The Duphly piece I mentioned in my last posting raised another interesting
point. I told PMX to use 3 pages and 15 systems, and here's what came out:
http://icking-music-archive.sunsite.dk/software/duphly/duph3.pdf

Look at the line on p.3 starting with bar 79. Pretty ugly, isn't it? But
this IS what PMX's basic algorithm comes up with. The easy fix would be to
go to 4 pages and 16 systems. But there's another option, i.e., trying to
find a better set of linebreaks. There's an undocumented command-line option
"-o" in PMX to do just that. It activates an iterative optimization scheme
that tries to find the set of line breaks that leads to the least variation
in the size of \elemskip among the systems. When I ran it on this example,
still with 15 systems, it gave a >50% reduction in the root-mean-square
variation of \elemskip:

http://icking-music-archive.sunsite.dk/software/duphly/duphopt.pdf

The option probably won't work if you have any forced linebreaks, which
therefore means no movement breaks either. And there's also no way to
preserve the new linebreaks for future runs, short of re-running the
optimization or manually inserting them as forced linebreaks. In other
words, it's not ready for prime-time, and at the moment I don't intend to
upgrade it. But feel free to play with it as is.

Another observation: When I use NS 4.7 with the acrobat plugin to view the
above links, they look different (worse) than when I use NS to view the
local copies of the SAME files. You can make the same comparison if you
download and save the file. Any idea what's going on?

--Don Simons

_______________________________________________
TeX-music mailing list
[EMAIL PROTECTED]
http://sunsite.dk/mailman/listinfo/tex-music

Reply via email to