Re: Discussion on tetrits motion
Dan
On 24 Apr 2010, at 21:39, Dan Zuras Intervals wrote:
>> I fear such an analysis demonstrates the proposed propagation mechanism
>> would be a setback for P1788; it deviates too far from the semantics we
>> agreed on in Motion 8. For example, Motion 8 specifically identifies the
>> purpose of decorations is to help disambiguate between
>> X \union {empty} = X and X \union NaI = NaI,
>> where "NaI" is really some bare decoration. But the definitions being
>> proposed in the current motion implicitly "undo" this disambiguation,
>> leaving us, to some extent, back where we originally started about a year
>> ago.
>
> We have not yet discussed a definition of an NaI.
> Nor even the need for one.
>
> So I'm not sure how to sensibly respond to this.
See my previous posting. I find your comment hard to comprehend. We need a way to handle invalidly constructed intervals. Especially in large-scale work, like in your nuclear supercomputer where it may construct 1,000,000 intervals on 1,000 machines in parallel, and owing to a bug, a random 50% (or 0.5%, or 0.005%) are invalid. And you want to find just which ones, which is a necessary start to finding out why.
BTW the 128 garbage bits of the interval part of an invalid "double" interval are a fine place to dump diagnostic data. I guess that if done in software, this would slow a vectorised calculation down by an order or two of magnitude, but in principle with hardware support it could be done without slowing things down at all, e.g. to give a linked list leading back to the address of the code line where "invalid" first appeared?
We had NaI till the Hayes-Neumaier scheme of Motion 8 replaced it. NaI's still there in concept. Just implemented more efficiently.
John