This is probably another bug from Andy's "optimization"
of buffer
reversion. If so, it is at least the third, maybe fourth, I can
remember. I think we should revert Andy's patch.
And I fixed all of them
What do you think, Vin?
I don't think you should. Its also not clear to me why my patch is to blame
here. font-lock doesn't get run if the reversion is "optimized", otherwise
the normal rules apply.
andy