error : conversion to 'float' from 'double'
vtorri opened this issue · comments
with graphene 1.8.6, cross-compiling on linux and targetting windows 64 bits (using mingw-w64), i get this error when returning INFINITY in graphene-ray.c, lines 285 and 292.
I can provide the cross compilation file, meson output and ninja output if needed
Yes, please: attach the build logs.
here they are :
the cross toolchain file :
cross_toolchain.txt
meson is invoked like that :
meson .. --prefix=$3 --libdir=lib --buildtype=release --strip --cross-file ../cross_toolchain.txt --default-library shared -Dintrospection=false
the output is :
and ninja output :
thank you
I tink that i understand what the problem is: in mingw-w64's math.h (on fedora 27), if GNUC is defined (which is the case), INFINITY is defined as __builtin_inf and not __builttin_inff
when I compile on Windows, INFINITY is correctly defined as __builtin_inff
maybe i will do a sed when cross compiling on fedora
Closing as it is not a bug of graphene
may
Right, I was about to leave a reply after looking through mingw's issue tracker. 😃
C99 says that INFINITY is a float value; apparently, mingw used the wrong compiler built-in. It should have been fixed upstream.
on Windows, with MSYS2, INFINITY is correctly defined. I'm upgrading fedora to f30 and i'll see if INFINITY value has been fixed.