Re: Emacs vs. TextMate (not trying to start an editor war)

[Lennart Borgman (gmail)]

Tom Tromey wrote:
> > > > > > "Miles" == Miles Bader <miles@gnu.org> writes:
>
> Miles> What is really desirable, I think, is some way of transparently fooling
> Miles> "whole buffer" modes into thinking they have control of a whole buffer,
> Miles> when in fact, they just have control of part of one.  I think it's
> Miles> probably possible to do this (perhaps using an extension of the indirect
> Miles> buffer mechanism?) but afaik nobody has really looked at the problem
> Miles> closely.
>
> Have you looked at MuMaMo?  I thought it did this kind of thing.  But,
> I have not looked at it and I don't know the details.


MuMaMo (which is part of nXhtml) does something like that, yes.

But it does this by trying to handle buffer local variables and 
fontification carefully. Doing it this way I could make it to work with 
Emacs 22. I think somewhere in the future with some more experience with 
this we may want to rewrite some structures in Emacs to better support this.

However my impression is that such structures should hide part of the 
buffer rather than just giving one continuous part of the buffer to the 
major mode. This will make it easier to handle major modes that parses 
the buffer.

Take for example XHTML buffers versus XHTML template buffers. For an 
XHTML buffer (ie containing an .xhtml file) the whole buffer can be 
parsed by nxml-mode (or a mode derived from it). For a template language 
the situation is different. In that case only (non-consecutive) parts 
can be parsed by nxml-mode.

This situation is today handled by MuMaMo + a patch to nxml-mode.

There are of course also other details to think of.