1's comp

[Chuck Guzis]

On 05/22/2016 06:23 AM, Toby Thain wrote:

> I didn't mean performance - I meant programmer effort to ensure 
> robust/correct code. The Grishman book didn't reassure me, but thanks
> for the additional info.

Well, if you consider it carefully, does zero really have a sign?  So,
the convention of considering 0 to be a positive value is, in a sense,
illogical.  But no, I don't recall that the issue of -0 ever raised much
of a problem in either CP or PP code.

I've long wondered if the abandonment of ones' complement arithmetic on
modern ISAs was more due to the requirement that ones; comp requires an
end-around carry, where two's comp does not.

--Chuck