>Maybe there's an easy solution to this. Observe that in all cases where
>combine="group" is allowed by this proposal, it has exactly the same
>semantics as combine="interleave". So we could save ourselves a bunch of
>complexity in the spec and in implementations without forgoing any
>functionality just by getting rid of combine="group" altogether.
Yes, this would work.
>A grammar-level include would also need to be able to contain a <start>
>element to replace any <start> element from the include.
Yes.
Cheers,
Makoto
