On Oct 29, 2008, at 10:51, Dario Laera wrote:
Il giorno 28/ott/08, alle ore 18:00, Andreas Delmelle ha scritto:
Have you investigated whether the column-balancing approach is of
influence here? I mean: are there significant differences between
using one narrower column (the same width you would get when using
two columns, only it will roughly yield double the amount of pages).
A quick test haven't raised any difference except the fact that 2
columns used slightly less memory, but I think it's not related to
breaking as the number of active nodes is the same.
OK. The fact that less memory is needed for 2 columns, can probably
be explained by the number of pages being roughly doubled.
As Vincent pointed out, the main cause is then the fact that (3 *
space-width) represents a bigger portion of short lines. Basing that
width on the actual line-width indeed sounds like the right idea to
begin with.
If no one beats me to it, I'll have a look at changing that one of
the coming days.
Cheers
Andreas