bug#46819: 28.0.50; Undesireable warning: Empty let body from the `benchmark-run' implementation

[Matt Armstrong]

Lars Ingebrigtsen <larsi@gnus.org> writes:

> Stefan Monnier <monnier@iro.umontreal.ca> writes:
>
>> It'd be easy to put an explicit nil in the loop's body, but....
>>
>> The idea was that in actual (non-macroexpanded) source code, empty
>> let bodies are usually the result of a misplaced paren.  But it's
>> probably more often annoying than useful since we can't easily
>> distinguish a "let from source code" from a "let from a macro
>> expansion".
>>
>> So I'm also voting to revert that change.  Any objection?
>
> It is a nice and helpful warning, though, so if it could be fixed in
> another way, that'd be great.  (But I have no idea how, of course.)

It might be hard to fix in the general case.  Think about this example:

(let (...)
  (some-macro blah))

What if that macro expands to nothing, based on some criteria not
entirely under programmer control?  Is the code "wrong"?