RE: [patch] minor patch for register.el

[Drew Adams]

> > Personally, I don't really care one way or the other - (a) 
> > show a message and do nothing or (b) raise an error.
> 
> Raising an error would be tantamount to suggesting that it is user's
> responsibility to _know_ a priori that the set is empty and to avoid
> invoking the function.  Encountering a normal boundary state is not an
> error.
> 
> I run with debug-on-error set.  Raising errors for non-error
> conditions is very distracting.

Again, I don't care which we do here: message or error.  But if you are worried
about an error then you'd better look throughout register.el at the other
commands.  They pretty much all already raise errors for similar situations -
see the cl assertions.

Feel free to change the error raising in the patch to `message' calls, as far as
I am concerned.  Or use `user-error', as Christopher suggests.

The point, for me, is not to open an empty window/frame here: display a message
instead.