Should sm::range be more flexible than a closed interval?

[sebsjames]

Started life as "Consider ability to initialize sm::rand_uniform using sm::range as constructor arg".

range specifies a closed interval; rand_uniform sets a semi-open interval. Should range have the option for each end of the range to be specified as closed/open?

[sebsjames]

This raises the question of whether sm::range should really act like an interval, with a compile-time setting for the interpretation - is it closed, semi-open, open? etc.

[sebsjames]

For range to have closed and open intervals, many of the functions would require extending. Lots of <= and >= would need to be modified at compile time based on template parameters.

Maybe it would be better to create sm::interval, to be an interval class?

[sebsjames]

Maybe templated versions of the member methods to interpret as closed, semi open etc?

[sebsjames]

There's progress on this in dev/range_as_interval