X-VM-v5-Data: ([nil nil nil nil nil nil nil t nil]
	[nil nil nil nil nil nil nil nil nil nil nil nil nil nil nil nil])
Date:         Wed, 18 Apr 90 14:33:06 MET
Reply-To: LaTeX-L Mailing list <LATEX-L%DHDURZ1@unido.bitnet>
From: Victor Eijkhout <U641000@HNYKUN11.bitnet>
Subject:      Aux file blues.
To: Rainer Schoepf <BK4@DHDURZ1.uucp>
Status: R
X-Status: 
X-Keywords:                  
X-UID: 78

Hello,

I would like to raise a new topic:

            Using or Abusing the AUX File.

Small introduction.
I have written replacement styles for the standard LaTeX document
styles (check out the TeX-nl FILELIST of the LISTSERV@HEARN.BITNET),
and in some of these I implemented the following layout:

  1.     Section
  1.1.   Sub
  1.1.1. Subsub

  Paragraph.....
         Next paragraph is
  rather indented....

Now the problem arises: I take a certain parindent that is sufficient
up till 9.9.9., after that I'm in trouble. Solution taken:
everytime a heading is set I measure the label width (including 1em
or so), and if at the end of the run the original parindent
has been changed I (want to!) tell this to the next run.

Now for the real problem: I can write an instruction to the
aux-file that will set the parindent in the next run, but the
aux-file is read inside a group. So I will have to \global
the assignment. However (now comes the fun part!),
the Aux is also read at the end of the run, and then
there is a line in \enddocument which says:
\def\global\@namedef#1#2{}
Raaaaaaaaaaaaaaaggggggggggggggghhhhhhhhhhhhh!!!!!!!!!!!!!!

So I have my own version of \enddocument which simply
leaves out this line, and it doesn't seem to hurt anyone.

Should we discuss to what extent this sort of use
of the aux file should be permitted? Would it be appropriate
to support such activity?

Should hooks for this sort of thing be provided? Certainly
this would come in handy, because now the people who use
my styles run into (very occasional) problems when they
use the old 'article' etc. styles: the aux-file contains
an instruction that is undefined.

Give it a thought or two.

(remark aside: Frank, the discussion on this list does feel
a bit strange. After all, we can discuss all we like, you are still
the sole implementor. Maybe if you give some more hints of
your decision processes that arguments will flow more freely.)


Victor Eijkhout                 Department of Mathematics
                                University of Nijmegen
                                Toernooiveld 5
  "Far out in the uncharted     6525 ED Nijmegen, the Netherlands
   backwaters of the unfash-
   ionable end of the           +31 80  61 3169 (switchboard: 61 1111)
   western spiral arm           fax: +31 80 55 3450
   of the galaxy"
                                u641000@HNYKUN11.BITNET