The current design for the non-decimal radix formats assumes the user needs to see the actual implementation bit pattern, so its whole purpose is to expose that information in a more human-readable fashion (and we could certainly make a couple further improvements in that direction).
The human readability you're asking for is a different use case, so more detail on some situations where you need to see hex that is not directly describing the underlying bit pattern would be helpful (sorry, I didn't get much out of the French thread--maybe that has the answer). I agree that -2A is more readable than FFFFFFD6, but in what situations is that superior to the options you get from -42 (most readable) or FFFFFFD6 (most explicit)?
For sign-magnitude representations like floating-point, this would definitely be nice, but for 2s complement values we need some more background info, apparently from the mathematicians and Python users.
The point of hex is that it's easily related to the underlying bits, in that each hex digit represent a nibble (4 consecutive bits). Once you bring in a minus sign, the relation is not so simple so it kind of defeats the point IMO.
ToeCutter: That's the point of hex *for programmers*, not for mathematicians. You think bringing in the - sign is bad... try sigmas and integration marks, all in hex... there's a proof I saw in college of some number theorem that was "easier" to solve in hex. I've also seen a geometry proof that was worked out in base sixty.
