Motion P1788/M0034.01 Notation: YES
My vote on Motion M0034.01 Notation is YES,
provided that it is consistent with "I" being an oprator.
-Dima
----- Исходное сообщение -----
От: eliasen@xxxxxxxxxxxxxx
Кому: stds-1788@xxxxxxxxxxxxxxxxx
Отправленные: Среда, 6 Июнь 2012 г 12:12:42 GMT +04:00 Абу-Даби, Маскат
Тема: Motion P1788/M0033.01 Number Format:NO
My vote on Motion M0033.01 Number Format is NO.
Currently, the text of the motion is ambiguous or impossible to meet
if the implementation of the numerical type uses arbitrary-precision
numbers. (And good numerical implementations will have arbitrary
precision numbers.)
The section A5 says:
===============begin quote======================
A5. An element of Rbar (Level 1) is mapped (rounded) to an element
of F (Level 2) according to the following rules:
* A constraint can be given on the rounding direction. It must
be satisfied.
* Following this constraint, the rounded result should be an
element of F that is the closest to the exact result (with
a special implementation-defined rule for the distance to
an infinity).
==============end quote=========================
"The closest to the exact result" is ambiguous or impossible or
wasteful to implement. For example, what is the closest result to
1.0/3.0 , rounded up? Is it 0.33333333333334, or
0.33333333333333333333333333333333333333333334, or the same result with
2 billion digits which is as close as my language's implementation might
get?
The standard always needs to allow superior numerical
implementations; that is those which contain exact rational numbers,
arbitrary-size integers, and arbitrary-precision floating-point numbers.
In order for me to vote YES on this motion, it would need to be
amended to allow a constraint on the *precision* as well as the rounding
direction, and remove text about "closest" value. For example, the
constraint might require that we get the closest value within 20 decimal
digits, or similar. The rounded result should be an element of F that
is rounded in the appropriate direction, with an attempt made to meet
the precision constraints. If the exact precision constraints are not
meetable for any reason, then the implementation should be allowed to
return a less "sharp" value that still preserves containment.
Throughout all of these discussions, we always need to consider
arbitrary-precision implementations, systems that preserve exact
rational numbers, etc.
--
Alan Eliasen
eliasen@xxxxxxxxxxxxxx
http://futureboy.us/