Re: min / max and empty intervals
Dan Zuras wrote:
You know, that reminds me of something.
This is totally off the subject but, Nate, you once argued for
the need for non-standard intervals for your graphics work.
And I know other have discussed them for solving equations.
Have you figured something else out or is this just a subject
for some other day?
I remember we once kicked arouond a decoration called
'non-standard'.
Do we still need that?
No. I don't see any reason a "non-standard" decoration is needed.
We just need Kaucher arithmetic. Interval processors without Kaucher
arithmetic will never run the various graphics algorithms as fast as those
with it. P1788 will at some point have to decide if it cares about this or
not.
Motion 25 would be compatible, regardless if P1788 decides to include
Kaucher intervals or not (the motion is not motivated by Kaucher
arithmetic).
Or is this a subject better discussed after we have settled
much of the other decoration issues?
I believe Kaucher arithmetic can be discussed independently (and at a later
date) of the current issues surrounding decorations, property tracking, and
FTDIA. In fact, I recommend this.
In the meantime, I still believe very strongly that P1788 needs to pass
Motion 25, regardless.
I've reviewed John's v3.02 this morning. I appreciate the efforts, but it
seems to still miss the point and I think with this draft the gap between us
is widening, not shrinking. :-(
So my hopes that P1788 is going to end up in the right place are a bit shaky
right now. But I'm along for the ride until the end, for better or worse.
I am just one of the many voices crying in the wilderness...
I know we are all trying in our own ways to make a good standard, so I don't
doubt anyone's intelligence, experience or motivations. At the end, our
company (Sunfish) will examine where P1788 has landed, and conform to the
standard if it meets our needs or ignore the standard if it doesn't. That's
really what it all boils down to.
Nate