Re: PDF size

[Harald Wellmann]

Thanks, that was it!

Still, since the lilypond installer already contains a patched gs, there 
had to some other problem.

As it turned out, the method lilypond uses to locate its gs binary is not 
fool-proof: in my case, the lilypond 2.8.1 installation picked up the old 
gs 8.15 binary from the 2.6.3.1 installation.

This is probably due to the following definition in 
/usr/share/lilypond/current/scm/backend-library.scm

(define-public (search-gs)
  (search-executable '("gs-nox" "gs-8.15" "gs")))

lilypond 2.8.1 does not have gs-nox, but lilypond 2.6.3.1 does.

The old installation path was still in my PATH, so even though lilypond 
2.8.1 prepends its own path to the PATH, the gs search first finds the old 
gs-nox.

By the way, "gs-8.15" does not really make sense here when the gs version 
bundled with lilypond is 8.50.

On the whole, since lilypond comes with so many required libraries and 
executables which are very likely to be already installed on the user's 
system in different versions, it is a bit dangerous to rely on the PATH or 
other enviroment variables. It would be safer to refer to all required 
binaries relative to the installation root.

Harald

Mats Bengtsson schrieb:
> I should have hinted you to also search the archives of
> bug-lilypond. For example, you will find:
> http://lists.gnu.org/archive/html/bug-lilypond/2006-03/msg00296.html
> which tells that the culprit is Ghostscript and how to
> fix it (if you're prepared to recompile Ghostscript).
>
>   /Mats