PATCH: Fix LoadCursor() under win9x

[Andreas Mohr]

On Fri, Oct 12, 2001 at 08:13:20PM +0200, Marcus Meissner wrote:
> On Fri, Oct 12, 2001 at 03:16:04PM +0200, Johan Gill wrote:
> > LoadCursor() should return 0 if we emulate win9x and the second parameter
> > is an invalid pointer.
> 
> > +    if (HIWORD(name))
> > +       if (GetVersion() & 0x80000000)  /* win9x */
> > +           if (IsBadReadPtr(name, 4))
> > +               return 0;
> 
> This code is unclean:
> 	It should not check for versions.
Hmm.

> 	It should not use IsBadReadPtr(), but IsBadStringPtrA().
Oops, yep.

> 	It should not use IsBad* functions, but an exception handler.
Hmm, probably.

Oh no, not the version difference discussion again ;-)

Excuse me, but our testing revealed that there *is* a difference between
Win9x and NT 4, namely that Win9x returns 0, whereas NT 4 crashes
(the program, that is) !

Using your exception handler, LoadCursor() *always* returns 0
on invalid parameter, which is WRONG in the NT 4 case
since it continues program execution without crashing.

--> GetVersion() is required, check was (nearly) right IMHO.
Probably the best solution would be to use both an exception handler and
call GetVersion() on exception.

Well, you'll probably now argue that the NT 4 crash case is unneeded,
however I'm not that convinced about it, since it's a real Wine
incompatibility versus NT if we don't check on winver.
Any reasons for doing it this way and *not* (more or less) our way ? ;-)

-- 
Andreas Mohr                        Stauferstr. 6, D-71272 Renningen, Germany
"Point is, nothing here is unfamiliar or unexpected.  How long does
it take before there's general recognition that Microsoft software
has no business on the Internet? - Dennis E. Powell re: Code Red