Re: Ghostscript/GhostPDL 9.22 Release Candidate 1

Hi everybody!

The removal of the PDFDontUseFontObjectNum option in ghostscript caused a lot of headache during the last days.

Lilypond provides since early 2015 a way to persuade ghostscript to remove duplicated fonts from pdfs.
Later the extractpdfmark program was introduced to allow links to other pdfs to survive this process.
Also the -dgs-never-embed-fonts was introduced to force ghostscript never to embed fonts.
We also have a -dgs-load-fonts option.

All those options influence important details of our postscript code generation.

For ghostscript versions 9.16 and older --bigpdfs might be used with or without the -dgs-never-embed-fonts.

For ghostscripts versions 9.17 until the removal of ghostscripts PDFDontUseFontObjectNum, --bigpdfs might successfully be used with or without -dgs-never-embed-fonts as long as PDFDontUseFontObjectNum is given to ghostscript.

For a recent ghostscript without the PDFDontUseFontObjectNum only the combination of lilyponds --bigpdfs with -dgs-never-embed-fonts gives the desired result. The necessity for the extractpdfmark utility has not changed - use it if you link to destinations in other pdf files. Unfortunately our 
lilypond build system missed to use the --bigpdfs option together with the -dgs-never-embed-fonts.

If other people agree that my patch <https://codereview.appspot.com/325630043/> solves the problem of exorbitant file sizes, we might live well without ghostscripts PDFDontUseFontObjectNum.

Knut