cabinet: Include wine/port.h for strcasecmp().

[Kirill K. Smirnov]

> The rules are as follows:
> - For case insensitive comparisons of 'windows strings', a Windows 
> function must be used (stricmp() or similar).

and
> In this case, KERNEL32 lstrcmpiA() would be more appropriate. In
> general, using stricmp() in Win32 code should be discouraged since it
> may cause unnecessary dependency on MSVCRT.

Wine lstrcmpi is broken (bug #10767). There are cases GLIBC strcasecmp and 
Windows lstrcmpi handle correctly but wine not. Should we fix this first?

--
Kirill

[Dmitry Timoshkov]

"Kirill K. Smirnov"  wrote:
> > The rules are as follows:
> > - For case insensitive comparisons of 'windows strings', a Windows 
> > function must be used (stricmp() or similar).
>
> and
> > In this case, KERNEL32 lstrcmpiA() would be more appropriate. In
> > general, using stricmp() in Win32 code should be discouraged since it
> > may cause unnecessary dependency on MSVCRT.
>
> Wine lstrcmpi is broken (bug #10767). There are cases GLIBC strcasecmp and 
> Windows lstrcmpi handle correctly but wine not. Should we fix this first?

The bug 10767 is a corner case, and nothing prevents using kernel32 APIs
to perform string comparisons in Wine (which we do). I don't think that
you want to use glibc string comparison functions to compare for instance
cyrillic strings in CP1251 (Wine) and UTF-8 (Linux) encodings. Still, the bug
10767 has an explanation for an incompatibility, so the problem should be
either solved on unicode.org side, or by patching libs/wine/cpmap.pl.