unfill-paragraph(region) is different in GNU emacs and xemacs (was: unfill-paragraph (region) and indentation, )

[Uwe Brauer]

>>>>> "Stefan" == Stefan Monnier <monnier at iro.umontreal.ca> writes:

   > It depends on the definition of what is a paragraph, so it depends on
   > paragraph-start and paragraph-separate.

It should not since the unfilling function should (also) act on the region.

   > But FWIW, I can't reporduce the behavior you describe in
   > fundamental-mode in Emacs-23.1.

Right, as I will describe below indeed the basic unfill functions behave
*differently*  in GNU emacs (I used Emacs 22.3) and Xemacs (21.4.22 Mule), so
I will also CC to xemacs beta, which seems to be the better place to
continue the discussion.

   >         Stefan


   >    (let ((fill-column 9999999)) (fill-region (point) (mark)))

   > However Uwe's sample code uses the function "filladapt-mode", which I
   > don't seem to have, and can't seem to load by doing (load-library
   > "filladapt") or (load-library "filladapt-mode"); is it an external
   > library?  If so, maybe Uwe's problem lies there...

No it does not surprise, it is Xemacs filling function itself with
behaves different.

The following I did with 
emacs -q (GNU emacs 21.3)
and 
xemacs -vanilla (Xemacs 21.4.22 Mule)

In both cases I set the  fill-column to 10000000 and called.

GNU emacs (fill-paragraph)

Xemacs (fill-paragraph-or-region)

The example is (a buffer in fundamental mode)


See http://front.math.ucdavis.edu/ for a math specific front-end
	 or point your web browser at http://arxiv.org/ for the generic front-end.
	 To unsubscribe, e-mail To: math at arXiv.org, Subject: cancel
	 We consider ballistic aggregation equation for gases in which each particle
	 is either by its mass and impulsion or by its sole impulsion. For
	 the constant aggregation rate we prove existence of self-similar solutions as
	 well as convergence to the self-similarity for generic solutions. For some
	 classes of mass and/or impulsion dependent rates we are also able to estimate
	 the large time decay of some moments of generic solutions or to build some new
         classes of self-similar solutions.

A text which has indentation.

GNU emacs filled the line as expected Xemacs did not.

The behavior of both filling functions was identical if the text had no
indentation!

Could somebody from the Xemacs dev team comment on this?


thanks

Uwe Brauer