[std-interval] C++ committee feedback

[George Corliss]

Sylvain,

>> I would encourage you to send this report also to the reliable_computing
>> alias with a renewed invitation to subscribe here.
> 
> Yes, but I would wait a bit more.  I think that there are still a
> few big questions that need to be discussed here (and I would
> appreciate if new people give a fresh look at a more polished
> version of the proposal in some time):
> 
> 1- the set of functions which we want to propose for standardization.
> 2- the required precision for these functions.
> 3- the mathematical semantic of intervals (cset versus other models...).
> 4- how to deal with comparisons of intervals.
Good suggestion.  I think we benefit from as many participants in the
discussions as possible.  In particular, a friend notes that "the Germans"
have not been represented, although Juergen has.  I'd encourage sending the
meeting feedback and this agenda to reliable computing with a renewed
invitation.  You want discussion while the issues are debated, not after it
is too late.

Perhaps we should circulate the response John Pryce had sent you directly?
Many of his issues may be improved by further discussion, too.
> 
> I propose to begin by discussing the list of functions here.
> The current proposal does not include cos/sin/exp/log for example,
> which I think now is a mistake, but the question is to determine
> what set we want exactly.
> In a couple of days I expect to send a list of functions
> (after having a look at the C89 and C99 standards "math.h"),
> then proposing a list which matches the criteria of usefulness,
> implementation difficulty, completeness, and existing implementations.
> Does this sound OK ?
Yes.

Dr. George F. Corliss
Electrical and Computer Engineering
Marquette University
PO Box 1881 
1515 W. Wisconsin Ave.
Milwaukee WI 53201-1881 USA
414-288-6599; Fax: 288-5579; Dept. 288-6280
Office: Haggerty Engineering 296
George.Corliss at Marquette.edu


> 
> 
> BTW, so far we have 51 subscribers here, and we got more than 40
> answers to the questionnaire.