Brian Smith wrote: > I agree that @scheme seems pretty useless as it is defined in the > specification. Luckily, if you avoid using @scheme, and you use IRIs > for @term, you can avoid worrying about it.
Just to be sure we're clear, I DON'T think that it's useless -- I agree only that if the spec meant what you're saying it means, @scheme would be worthless.
But I think you're reading a different meaning into it than we intended when we created it. Which is more likely:
1) Everybody working on the spec completely missed the fact that @scheme was useless and the fact that the definition of @term said nothing about how it should be sufficient by itself to uniquely identify a category, and should be constructed in a way that prevented collisions (even though we put a lot of effort into working out how atom:id would act as a unique identifier for a feed or entry). And it didn't occur to anyone that since we wanted a globally unique identifier for a category, we could use atom:id instead of @term.
2) @term was intended to identify a category within a defined scheme which was indicated by @scheme, or within an ad hoc scheme when @scheme was absent.
Especially since I believe I remember the discussion running along the lines of #2, I find #1 highly implausible.
Antone
