This is longer than I wanted but has the benefit of defining a satisfiable ranges-specifier in general (instead of only for bytes requests), how to handle an invalid ranges-specifier as a bad request, and defines which code to use for an unsupported range-unit.
I believe all of this was implied by the existing text and how we have used valid/invalid, but I think it is clearer stated this way instead of the prior generalizations.
YMMV. This probably should be reviewed by the WG (just to be sure).
Fixes #1011
This is longer than I wanted but has the benefit of defining a satisfiable ranges-specifier in general (instead of only for bytes requests), how to handle an invalid ranges-specifier as a bad request, and defines which code to use for an unsupported range-unit.
I believe all of this was implied by the existing text and how we have used valid/invalid, but I think it is clearer stated this way instead of the prior generalizations.
YMMV. This probably should be reviewed by the WG (just to be sure).