Created attachment 28179 [details] [review]
the patch

Gfx.cc has three places where it assigns a double expression to an int variable.  gcc complains with
Gfx.cc: In member function 'void Gfx::doAxialShFill(GfxAxialShading*)':
Gfx.cc:2515: warning: converting to 'int' from 'double'
Gfx.cc:2529: warning: converting to 'int' from 'double'
Gfx.cc: In member function 'void Gfx::doImage(Object*, Stream*, GBool)':
Gfx.cc:3964: warning: converting to 'int' from 'double'
This code is a little risky because on some systems, assigning a double outside the range of INT_MIN .. INT_MAX to an int can cause a core dump.
I added an (int) cast to make gcc quiet.
To be safer, maybe teoricalj should be changed from int to double and the test should be
  if (floor(teoricalj) <= (double)i) ...