Re: [PATCH 4/8] msi: Fix merging string type checks.

[Alexandre Julliard]

Hib Eris  writes:
> +static BOOL merge_type_match(LPCWSTR type1, LPCWSTR type2)
> +{
> - const WCHAR szl[] = {'l', 0};
> - const WCHAR szs[] = {'s', 0};
> - const WCHAR szL[] = {'L', 0};
> - const WCHAR szS[] = {'S', 0};
> -
> - if ((!StrCmpNW(type1, szl, 1) || !StrCmpNW(type1, szs, 1)) &&
> -
>    (!StrCmpNW(type2, szl, 1) || !StrCmpNW(type2, szs, 1)))
> -
>    return TRUE;
> -
> - if ((!StrCmpNW(type1, szL, 1) || !StrCmpNW(type1, szS, 1)) &&
> -
>    (!StrCmpNW(type2, szL, 1) || !StrCmpNW(type2, szS, 1)))
> -
>    return TRUE;
> -
> - return !StrCmpNW(type1, type2, 1);
>
> +}

Surely you don't need StrCmpNW to compare one character...

[PATCH 4/8] msi: Fix merging string type checks.

[Hib Eris]

On Tue, Oct 20, 2009 at 3:06 PM, Alexandre Julliard  wrote:
> Hib Eris  writes:
> > +static BOOL merge_type_match(LPCWSTR type1, LPCWSTR type2)
> > +{
> > - const WCHAR szl[] = {'l', 0};
> > - const WCHAR szs[] = {'s', 0};
> > - const WCHAR szL[] = {'L', 0};
> > - const WCHAR szS[] = {'S', 0};
> > -
> > - if ((!StrCmpNW(type1, szl, 1) || !StrCmpNW(type1, szs, 1)) &&
> > - (!StrCmpNW(type2, szl, 1) || !StrCmpNW(type2, szs, 1)))
> > - return TRUE;
> > -
> > - if ((!StrCmpNW(type1, szL, 1) || !StrCmpNW(type1, szS, 1)) &&
> > - (!StrCmpNW(type2, szL, 1) || !StrCmpNW(type2, szS, 1)))
> > - return TRUE;
> > -
> > - return !StrCmpNW(type1, type2, 1);
> >
> > +}
>
> Surely you don't need StrCmpNW to compare one character...

I guess not. This may sound as a stupid question to you, but how should I do it?
Can I just do

if (type[0] == 's')  ?

I am a bit worried that this comparison on WCHAR is not guaranteed to work?

Hib Eris

[PATCH 4/8] msi: Fix merging string type checks.

[Andrew Eikum]

Hib Eris wrote:
> On Tue, Oct 20, 2009 at 3:06 PM, Alexandre Julliard  wrote:
> > Surely you don't need StrCmpNW to compare one character...
>
> I guess not. This may sound as a stupid question to you, but how should I do it?
> Can I just do
>
> if (type[0] == 's')  ?
>
> I am a bit worried that this comparison on WCHAR is not guaranteed to work?
>
> Hib Eris

WCHAR is just unsigned short, so the char will promote to short, padding 
with 0s on the high byte.  The comparison will work fine.