GNU Emacs Developement Inefficiency (commentary)

[Tom Lord]

On Jul 17, 11:11 am, Emmy Noether <emmynoeth... at gmail.com> wrote:

> Well, there is a lot of resistance from the emacs community in sharing
> information. Richard Stallman is a true STALLER of progress. He has
> held the whole process hostage by not sharing information. He has
> RENEGED on his promise to make it truly open by suppressing
> documentation of his softwares.

I used to think similarly but differently.  In my case
I thought of him as a "STALLER" for not more aggressively
changing Emacs to keep up with GUIs, with better languages
than elisp, with better ways to implement text buffers,
and so forth.

One day I woke up and realized: hey, I've been using this same
damn program for 15 years (and now, today, more than 20).  I
use it every day.   I'm annoyed on systems that lack it.  It
satisfies me, as a user, in ways that almost no other program
I've used for a long time does.

Aha - I realized.  He must be doing something right.

And it became way more interesting to try to understand
what he's doing right than to focus exclusively on what he's
doing wrong.

On the documentation thing ... well:

Stallman's code (and code descended from Stallman's code)
tends to annoy me with its sparcity of good commenting,
its long, convoluted functions, its sometimes wacky choices
of how to structure module dependencies,  its horrific approach
to portability.....

Problem is, though, than when I've had to actually work
on any of that code?  Yeah, it takes some hours or days
of study to puzzle it out - weeks even - but then once
you've "got it" you've got it.   It's actually fairly lucid
in spite of the style.

One of the things he does right is not overly documenting
the code.   First, it's a lot of work.   Second, the folks
that wind up contributing the most to it learn a lot and learn
it well by puzzling it out a bit first on their own.
Yes, really.   I'm not suggesting weak internals documentation
and all the other aspects as a general style that everyone
should adapt.   I do think it slows some things down.
Yet in some cases, like Emacs, .... well, it's hard to argue
with success, ain't it?





> All Richard Stallman has to do is to hand draw the data-structures and
> architecture of the various programs. Give references to the places
> where he got the ideas.

Isn't that (the drawing) something you could do?

As for where he got the ideas: a lot of Emacs architecture is
based on common knowledge ideas, some on the order of 40 years
old.   There's not a lot that isn't original in the core
architecture that one would normally want cites about.