There are several warnings that get flagged when
building with MinGW32 3.2.
One of them stumps me, and that's a set about "warning:
cast does not match function type", on the WIN32
MAP_CURSOR() macro invocations.
The rest look fairly minor.
When all is said and done, MinGW won't build a shared
library because the MinGW distribution is missing a
proper libm.a (and other files). I may need to remove
the -no-undefined optin for MinGW (it was ironically a
necessary option for CygWIN's libtool).
Or MinGW may simply have to build static libraries.+
It would be nice to support this so that one can use GCC
to build hardware-accelerated OpenGL OpenGLEAN
applications on WIN32.
Anyway, I've been toying with this a bit today. I'll
commit changes, soon, and test that WIN32 and CygWIN
aren't hurt by them.
Logged In: YES
user_id=854844
Oh, with NetBSD/amd64, having built the MinGW cross compiler
from pkgsrc in the usual way, I can build OpenGLEAN for WIN32
(untested) after using this ./configure line:
PATH="/usr/pkg/cross/i386-mingw32/bin:$PATH"
CC='/usr/pkg/cross/i386-mingw32/bin/gcc'
CPPFLAGS='-I/usr/pkg/cross/i386-mingw32/include'
LDFLAGS='-L/usr/pkg/cross/i386/i386-mingw32/lib' ./configure
--host=mingw32 --without-x --disable-warnings
...the --without-x is necssary to avoid binding to X; we
want WIN32 bindings. The --disable-warnings is necessary
because of problems with cursor.c (as noted above).
Logged In: YES
user_id=854844
Because of things like -lGL and -lGLU in the Makefile for the
core library, the demos do not compile. Probably this can be
hacked around by copying, e.g., MinGW's libopengl32.a to
libGL.a, but this suggests a deeper problem with how the
auto* tools are generating Makefiles---and how they are
detecting MinGW.
For now, I'm putting this on a back burner, but would welcome
further input from MinGW users.