Ranges - Question about desing choices

[Michal Minich via Digitalmars-d-learn]

On Monday, 24 August 2015 at 15:16:12 UTC, Adam D. Ruppe wrote:

> One advantage of the current design is you can statically 
> determine if something is an infinite range by seeing if empty 
> is a constant false.

That is important aspect! By having this information at compile 
or runtime, you can enforce that algorithms which consume entire 
range, and are expected to finish, would not accept this range as 
an argument, for example saveToFile(randomNumbersRange).

I see two possible way to model this:

1) provide another property 'bool isInifinite'. This of course 
causes dependency on empty property: when isInfinte is false, 
empty also needs to be false. This can be complex to enforce.

2) instead of empty property, have state property, returning enum 
{ empty, full, infinite }. This might complicate algorithms 
implementation, and of course this enum could not be never 
extended, since it would be breaking change. For example if 
algorithm ask currently for !empty, it would ask for just state 
== full, so it might not complicate many algorithms

Any more insights? :)