[ft-devel] bcc and FreeType2 (Re: freetype-2.3.7 -- ftconfig.h for biarch systems)

[mpsuzuki]

Hi,

As a subside discussion of 32/64bit bi-arch header,
I'm interested in the investigation of FreeType2
on 16bit integer platforms. Although it's not the
highest priority task, I will check in my sparetime.
Here is a brief report on a C compiler "bcc".

Just I've tried to build FreeType2 for 16bit ELKS,
with Bruce's C compiler. The behaviour of bcc was
out the cases Autoconf developers expected, so
configure script in cross-building mode didn't work
correctly. It's recommended to use Linux's binfmt_misc
trick, to execute ELKS binary as if it's executed
natively (for detail, please check autoconf ML archive).

After the execution of configure in host-building mode,
I received many errors about the broken preprocessed
lines. For example, the error code definitions in ftmoderr.h
are expanded aslike...

  FT_Mod_Err_ Base = 0, 
  FT_Mod_Err_ Autofit = 0, 
  FT_Mod_Err_ BDF = 0, 
  FT_Mod_Err_ Cache = 0, 

Oops. If I rewrite
          FT_MODERRDEF( Base,      0x000, "base module" )
to
          FT_MODERRDEF(Base,      0x000, "base module" )
, the strange expantion "FT_Mod_Err_ Base" is changed to
"FT_Mod_Err_Base". I guess FreeType2 developers don't
want to rewrite the source for such broken cpp. At least,
bcc is not good starting point, because we cannot sure
the wrong result comes from FreeType2 source or bcc.
I must consider another C compiler.

Regards,
mpsuzuki