Re: oleview: The PSDK's ELEMDESC cannot be used with compilers that don't support nameless unions. So define our own structure so the test compiles with both the Wine and the PSDK headers.

[Alexandre Julliard]

Francois Gouget  writes:
> @@ -537,7 +547,7 @@ int EnumFuncs(ITypeInfo *pTypeInfo, int 
>              }
>              bFirst = TRUE;
>  #define ENUM_PARAM_FLAG(x)\
> -
>        if(pFuncDesc->lprgelemdescParam[j].paramdesc.wParamFlags & x)\
> -
>        if(((myELEMDESC*)&(pFuncDesc->lprgelemdescParam[j]))->u.paramdesc.wParamFlags & x) \
>        {\

Why can't you use U(x) instead?

oleview: The PSDK's ELEMDESC cannot be used with compilers that don't support nameless unions. So define our own structure so the test compiles with both the Wine and the PSDK headers.

[Francois Gouget]

On Fri, 20 Oct 2006, Alexandre Julliard wrote:
> Francois Gouget  writes:
> > @@ -537,7 +547,7 @@ int EnumFuncs(ITypeInfo *pTypeInfo, int 
> >              }
> >              bFirst = TRUE;
> >  #define ENUM_PARAM_FLAG(x)\
> > -
> >        if(pFuncDesc->lprgelemdescParam[j].paramdesc.wParamFlags & x)\
> > -
> >        if(((myELEMDESC*)&(pFuncDesc->lprgelemdescParam[j]))->u.paramdesc.wParamFlags & x) \
> >        {\
>
> Why can't you use U(x) instead?

My basic assumption is that we want to be able to compile oleview on 
Windows. But then I had in mind that this was a conformance test so I 
was pretty confused. Still it seems like cmopiling our programs on 
Windows should be feasible.

So I modeled my approach on what I remembered of an old commit to 
dlls/user/tests/input.c (and menu.c) which had the same issue with the 
INPUT type.

The PSDK uses a nameless union whether NONAMELESSUNION is defined or 
not. But if we assume that NONAMELESSUNION will never be defined when we 
compile with the PSDK headers, then we can use U(x).

oleview: The PSDK's ELEMDESC cannot be used with compilers that don't support nameless unions. So define our own structure so the test compiles with both the Wine and the PSDK headers.

[Alexandre Julliard]

Francois Gouget  writes:
> The PSDK uses a nameless union whether NONAMELESSUNION is defined or 
> not. But if we assume that NONAMELESSUNION will never be defined when we 
> compile with the PSDK headers, then we can use U(x).

That's what the rest of the oleview code already does, so it should be
OK.