On 20 Jun 2013, at 16:57, G. William (Bill) Walster wrote:
Or what about an implementation that implements "crude range tests"
<http://www.deepdyve.com/lp/springer-journals/using-pillow-functions-to-efficiently-compute-crude-range-tests-4G0VslI8nH>
and/or endpoint evaluation for expressions that have been proved by a compiler to be monotonic over given intervals or boxes?
In my view, a compiler should have "no holds barred" except for failure to contain.
Again, splendid techniques but again, in my view, the standard has no opinion on them.
John P