> I considered the same thing :o) My only concern is that this sort of > nondeterminism isn't all that helpful to people writing process > handlers. I mean, having something in the docs like "your process > handler *might* be executed in the buffer the process was called from" > would mean that the caller would still have to check which buffer they > were in. > > I can't really think of a sensible default buffer... maybe we shouldn't > encourage people to assume their handler will be executed in any > particular buffer. Any ideas?
You may mean that if you want a buffer, create it in your code. I don't check all buffer usage in xtla, however, sometime we want to use existing buffer in our code in stead of creating. In other hand, if an user of xtla wants to kill the buffer, none can stop it. In such case what we can do is just warning like: (add-hook 'kill-buffer-hook 'tla-warning-dont-kill-buffer)
