 2/12/20 5:15 PM, Patrick Palka wrote: > > Two equal atomic constraint expressions do not necessarily share the same > > tree, > > so we can't assume that two ATOMIC_CONST_EXPRs are equal if and only if they > > point to the same tree. > > This is incorrect; comparison of atomic constraints is based on them coming > from the same actual tokens, which is why we use pointer comparison. > > In your concepts-partial-spec7.C, test2 is properly rejected. If you want to > be able to write the same constraint in multiple places, you need to use a > concept. > > 13.5.1.2: > > Two atomic constraints are identical if they are formed from the same > expression and the targets of the parameter mappings are equivalent according > to the rules for expressions described in 13.7.6.1.
I see, that makes complete sense now. I was stumbling over what the spec meant by "the same expression." Thanks for clearing up my misunderstanding.
