My first thought is that a variant script selector might be defined, which applies until cancelled or overridden, on the analogy of RLO...PDF. But I guess others will object to this. Does anyone have any other suggestions for how Unicode can support script variants, apart from encoding them as completely separate scripts?
Stateful controls do not belong in plain text. There are very few exceptions to that general rule, but variant indication is definitely not one of them.
I think it's up to you to develop a workable alternative to parallel coding, if you think that would be preferable.
As the discussion has gone on for a long time on the list and has involved only a small number of participants, I suggest that you contact the interested parties offline.
A./
