Re: Shrinking the C core

[Alan Mackenzie <acm@muc.de>]

Hello, Arthur.

On Wed, Sep 06, 2023 at 07:04:43 +0200, Arthur Miller wrote:
> Richard Stallman <rms@gnu.org> writes:

> > [[[ To any NSA and FBI agents reading my email: please consider    ]]]
> > [[[ whether defending the US Constitution against all enemies,     ]]]
> > [[[ foreign or domestic, requires you to follow Snowden's example. ]]]

> >   > Would you please stop arguing for rewriting Emacs in Common Lisp?  It
> >   > is a non-starter.

> >   > It would be an enormouse job -- including rewriting the Emacs Lisp
> >   > Referance Manual.

> > Also, there are aspects of Common Lisp which i rejected as clumsy and

> With all respect to you, but sound to me like an emotional argument, not
> a rational one.

It's a rational argument expressed in emotional terms for simplicity.

> I don't know why you reject them as clumsy, but why does it matter to us
> if a tool includes features we don't use?

It matters a very great deal.  In practice you cannot avoid "using" these
features if you have to understand or debug somebody else's code.

> I don't know if I understand it correctly, but CL was made so other
> Lisps can be abstracted on top of it; so we use those features to
> abstract stuff on top of those lengthy verbose functions? In other
> words we can hide those keyword params behind elispy abstractions if we
> don't like them?

That's complexity and bloat.  Far better not to have them in the first
place.

> > best avoided.

> Why are they best avoided? Is there some technical reason or is it
> psychological?

It's because they are not needed.  Bear in mind, Common Lisp is a massive
language, much like C++ or PL/1 or Algol-68.  With such a language,
nobody uses all of it (it's just too big), but everybody has her own
personal subset.  This creates difficulty when somebody else has to
understand that first hacker's code.

CL is a niche language; it has not captured the hacker mindset.  I think
it is just too big and too unwieldy.

> > best avoided.  All those keyword arguments!  I intentionally excluded
> > tham from Emacs and I am not going to let them in.

> I don't want to be impolite, but they are already in; via cl-lib.el.

Yes.  There was a time not so long ago when cl.el was banned from use in
our Lisp code, except for at compile time.  Our Emacs Lisp was small,
simple to understand, and easy to learn.  Now things in cl-lib.el get
used as if they are just a normal part of Emacs Lisp.  Our language is
thus MUCH more difficult to understand, perhaps by a factor of somewhere
between 3 and 10.  When perusing even established parts of Emacs I groan
inwardly every time I encounter one of these needless cl-lib features.
It stops me dead, forcing me to consult doc strings (which are often
missing and often inadequate even when they are present) or even manuals.

Were we to rewrite Emacs in Common Lisp, these things would get worse
fairly quickly.

-- 
Alan Mackenzie (Nuremberg, Germany).