On Mon, 2004-01-05 at 14:46, Ciprian Popovici wrote: > On Sun, 4 Jan 2004 18:39:51 -0800 (PST) Jean Hernandez <ean79@xxxxxxxxx> > wrote: > > /usr/local/include/freetype2/freetype/freetype.h:40: ft2build.h: No > > such file or directory > > By the looks of this, you're missing freetype2, or fontconfig doesn't > know where to find the headers for it. I'm not familiar with the way > FreeBSD does the compilation, normally I would suggest to pass some more > info to "configure", something like this: > > CPPFLAGS="-I/path/to/freetype2/include" ./configure --your-options-here On my system, I have /usr/include/ft2build.h and /usr/include/freetype2/freetype/freetype.h. Both files are part of the Redhat package freetype-devel-2.1.3-6. The OP seems to be missing /usr/include/ft2build.h for whatever reason, To compile code that uses freetype 2 I use -I/usr/include/freetype2 (or whatever is returned by the shell command `freetype-config --cflags`), and in the source files, #include <ft2build.h> #include FT_FREETYPE_H. where FT_FREETYPE_H is a macro defined (indirectly) in <ft2build.h>. I believe this is the canonical way of compiling with Freetype 2. I have no idea what should be the advantage of all this convolution. The use of these macros do not remove the need to add an element to the include path. Of course, each level of indirection added is an added opportunity for the administrators to relocate file trees, but that possibility already exists in freetype-config. I guess the advantage is that a future freetype3 needs not redefine all the headers, and so the macros in ft3build.h can point at some ft2 files and some ft3 files. Hooray. The file ft2build.h must also be in the include path. Regards, Enrique
