GCC for Windows on Mac OS X

Two toolchains based on GCC are popular on Windows: Cygwin, a whole development environment which includes also many other programs and a shell; and MinGW, a complete toolchain for the development of native Windows applications.

We'll describe here how we built MinGW as a cross-compiler for C and C++. It is likely that other GCC compilers can be added without problem.

Download the required elements of MinGW

Go to the download page of MinGW, and download the following files from the Current set:

• src gcc-core
• src gcc-g++
• bin mingw-runtime
• bin w32-api
• src binutils

Launch Terminal

The Terminal application is in the Utilities folder. In the Finder, type Command-Shift-U to open it.

Make a directory for MinGW

We'll store the MinGW toolchain in a disk image, which we find convenient since it makes its deployment on other Mac computers very easy. But depending on your needs, you may disagree; you'll have to adjust the directory names accordingly. We create a directory with the same path as where the image will be mounted:

mkdir /Volumes/MinGW

Change directory to the download folder

In the Terminal application, change directory to the folder where the downloaded elements are stored, typically on the Desktop:

cd ~/Desktop

Unarchive and build MinGW

Type the following commands (you can copy them from this page and paste them to the Terminal window):

# expand archives in temp directory
mkdir /tmp/xgccbuild
mkdir /tmp/xgccbuild/usr
cp binutils-*tar* gcc-*tar* /tmp/xgccbuild
cp mingw-run*tar* w32api-*tar* /tmp/xgccbuild/usr
cd /tmp/xgccbuild
gunzip *.gz
tar xf b*
tar xf gcc-c*
tar xf gcc-g*
rm *.tar
cd usr
gunzip *.gz
tar xf m*
tar xf w*
rm *.tar
cd ..
# copy MinGW headers and libraries to final place
cp -Rf usr/* /Volumes/MinGW
# rename directories for easier access
mv b* binutils
mv gcc-* gcc
# configure and build binutils
mkdir build
cd build
../binutils/configure --target=mingw32 --prefix=/Volumes/MinGW \
 --with-gcc --with-gnu-as --with-gnu-ld \
 --disable-shared --disable-nls
make  CFLAGS="-O2 -D__USE_CRTIMP -fno-exceptions" LDFLAGS=-s
make install
cd ..
rm -Rf build
# add a symbolic link to include in mingw32
ln -s /Volumes/MinGW/include /Volumes/MinGW/mingw32/include
# configure and build gcc
mkdir build
cd build
../gcc/configure --with-gcc --with-gnu-ld --with-gnu-as \
 --target=mingw32 --prefix=/Volumes/MinGW \
 --enable-threads --disable-nls \
 --enable-languages=c,c++ --disable-win32-registry \
 --disable-shared --without-x \
 --with-sysroot=/tmp/xgccbuild
make
export PATH=$PATH:/Volumes/MinGW/bin
make install
cd ..
rm -Rf build

Test compiler

Write a small program such as the one below and save it as test.c

#include <stdio.h>
int main(void)
{
  printf("Hello, Windows!\n");
  return 0;
}

Compile it with the following command:

/Volumes/MinGW/bin/mingw32-gcc -o test.exe test.c

Transfer test.exe to a Windows computer and execute it. Do the same with a C++ program named test.cpp:

#include <iostream>
using namespace std;
int main()
{
  cout << "Hello++, Windows!\n";
  return 0;
}

The C++ compiler is invoked with

/Volumes/MinGW/bin/mingw32-g++ -o test.exe test.cpp

Make a disk image

Create a .dmg image from /Volumes/MinGW, remove MinGW directory and mount image:

cd
hdiutil create -srcfolder /Volumes/MinGW MinGW.dmg
rm -Rf /Volumes/MinGW
open MinGW.dmg