Assorted problems with some machine-generated Lout

[Ian Jackson]

I've been working on some manuals for Debian (the Linux distribution).
The manuals are in an SGML DTD mainly of my own devising and are
converted to Lout for PostScript output.  Because of this I'm not
really able to use Lout's built-in chapters, sections, subsections,
&c.  This isn't a bad problem, because my Lout-generating SGML backend
can generate the toc and headings itself.  However, it does seem to
have had some bad side-effects.

1. I have to use address@hidden { Yes }' to get reasonable-looking
output (for example, to avoid page breaks just after headings) - which
is fine by me.  However, this breaks the table of contents into two
pieces for no reason that I can easily see.  @OptimizePages claims to
use TeX's breaking algorithm, but that algorithm is based on having
costs associated with page breaks in particular places.  How does
@OptimizePages assign these costs, and is there any way to change
them ?

2. I want to get my own section headings into the running headers and
footers.  I can't figure out how to do this without using the standard
@Section symbols and so forth - which I don't want to do because
(amongst other things) I need subsubsubsections and generally want
more control over the formatting, numbering and so forth.  I tried
using a symbol @Runner which I found in various of the standard Lout
include files but didn't have much luck.

3. For no readily apparent reason occasionally the last line of a
paragraph will get right-justified.  Eg,
 The Priority and Section control file fields give
 information for classifying the package in dselect and say
 which directory to place it in the FTP archive.
is rendered as
 The Priority and Section control [deleted to get <80 - iwj] in dselect
 and   say  which   directory   to   place  it   in  the  FTP  archive.
It actually looks a lot worse in the PostScript.  There have been even
worse examples, with the three words on the last line of a paragraph
spread across the page, but the problem is intermitted and sensitively
dependent on the input file.  This may have something to do with the
fact that many words and phrases in the document are marked up with
 {{0.7 1.0} @Scale {Courier Bold} @Font {whatever}}
to get look that I like for `typewriter style' output.

I've put all of the Lout-related files I'm using up for anonymous FTP
on chiark.chu.cam.ac.uk in /users/ijackson/lout.  The
automatically-generated Lout input file is problems.lout, and the
other files are the result of running
 lout problems.lout >problems.ps
three times one after the other in a fresh directory.  The example of
a badly formatted paragraph that I mention above is on page 6 of the
PostScript file.

I'm using the version of Lout 3.08 that I built for Debian myself.

Ian.