Re: Range excluding 0.0

[Robert A Duff <bobduff@shell01.TheWorld.com>]

csampson@inetworld.net (Charles H. Sampson) writes:

> vlc <just.another.spam.account@googlemail.com> wrote:
>> 
>> I would like to declare a subtype of Float which ranges from 0.0
>> (excluded) to Float'Last. Is there a way to exclude the 0.0 from the
>> range?
>
>      You've already got the answer you need, but your question brings up
> another that I've recently been puzzled about.  Why doesn't Ada allow
> all four flavors of ranges commonly used by mathematicians (and others):
> individually including or excluding the endpoints?

For that matter, why doesn't Ada allow arbitrary subsets?
For example, "an integer in the range -1_000_000 to 1_000_000,
except 0 is not allowed".  Or, "an integer for which Is_Good
returns True".

ARG is thinking about adding type invariants and/or user-defined
constraints, which could express such things.

>...I've been around Ada
> since close to the beginning and I don't recall that this was ever
> discussed.  At the same time, I have this vague memory that one of the
> candidate languages might have had it, but I can't remember which one it
> might have been.
>
>      This is not just theoretical musing.  I currently work in the world
> of compasses, so a range of the form [0.0 .. 360.0) is quite natural.
> (That notation is neither mathematical nor Ada.  I hope its meaning is
> clear.)

It's certainly not Ada, but why do you say it's not mathematical?
I find it ugly, but I've certainly seen in maths textbooks.
Ignoring the ugly syntax, the concept seems quite useful.

- Bob