Re: [MMM] CVS Edition Troubles

[Michael A. S. <ms...@ca...>]

On Fri, Mar 14, 2003 at 04:36:27PM -0800, Charlie Reiman wrote:
> Having said that, let me say I've never successfully used MMM mode. I've
> tried but never been very happy. I suppose this is because I'm using
> HeathenEmacs rather than the One True Emacs, where MMM seems to be best
> supported.

I don't have anything against XEmacs in particular.  But it is quite
true that Emacs is best supported in FSF Emacs, because that is what I
use myself and am most familiar with.  I've done my best in the past
to try to make it work under XEmacs, and I will continue to do so, but
XEmacs users need to be aware that sometimes my best is less than what
might be hoped for, and that XEmacs support is not a top priority for
me.

I would love it if anyone who uses XEmacs and is familiar with (or
would like to become familiar with) the idiosyncracies of its version
of elisp would offer to take charge of XEmacs compatibility, and I
would give all the help I can.

> I decided to try the latest CVS version (0.4.7a?) under XEmacs 21.4.11 and
> I've hit a few snags. First, the FAQ mentions using configure to build the
> system. There is no configure script in the CVS edition.

Ahh.  This should be documented, possibly in a README.cvs or some
such, and/or in the FAQ.  The script "configure" is not in CVS because
it is automatically generated by automake/autoconf from configure.in.
In order to get a configure script, you need to run automake and then
autoconf.

> Being a savvy user, I figured I could just bytecompile the .el files
> on my own and go. 

This ought to work perfectly well, however.  There is actually no
need to byte-compile the files at all; it only makes things run a
little faster.  For debugging purposes, in fact, the backtraces are
sometimes more helpful when things are _not_ byte-compiled, because
then the functions are visible instead of showing up as "compiled
function."

> So I did. When I start mmm-mode, I get this message (about 90%
> of the time):
> 
> Args out of range: #<buffer ".zprofile">, 0, 2

Hmm... I think see what the problem might be.  Apparently XEmacs
starts numbering characters in a buffer from 1, and isn't able to
figure out that when we ask for the extents between 0 and 2, we mean
between 1 and 2.  I'll see if I can do something about this in the
next couple days, and then you can let me know if that fixes it.  I'm
a little worried about your comments on the inconsistency of this
problem, since if this is all it is, it ought to happen all the time,
but maybe when I take a closer look at the code it will become
clearer.

-- 
Michael A. Shulman
http://www.ugcs.net/~shulman

Come a stove boat and stove body when they will, for stave my soul,
Jove himself cannot.
        -- Herman Melville, Moby-Dick