Mail Archives: cygwin-developers/2002/09/05/15:53:28
Corinna Vinschen wrote:
>
> On Wed, Sep 04, 2002 at 08:11:17AM -0400, Earnie Boyd wrote:
> > Cygwin cannot use GetConsoleWindow because it doesn't exist before Windows
> > 2000 (W2K).
>
> Again Earnie: You ever heard of autoload.cc?
Heard but not seen.
> Did you have a look into that?
No.
> You know that this file loads functions on demand?
Do now.
> You know that we're already using functions of more recent Windows versions
> _if they are available_?
Do now.
> Cygwin obviously *can* use functions of
> 2K or NT which aren't available in 9x/Me.
Ok, yes.
> So, what exactly are you
> up to?
>
I'm only up to providing the appropriate _WIN32_WINNT guard as specified
by
the MSDN. My patch is appropriate for w32api.
> > We have WINVER set to 0x0400 to avoid the problems you mention.
>
> That doesn't sound like a plan to me.
>
The w32api isn't Cygwin specific. It is the plan that was established
before
I took over. What plan do you have? What plan can we give for others
to use
that use the w32api on older platforms? If you want newer API then set
the
_WIN32_WINNT constant as needed within the application. Sure the same
could
be said for older platforms, but that would cause more headache. It
seems to
me that as you've explained it Cygwin could set the constant to -1 and
survive.
> > doesn't exist pre W2K. The ChangeLog also needs adjusted to mention the
> > autoload.cc changes.
>
> The autoload change was just to fix the GetConsoleWindow AT 4 to a
> GetConsoleWindow AT 0. It was a typo which disallowed successful
> linking, nothing else. Your current version of w32api disallows
> even compiling the stuff.
>
Fine, that doesn't invalidate my change.
> > Yes, Cygwin is what is broken with this patch.
>
> Definitely not. But I'm probably not good enough in explaining things,
> I'll let others discuss that.
>
Funny, you just said that it was in the previous paragraph.
Danny, do you have anything to add to the _WIN32_WINNT issue?
Earnie.
- Raw text -