Re: Issue 1294 in lilypond: Version 2.13.35 does not properly represent lyric tie

[lilypond]

Comment #7 on issue 1294 by pnorcks: Version 2.13.35 does not properly  
represent lyric tie
http://code.google.com/p/lilypond/issues/detail?id=1294

The regression test "lyric-tie.ly" also has this problem, as you can see on  
the 2.13 regtest page.

Some observations:

- In the GUB regtest tarball for 2.13.35, I can see that lyric-tie.ly loads  
a Japanese font, "Sazanami-Mincho-Regular", to use for the lyric tie.  Just  
look in lyric-tie.eps.

- Testing lyric-tie.ly locally, I installed this particular Japanese font,  
compiled the regtest, and compared my generated EPS file with that from the  
regtest tarball.  They are *identical*, with a negligible bounding-box  
difference, but a valid PNG is generated for me, whereas the PNG generated  
for the docs is garbled (has small black squares instead of U+203F  
glyphs).  See the attached image.

Maybe this is a problem with Ghostscript in GUB?  Not sure.

Attachments:
        lyric-tie.preview.png  330 bytes