RE: Local Variables difference between 21.4.8 and 21.4.9/10
I dispute that. We have _always_ reinitted the local variables on
buffer reversion. I think it is very reasonable to suppose that in
We can still do this. Buffer-local variable changes count as a "change" to
the buffer so the optimization should probably not kick in. The quick fix
would be turn off the optimization for interactive. The better fix would be
to do a full revert when buffer-locals have changed (can we detect this?) Or
alternatively always reinit the buffer-locals. Maybe this would be better?
andy