mailing-list for TeXmacs Users

[Michael John Downes <address@hidden>]

Joris van der Hoeven <address@hidden> writes:

> I don't agree and I discussed this issue with other TeX/LaTeX specialists:
> the "shoving in" technique of TeXmacs is a welcome typesetting
> innovation.

It has been called "skyline fitting" or something like that in some past
discussions about possible improvements for TeX.

> When their is nothing below the fraction, there is no reason to descend
> the second line more than needed. I do not see what you mean by 
> "unbalanced".
> In my opinion, one should stick to a constant line separation whenever
> possible, that is, whenever there is sufficient room between the ink
> of successive lines.

There are two opposed constraints.

1. Use enough space between two lines of complicated math so that the
reader's eye can easily avoid confusion.

2. Don't use so much space that your book ends up taking 10% more paper
when printed and therefore costing 10% more to all the readers.

In making the typesetting rules of TeX, Knuth spent quite a bit of time
studying the line spacing used by mathematical publishers.
Traditional practice for displayed equations in professional publishing
has been to use more space between display math lines than for paragraph
lines. This is the difference in eqnarray that Phil Mendelsohn observed.
About three points of extra space (\jot in Plain TeX and LaTeX).

It was also normal for publishers to use a constant line separation for
paragraph text (not displayed math) partly because this tends to use
less paper but also because uneven line spacing of paragraph text simply
looks bad.

My organization in the 1980s was using a typesetting service that used a
constant line spacing: but collision between subscripts in one line and
superscripts in the next line was not automatically prevented, so the
collisions had to be fixed by hand whenever they occurred by putting in
some special coding that said "more space between these lines".

Fixing those collisions was the most common kind of correction in galley
proofs, and any changes in the text tended to create new collisions and
require additional corrections. It was worse than with TeX.

Now my organization uses TeX and instead of collisions there is extra
space added automatically between lines whenever the bounding box of the
*entire line* (or to be more accurate, bottom part of first line and top
part of second line) does not allow them to be placed at the normal line
spacing. In the old days this would have been fixed by hand, now we say
"let it go because fixing it would cost too much". So instead we use a
little more paper.

Good skyline fitting by TeXmacs would be better than both of the other,
inferior solutions.

Phil Mendelsohn wrote:

> Even then, it appears that we're a little too stingy with interline
> spacing.  I'd like to see about 5% more in _all_ of TeXmacs, I think.
> It is a small detail, I know, but one that gives me pause when using
> TeXmacs to generate finished products.

Be careful though. Do you still think the same if your desire is
reworded in other terms: "I would like to pay 5% more for all my books
and journals" ? :-) The compromise between clarity and cost that has
been worked out over many years by commercial publishers need not be
regarded as beyond question, but neither should it be dismissed without
first taking a hard look at the tradeoffs.

Regards, Michael Downes
American Mathematical Society