[Openexr-devel] notes: OpenEXR-1.0.6 + OSX + ProjectBuilder

[Scott Frankel]

These notes were taken from my experiences with OpenEXR on MacOSX,
using Apple's ProjectBuilder IDE.  I'm sure there are more elegant
solutions to the various issues addressed below.  Nontheless, these
steps yielded a successful build of the example program.

Happy exring -
Scott


Versions

    MacOSX v10.2.6
    OpenEXR 1.0.6
    ProjectBuilder v2.1
    GCC v3.1


Build & Install OpenEXR libraries and header files

    - as root:
      % ./configure --enable-osx-istream-hack --enable-shared=no
      % make
      % make install


    - API header hacks: disable win32 compatibility
      Comment-out the win32 compatibility term, "restrict," from
      ImathMatrix.h; i.e.:
      [near line 2356]

      #if defined (PLATFORM_WIN32)
      #define restrict
      #endif

          /*
          register const T * restrict ap = &a.x[0][0];
          register const T * restrict bp = &b.x[0][0];
          register       T * restrict cp = &c.x[0][0];
          */

          register const T * ap = &a.x[0][0];
          register const T * bp = &b.x[0][0];
          register       T * cp = &c.x[0][0];
          register T a0, a1, a2, a3;


    - example program header hacks: explicit casts to float
      <no matching function for call to 'max(float, double)'>
      Explicitly cast 2nd arg to float in drawImage.cpp:
     [line 92]

      void
      sp (Array2D<Rgba> &px, int w, int h,
          float xc, float yc, float rc,
          float rd, float gn, float bl, float lm)
      {
          int x1 = int (max (floor (xc - rc),  float (0.0) ));
          int x2 = int (min (ceil  (xc + rc),  float (w - 1.0) ));
          int y1 = int (max (floor (yc - rc),  float (0.0) ));
          int y2 = int (min (ceil  (yc + rc),  float (h - 1.0) ));

      [and line 138]
      void
      zsp (Array2D<half> &gpx, Array2D<float> &zpx, int w, int h,
           float xc, float yc, float zc, float rc, float gn)
      {
          int x1 = int (max (floor (xc - rc),  float (0.0) ));
          int x2 = int (min (ceil  (xc + rc),  float (w - 1.0) ));
          int y1 = int (max (floor (yc - rc),  float (0.0) ));
          int y2 = int (min (ceil  (yc + rc),  float (h - 1.0) ));


                                
For ProjectBuilder

    - In a shell window, make a symbolic link to /usr/ in the project's
      working directory.  The MacOSX file Open|Save browser can't find
      BSD-specific directories.  The symlink lets you navigate to 
/usr/lib
      and /usr/local/OpenEXR from the cwd to import headers and lib 
files.

          % ln -s /usr/ yoozer

    - To prevent ImathMath.h float variable errors, this string must be 
added
      to the Target's GCC Compiler Settings tab's "Other C Compiler 
Flags"
      field.  Note that the "-" is significant.

          -DPLATFORM_DARWIN_PPC



    - prevent undefined symbol errors, "_compress" & "_uncompress"
      Add /usr/lib/libz.dylib to the project's External Frameworks and
      Libraries group.


    - prevent undefined symbol errors
      Set the link order of the library files so that libz is linked 
before
      liblex.