Re: enriched-mode and switching major modes.

[Oliver Scholz]

address@hidden (Kim F. Storm) writes:

> Some of the existing hook properties may also be used to handle
> "delete hard newline betwee paragraphs with different categories" in
> some sensible way.
>
> If nothing else, such hooks could set some global value which causes a
> post-command-hook to fix whatever conflicts are created by the last
> modification -- at least I think that would be more efficient that
> relying on jit-lock to run "all the time".

I agree that it would work.  But why is it better than the solution
that I listed as #2, i.e. letting a special fill-function handle all
whitespace formatting issues?  I have experimented with this and it
works nicely.  It can also handle the problem of putting whitespace of
a specific height before and after a paragraph rather elegantly.

The way I see it we first create two different semantics for
paragraphs: one via the `category' property, one via hard new lines.
And then we take special pain to keep both representations in sync.
To me this seems like asking for trouble.

The only benefit I can see, is that `forward-paragraph' and its like
would DTRT.  But writing a version that works correctly for the other
solution and binding it to the appropriate keys is rather trivial.


...!


Oh, no!  I see now, this is a way to make a slightly modified
`fill-paragraph' work.  Some of my major concerns could be adressed
this way and /technically/ it could be handled by a minor mode.
(Though I don't see what good that should be.  How a paragraph should
be indented in an RTF document depends on its properties, not on
paragraph-indent-text-mode nor on paragraph-indent-minor-mode.)  It
could be a way to provide a user interface for RTF where I can type
four spaces to indent a paragraph like in text mode.

I don't like the resulting user interface.  I don't like it at all.
In fact I find it really, really horrible.

Would it be possible to handle XML with its tree-like structure this
way?  My own thoughts have let into an entirely different direction
(not explained in this thread); offhand I don't see how nested block
boxes would be possible with hard newline semantics.


    Oliver
-- 
Oliver Scholz               Jour des Récompenses de l'Année 212 de la Révolution
Ostendstr. 61               Liberté, Egalité, Fraternité!
60314 Frankfurt a. M.