Re: Wizards (was: Re: Oort Version)

[Per Abrahamsen <abraham@dina.kvl.dk>]

Karl Kleinpaste <karl@charcoal.com> writes:

> Do you still have the code?  It would be worth looking at today.  Rot
> can be cut out.

It may be in W3.  W3 is build upon the same widget library as
customize, and the primary idea was to have a html tag that expanded
into a customize widget, and the secondary idea was to have Emacs Lisp
as an HTML script language (instead of ECMAscript)...

Hey, I actually have a demo-page from back then!  Look at 

        http://www.dina.kvl.dk/~abraham/custom/gnus.html

with w3 to get an idea of what I was trying to achieve.

W3 *is* FSF signed, but I don't think anyone is working on integrating
it.  W3 need to be a "core feature" of both Emacs and XEmacs for this
approach to be feasible, simple minded customization should not
require any extra packages.

The alternative is to write an ad-hoc markup langauge, which I think
is silly.  I'd much rather have W3 used everywhere. 

> Do you mean by using a long (contorted?) list of alternate selectors?
> That could get really hairy, I would think.  But it would be good for
> a first pass.

I don't understand what you mean, but then again, I have never used
the : or ! features.  It was just a thought.

Maybe the best option would be to drop the fancy split, and write a
third value for nnmail-split-methods, say nnmail-split-common, which
understood some not very flexible but easy to understand options.

The default syntax is very simple, too simple for common tasks.

The recursive nature of nnmail-split-fancy is way cool and very
powerful if you have a CS degree, and utter gibberish if not.

The nnmail-split-common should understand some simple lists
corresponding to common splits.  Maybe junk, people, lists, and other.

> > A problem is that BBDB is FSF code, which makes it problematic to
> ( s/is/isn't/ ?)

Yes.

> So let's provide BBDB auto-graft iff we detect that BBDB is
> available.  Can "require" be wrapped inside "ignore-errors"?

Yes.  It doesn't really interest me though, the vast majorioty of
ordinary users just use what is bundled.

> I wish I had any experience using VM, but I've never once invoked it
> except by accident.

It is a long time since I used VM, but Kyle Jones write beautiful code
with nice and clear apis.  I just wish he would agree to sign papers.

> But how does this interact with users who run >1 simultaneous Emacs?

Badly.