[texhax] using larger units to determine breaks?

[William Adams]

On May 5, 2009, at 6:59 PM, D. R. Evans wrote:

> I know that the original TeX basically works in paragraphs and  
> outputs in
> pages.
>
> But the more (fiction) works that I typeset, the more it seems to me  
> that a
> considerable improvement -- or at least less manual intervention --  
> would
> likely be effected if the algorithms optimized page breaks over an  
> entire
> section or chapter.
>
> Are there any versions of *TeX that have or will at some point have  
> the
> capability to optimize breaks over larger units than does the  
> original TeX?
>
>  Doc
>
> PS Maybe I'm wrong and in practice this doesn't help, but it sure  
> seems
> like it should. A lot of the tweaking I end up doing seems like it  
> would go
> away if some attempt had been made to optimize over larger units.


Dr. Jonathan Fine has done some work along those lines, and of course  
LaTeX does have a very sophisticated algorithm for determining figure  
placement (not familiar w/ ConTeXt), but tweaking the page breaks  
pretty much falls into the last n% of effort which requires human  
intervention.

The problem is solutions tend to be intricate and inter-connected and  
I've occasionally found myself going back to the beginning of a  
chapter to make the last page work out properly.

Lessee, one has a 100,000 word novel w/ 20 chapters, so average 5,000  
words per chapter, each paragraph has on average of five sentences of  
five words each, so 200 paragraphs in a chapter, figure 50 characters  
(a bit long, but I'm trying to make the math easy) per line, and 10  
characters per word (ditto) so 1,000 lines, figure 25 lines per page,  
so 40 pages, so one has to work out how 200 paragraphs of more-or-less  
1,000 lines inter-act w/ 40 page breaks --- I'll leave it to someone  
who does combinatorial mathematics to work out how many possibilities  
there are if one-fourth the paragraphs can gain a line, one-fourth can  
lose a line, one-fourth gain _or_ lose a line and the balance are  
inflexible and unyielding and won't change their line lengths w/  
reasonable type settings.

It's probably w/in reach of modern equipment, but it won't be a fast  
typesetting run.

William

-- 
William Adams
senior graphic designer
Fry Communications
Sphinx of black quartz, judge my vow.