On Thu, 19 Nov 2020 at 14:51, Stefan Kangas <stefan@marxist.se> wrote:

> Neil Jerram <neiljerram@gmail.com> writes:
>
> >> I've been working on removing redundant `function' around `lambda' in
> >> Emacs core,
> >
> > I'm slightly curious about the history and reasoning around this.  If I
> > understand correctly, (lambda ...) on its own has always worked, and it's
> > never been strictly necessary to add (quote ...) or (function ...) around
> > it.  Then sometime (Emacs 19 or later, I think) it started being
> > recommended to use (function ...).
> >
> > Do you know why that recommendation started, and should I understand that
> > the reasoning for it has now evaporated?
>
> Correct, there is no reason to do this.
>
> I don't know the history here, and there are people on emacs-devel that
> would know better.
>
> I _suspect_ that the byte-compiler first got the capability to optimize
> calls to anonymous functions, but that it required to explicitly marked
> as such with `function'.  Later, it grew the capability to recognize
> lambda as such automatically.  But I don't know if that is correct; it's
> just a guess.  In any case, they are no longer needed as lambda and
> lambda+function are equivalent.
>
> (Note that the worst thing here is to do `(quote (lambda ...))' as that
> defeats byte-compiler optimizations altogether.)
>

Many thanks Stefan!