mode inheritence, cloning, hooks & OOP [was: Re: how to use cl-indent for CL and indent for EL files?]

[Richard Stallman]

The problem with defining a new major mode is that your new major mode
will in many cases not be recognized as a generalization of the mode
its based on because, e.g, many packages do tests like
(eq major-mode 'lisp-mode).

I prefer a different solution--avoid testing for specific major modes.
Instead, use local variables set by the major mode. This makes
inheritance possible between major modes.

[Richard Stallman]

While this provides a useful feature for major-mode inheritance, one
cannot help but question the design decision to extend the elisp
language using ad-hoc inheritance solutions when there are much better

I suspect the people who "cannot help but question" this are a fairly
small. There is no factual basis to claim that people in general, or
even programmers in general, will have this experience.

I installed derived.el because it does the job that actually needs
doing, and was a small amount of work.

Emacs Lisp is powerful enough. Adding OOP to Emacs is not clearly an
improvement; I used OOP when working on the Lisp Machine window
systems, and I disagree with the usual view that it is a superior way
to program. But even supposing that OOP would make Emacs better, it
is lower priority than many other changes (mostly in editing rather
than in Lisp).

[Marty Hall]

>>>>>> "WGD" == William G Dubuque <w...@martigny.ai.mit.edu> writes:
>
> WGD> Unfortunately RMS is of the opinion that OOP support is not
> WGD> needed in elisp. [...]

Where did you get this info? I'm not contesting it, just curious.
I was under the impression that RMS was moving toward a Scheme-based
extension language (GUILE), and that this would eventually be used in
emacs as well as in other GNU SW. In such a case, there are good OO
Scheme packages already existing that could be dropped in.

- Marty
(proclaim '(inline skates))

[Greg Limes]

In article <3fj9o9$b...@news.CNRI.Reston.Va.US> bwa...@cnri.reston.va.us (Barry A. Warsaw) writes:
> >>>>> "RS" == Richard Stallman <r...@gnu.ai.mit.edu> writes:
>
> RS> I prefer a different solution--avoid testing for specific
> RS> major modes. Instead, use local variables set by the major
> RS> mode. This makes inheritance possible between major modes.
>
> I'm looking at some of the cases where I test for major-mode in
> cc-mode, and trying to think about how these tests would be rewritten
> using this solution. It would require a large number of local
> variables containing lisp code to be eval'd. Definitely not very
> elegant (or easily maintainable). Not that it couldn't be done, but
> it would be ugly.
>
> A good example is how things like [] and // are treated differently in
> c-mode, c++-mode, and objc-mode. Various tests and actions can't be
> condensed to just a single variable (i.e. regexp or such), but is much
> more procedural.

One possible solution, then, is to have simple variables
"c-mode-likep", "c++-mode-likep" and "objc-mode-likep" which not only
would eval to t for their specific major modes, but also eval to t for
any newer major modes that descended from them.

Don't forget to do sane things when more than one of them is "t" ... ;-)
--
-- Greg Limes li...@3do.com, li...@netcom.com
Not speaking for my employer, of course
PGP key available on request