Cliff Miller wrote:
>
> the non-trivial case here is when -b is present in the form
>
> sort -k2.1b,2.3
>
> where the start and end positions could be in either order depending
> on the number of blanks that start field 2.
Right, good point.
I've added a test for that case also and pushed the result.
cheers,
Pádraig.