Re: [Gnash-dev] how to handle hash_map

[Rob Savoye]

Erik de Castro Lopo wrote:

> Someone please correct me if I'm wrong, but I think winegcc/wineg++ 
> generate Linux binaries from windows source code using the Wine
> libraries to handle the win32 API calls.

  Ah, that could be.

> I tried the cross compile path for my project, with the test suit
> being run using Wine. Unfortunately this never quite worked right
> and i didn't have the time or patience to pursue it.

  Setting up a project to be cross configurable and cross compilable is 
complex. Luckily I did huge amounts of this when I worked at Cygnus, so 
I added initial MingW cross compiling support the other day. I had to 
build up a MingW environment with everything Gnash needed, like SDL, 
XML2, libpng, libjpeg, pthreads, libz, etc... It seems like MingW is the 
way to go, and hopefully can also support being used for a Firefox 
plugin on win32.

  I added configure tests for all the differences I could find, and 
added the few things like gettimeofday() that Gnash needed. So now you 
can cross configure Gnash by setting --build, --host, and --target 
appropriately. I'll add something to the manual on this when I get a chance.

 I used this to configure successfully:

/home/rob/projects/gnu/gnash/configure --prefix=/usr/mingw 
--build=i686-pc-linux-gnu --host=i586-mingw32msvc 
--target=i586-mingw32msvc --disable-plugin 
--with-pthread-incl=/usr/mingw/include --with-pthread-lib=/usr/mingw/lib 
--with-sdl-incl=/usr/mingw/include --with-sdl-lib=/usr/mingw/lib 
--with-sdl-mixer-incl=/usr/mingw/include --with-sdl-mixer-lib=/usr/mingw/lib

  Currently executables aren't produced fully linked. I'll fix that on 
my next pass through this stuff. (unless somebody else gets the urge...) 
I think the hard part is done. I don't think it'll take much more work 
to beat it into shape, and it sure was easier than trying to use VC++.

        - rob -