[Guy Harris <guy@xxxxxxxxxxxx>]

On Feb 5, 2004, at 3:49 AM, Biot Olivier wrote:

> |  We can't test the GTK+ version in dissectors or libethereal code, as
> |  they don't include any GTK+ headers (and shouldn't do so, as
> |  they can't use GTK+); we use the GLib version as a proxy.
>
> I am not sure whether this will work as Windows builds use Glib 2.2.3
> and Gtk+ 1.x (or Gtk+ 2).

Windows builds use either GTK+ 1.3[.x] or 2.x, both of which use UTF-8 
internally - as the page at

	http://www.gimp.org/~tml/gimp/win32/

says:

	GTK+ 1.3 for Windows (and GTK+ 2.x on all platforms) uses Unicode 
internally. All strings you pass to GTK+ functions must be in UTF-8. 
You can not just use non-ASCII characters from some Windows single- or 
multi-byte codepage.

As we now use Glib 2.x for Windows, "are we using Glib 2.x?" should 
always be true on Windows, just as "do the GTK+ routines take UTF-8" is 
always true on Windows.

On UN*X+X, we presumably will be using Glib 2.x iff we're using GTK+ 
2.x, and GTK+ 2.x uses UTF-8.  GTK+ 1.2[.x] doesn't use UTF-8, even 
though the Windows GTK+ 1.3[.x] does.