[Ulf Lamping <ulf.lamping@xxxxxx>]

Guy Harris wrote:

> Gisle Vanem wrote:
>
> > I guess it's because WIN32 isn't a built-in in MSVC (in some
> > cases?). But _WIN32 always is.
>
>
> Is there any compelling reason not to replace all tests of WIN32 with 
> tests of _WIN32 in Ethereal?  (We already test _WIN32 in a number of 
> places.)  Does _WIN32 get defined by Mingw32 also?  (What about Cygwin?)
I don't know the ways cygwin/mingw does that part (but I would think 
they do it the same way as MSVC, or at least they should).

I think it's a mix a both _WIN32 and WIN32 in the code, as noone 
(including me) was thinking about a solution to that topic before.

There would be another candidate: G_OS_WIN32, but it's only documented 
from Glib2.0 on. As we only support GLib 2.0 on windows, we could also 
use that one too.


We should first decide which way to go, then put this info into the 
README.developer and then change the code that way at all places.

Regards, ULFL