Your suggestion is an improvement but it still seems wrong to me, operator< could be a member. We could talk about unary/binary operators not member/non-member, but since the rule applies to both unary and binary forms of operator+ and operator- we don't really need to distinguish between them at all.

That operator< (described in 24.5.1.3.14[reverse.iter.op<]) is declared as a non-member. I don't think an implementation is allowed to declare it as a member.

I agree that there's no need to distinguish between members and non-members here.

24.5.3.2[move.iter.requirements] reads

The template parameter Iterator shall meet the requirements for an Input Iterator (24.2.3). Additionally, if any of the bidirectional or random access traversal functions are instantiated, the template parameter shall meet the requirements for a Bidirectional Iterator (24.2.6) or a Random Access Iterator (24.2.7), respectively.

Perhaps 24.5.1.2[reverse.iter.requirements] could be phrased in a similar way. I mean, 24.5.1.2/2 could simply say

Additionally, if any of the random access traversal functions are instantiated, the template parameter shall meet the requirements for a Random Access Iterator (24.2.7).

@cpplearner: Are you still interested in pursuing this PR?

@zygoloid, @jwakely: Close this PR as abandoned?

Here's a p/r: globalnamespace scope operators