We hacked Clang to return a random number for any undefined integer
operation. Stuff breaks. Most of SPEC CINT breaks.

There's no way Unisys's compiler implements non-two's-complement if they
run any substantial amount of C/C++ from the wild.

John



On 10/24/13, 4:40 PM, Richard Smith wrote:
> On Sat, Oct 19, 2013 at 8:51 PM, Jason Merrill <jason_at_[hidden]
> <mailto:jason_at_[hidden]>> wrote:
>
> On 10/18/2013 11:15 PM, Lawrence Crowl wrote:
> > There were machines being sold in the 90's that were not two's
> complement, but I've lost track
> > since then.
>
> http://stackoverflow.com/questions/6971886/exotic-architectures-the-standard-committee-cares-about
>
> The Unisys machine mentioned in the top-rated answer still seems to
> be sold.
>
>
> Has anyone ever written a non-twos-complement C++ compiler for this machine?
> Has anyone ever written *any* non-twos-complement C++ implementation?
> And... does anyone intend to write a non-twos-complement *C++17*
> implementation?
>
> If the answer to these questions is "don't know", how do we find out?
> Strawman suggestion: remove non-twos-complement support from the
> standard and wait to see if anyone complains (with the intent that we
> would restore such support if there are actual complaints, not just
> hypothetical ones). The world would have nearly four years to notice and
> react. Thoughts?
>
>
> _______________________________________________
> ub mailing list
> ub_at_[hidden]
> http://www.open-std.org/mailman/listinfo/ub
>