I was incorrect in thinking that cairo_win32_printing_surface_create() did not need an HDC, but it does. Our bug comes from that value being null.
I'm sure that the new stuff would also fail with a null HDC. I still suspect a stack or malloc problem causing an alignment problem, though I am probably wrong about that, too. ^^
Looking at the Gtk patch, it doesn't look like it would solve our problem:
http:// bugzilla. gnome.org/ attachment. cgi?id= 99752
I was incorrect in thinking that cairo_win32_ printing_ surface_ create( ) did not need an HDC, but it does. Our bug comes from that value being null.
I'm sure that the new stuff would also fail with a null HDC. I still suspect a stack or malloc problem causing an alignment problem, though I am probably wrong about that, too. ^^