Re: But I DO like the smell of Motion 7...
On 2009-09-01 06:52:14 -0700, Dan Zuras Intervals wrote:
> Some observations about the form of Motion 7:
>
> I should say that, in spite of voting against this motion
> along with John on the grounds that we should have our
> discussion first, I DO like the form given by Prof Gudenberg.
>
> Namely that NaIs should (in my language, not his):
>
> (1) Never be returned by an operation on F intervals.
This would probably be a consequence of the other motions on
operations. In fact I prefer: any operation on F intervals returns
F intervals (so that this excludes things like undefined behavior).
> (2) Can only be created from outside the world of intervals.
> That with one NaI the only example is constructors.
> With multiple NaIs, there may be more.
This is a bit ambiguous, unless (2) is seen as a rewording of (1).
> (3) That once created are propagated universally.
Some people may disagree (e.g. min/max on a list with NaI representing
missing data -- but one may want to use the empty set for that).
> (4) That all comparisons on NaIs are false except for
> isNaI(NaI). I would extend this to all predicates.
> For example: isSingleton(), isInfinite(), & the like.
Even NaI != NaI? But before defining comparisons on NaIs, one should
define comparisons on sets. There could be some similarities (e.g.
with the empty set).
> (5) That there is an empty set constructor.
This has nothing to do with NaI's.
--
Vincent Lefèvre <vincent@xxxxxxxxxx> - Web: <http://www.vinc17.org/>
100% accessible validated (X)HTML - Blog: <http://www.vinc17.org/blog/>
Work: CR INRIA - computer arithmetic / Arénaire project (LIP, ENS-Lyon)