Shouldn't CWG2458 resolution also modify [basic.lval] wording, as in #3262?

Shouldn't CWG2458 resolution also modify [basic.lval] wording, as in #3262?

Maybe it should, but it's not part of the motion, so we should revisit after we're done with the motions.
I think C::f, when it is an lvalue, still "identifies a function" per [basic.lval].

I think C::f, when it is an lvalue, still "identifies a function"

This is fine. The thing is that C::f identifies a function when it is a prvalue, whilst [basic.lval] says that evaluated expressions denoting functions are glvalues. And I think class member access expression e.C::f is where C::f is evaluated and is a prvalue.

And I think class member access expression e.C::f is where C::f is evaluated and is a prvalue.

Ok, but then the proposed change in #3262 isn't actually what we want post-CWG2458.
Instead, we want to augment the definition of "prvalue" such that C::f fits.

Instead, we want to augment the definition of "prvalue" such that C::f fits.

The current definition — A prvalue is an expression whose evaluation ... computes the value of an operand of an operator — looks fine to me, since C::f computes the value of the right operand of the . or -> operator. The definition of «glvalue» should say that some expressions denoting a non-static member functions are not glvalues.

The "glvalue" and "prvalue" definitions do not exhaustively describe which expressions yield which kinds. I agree that C::f could be the operand of an operator. But C::f also designates a function (in case we take its address). So, this seems fine both ways.

The "glvalue" and "prvalue" definitions do not exhaustively describe which expressions yield which kinds.

I'm fine with it, but let me remind that there was CWG1076 issue about value category inconsistency:

The taxonomy of value categories in 7.2.1 [basic.lval] classifies temporaries as prvalues. However, some temporaries are explicitly referred to as lvalues (cf 14.2 [except.throw] paragraph 3).

which can be reformulated for CWG2458 resolution:

The taxonomy of value categories in [basic.lval] classifies function designators as glvalues. However, some non-static member function designators are explicitly referred to as prvalues