On Tue, Dec 3, 2019 at 4:26 PM Stefan Monnier <monnier@iro.umontreal.ca>
wrote:

> I'd rather fix the code not to rely on the return value.

Agreed, but I wasn't talking specifically about this case. There are others
in the code,
I think. If we don't want to document the return value of push, we should
fix them all.

>     +                    (and class
>     +                         (progn (cl-pushnew class classes) t))))

Thinking about converting it to cl-pushnew is what made me look closer at
this code in
the first place ;-)