Re: save-excursion and the mark

[Stefan Monnier]

> That's true, if it is _never_ used that way.  But it is hard to verify
> _never_, and we can expect some things to break.

Definitely.  This said, the same holds for changing the function
slot of symbols so that they can't be unbound anymore (and use Qnil
instead of Qunbound).  Yet when I did that in 24.4 noone noticed.

> I think the principle of avoiding incompatible changes is enough
> reason to pay the small price of adding one more very
> simple construct.

Then again, the benefit of adding a new construct is rather minimal,


        Stefan