Re: Elispers all around the world (was: Re: describe-mode "some-mode"?)

[Emanuel Berg]

Drew Adams <drew.adams@oracle.com> writes:

> All of this applies *IF* you care about affecting
> vanilla Emacs - see that qualification in the first
> line you quoted, above.

No, I don't really care if it is in the vanilla Emacs
or not. It it is good, it would make sense for it to
be. But as long as it is available and easily
integrated it isn't that important. Example: I use
Emacs-W3M and Gnus every day. Gnus is part of Emacs and
Emacs-W3M is not. That doesn't bother me, I can't say.
To me they are as much parts of Emacs. For other people
that might not find Emacs-W3M for this reason or might
run into problems installing/upgrading it, perhaps it
is a problem, I don't know. (Note: What I remember
installing Emacs-W3M is child's play.)

The way I would to it are along the lines:

1. You have a problem.

2. By specifying the problem, you are presented with
   the available solutions to that problem. Some of
   those solutions may be only partial matches, but
   they can be beneficial to you, still.

3. If you don't find a solution, you solve it, but it
   doesn't stop there: you submit your solution and
   describe formally or semi-formally what problem your
   solution aims to solve. So the next time someone
   faces this problem, your solution will be displayed
   at (2).

In time, it'll be clear that most problems only get one
or two solutions that are serious and have passed the
test of time, also receiving the famous distributed
peer-to-peer review to get the last details figured out
and perfected.

In even more time - yes, the most popular solutions
should (most often) be added to vanilla Emacs if they
solve problems that are general in nature. (But this
shouldn't be automatized, of course.)

As an experiment why such a system is beneficial, just
code whatever in Elisp. Then pick a line at random.
Pretend you didn't know how to do that, or that that
function didn't exist. Use Google to search for a
solution. Many times you get Common Lisp or even C
solutions! (And to use Google to find Elisp code -
brr!)

In the AI world, this type of system is called an
expert system. It is usually implemented as a finite
state machine with the transitions based on probability
theory.

-- 
underground experts united