Re: LilyPond 2.13.10 (OSX) Font Issues

[Patrick McCarty]

On 2010-01-20, Keaton Mowery wrote:
> Okay, I think I've tracked it down... Whew, autotools is a mess.
> Anyways, here's the short story:
> 
> python-config reports information about the local python install.
> Lilypond's configure scripts already use python-config, but in a
> slightly different manner than my original hack (which is now
> obsolete).
> 
> On OSX, "python-config --cflags" returns
> "-I/System/Library/Frameworks/Python.framework/Versions/2.6/include/python2.6
> -I/System/Library/Frameworks/Python.framework/Versions/2.6/include/python2.6
> -fno-strict-aliasing -fno-common -dynamic -DNDEBUG -g -fwrapv -Os
> -Wall -Wstrict-prototypes -DENABLE_DTRACE -arch i386 -arch ppc -arch
> x86_64"
> 
> stepmake/aclocal.m4 attempts to remove some of the useless flags from
> this, but my testing indicated that the sed command wasn't doing
> anything. Also, the "-arch i386 -arch ppc -arch x86_64" flags were not
> removed and caused errors elsewhere in the configure. I've attached a
> small patch for aclocal.m4 that should properly strip unneeded and
> dangerous cflags returned by python-config.

Thank you.  I tested your patch, both with your `python-config
--cflags' output and my own, and it works just fine in both cases.  I
just pushed it to master.

I made one small change that's just a nitpick: the character class
[OWDOm] has an extra `O', so I removed the first one.

> Unfortunately my font issues remain. Here's the configure output from
> my new patched version of configure:
> http://pastebin.com/m6c2433da
> And the make output:
> https://dl.dropbox.com/u/41619/homebrew_build.txt
> 
> Here's ./configure from a lilypond install via MacPorts (whose fonts
> work properly):
> http://pastebin.com/m2b4ce81
> And the make output:
> https://dl.dropbox.com/u/41619/macports_build.txt

Thanks for sending these.

The most obvious difference I see in the configure logs is the
presence of

  -L/usr/X11/lib

under PANGO_FT2_LIBS for your homebrew build.  Since the fonts failing
to load correctly for you are loaded via the Pango interface, I think
this may be an issue related to Pango.

I couldn't see anything wrong with the make logs after a casual
inspection.

> Also, if you could point me in the direction of the second-phase font
> selecting code (where my lilypond decides to print
> [/usr/X11/lib/X11/fonts/TTF/Vera.ttf], etc), I can start poking around
> in there to see what's going on. Thanks for your help!

Sure.

Have a look in scm/framework-ps.scm at the procedure "supplies-or-needs".

It looks like the fonts for the PostScript output are picked up with
"ly:paper-fonts", which is a C++ callback.  This function is defined
in lily/output-def-scheme.cc.

Hope this helps!  Let me know if you have any more questions.


Thanks,
Patrick