Defining "common interval literal"
Why should a *common* bare interval literal just be one of the forms defined in 9.5.4, i.e. inf-sup & uncertain form?
This goes back to Dmitry's original text of 20150222:
> Each flavor defines a set of portable bare and decorated interval literals. Portable interval literals shall be valid interval literals in each implementation of the flavor. An implementation may support a more general form ...
>
> This subclause defines common bare and decorated interval literals. Their values are common bare or decorated intervals. They shall be portable interval literals in all flavors.
I agree with the substance, but am unhappy about the use of language. Let s be a literal, whether number, bare interval or decorated interval.
- At present "s is common" means s is understood and denotes same value
in all flavors. Also the definition is phrased so that when s denotes
an *interval* x, this x is a common interval.
- "s is portable" means s is understood and denotes same value in all
implementations of a given flavor.
A flavor might define, the string s="[pi]" to denote the one point interval containing pi, which translates to a tight enclosure of this at Level 2. This s denotes a common interval. But at present we can't call it a common interval literal, even one "of that flavor"?
I think this could be confusing to users and will welcome suggestions to revise the naming.
John Pryce