Re: false warnings when compiling with lexical-binding and cl-lib.

[Stefan Monnier <monnier@iro.umontreal.ca>]

>> Also the warning is only at compile time, should we return an error on
>> evaluation ?

I don't see why we should bother with an error.

   (let* ((a 1) (a 2)) ...)

is not very useful, but it's not an error.  The warning we currently
have is not perfect because it applies to the post-macroexpansion code,
so the user may not immediately understand what's going on, but your
suggested patch has the following downsides:
- it breaks previously working code.
- it does not tell you where (approximate line number) the error is located.
So if we want to improve on the current code, I think rather than
`cl-assert' we should use macroexp--warn-and-return (grep for it to see
sample uses).  But the patch won't be as trivial.


        Stefan