Re: jscript: Added NAN definition (1/10)

[Alexandre Julliard]

Piotr Caban  writes:
> diff --git a/dlls/jscript/jscript_main.c b/dlls/jscript/jscript_main.c
> index 37266e5..740bb06 100644
> --- a/dlls/jscript/jscript_main.c
> +++ b/dlls/jscript/jscript_main.c
> @@ -38,6 +38,21 @@ static const CLSID CLSID_JScriptAuthor =
>  static const CLSID CLSID_JScriptEncode =
>      {0xf414c262,0x6ac0,0x11cf,{0xb6,0xd1,0x00,0xaa,0x00,0xbb,0xbb,0x58}};
>
> +#ifndef NAN
> +DOUBLE NAN;
> -
>
> +static inline void set_jsnan(void)
> +{
> - VARIANT v;
> - V_UI8(&v) = (ULONGLONG)0x7ff80000<<32;
> - NAN = V_R8(&v);
>
> +}
> +#else
> +static inline void set_jsnan(void)
> +{
> +}
> +#endif
> -
>
> DEFINE_GUID(GUID_NULL,0,0,0,0,0,0,0,0,0,0,0);
>
> static HINSTANCE jscript_hinstance;
> @@ -110,6 +125,7 @@ BOOL WINAPI DllMain(HINSTANCE hInstDLL, DWORD fdwReason, LPVOID lpv)
>          return FALSE;  /* prefer native version */
>      case DLL_PROCESS_ATTACH:
>          DisableThreadLibraryCalls(hInstDLL);
> -
>    set_jsnan();
>    jscript_hinstance = hInstDLL;
>    break;

The previous implementation was better, having to explicitly initialize
the constant on startup is ugly.

jscript: Added NAN definition (1/10)

[Piotr Caban]

Alexandre Julliard wrote:
> Piotr Caban  writes:
> > diff --git a/dlls/jscript/jscript_main.c b/dlls/jscript/jscript_main.c
> > index 37266e5..740bb06 100644
> > --- a/dlls/jscript/jscript_main.c
> > +++ b/dlls/jscript/jscript_main.c
> > @@ -38,6 +38,21 @@ static const CLSID CLSID_JScriptAuthor =
> >  static const CLSID CLSID_JScriptEncode =
> >      {0xf414c262,0x6ac0,0x11cf,{0xb6,0xd1,0x00,0xaa,0x00,0xbb,0xbb,0x58}};
> >
> > +#ifndef NAN
> > +DOUBLE NAN;
> > -
> >
> > +static inline void set_jsnan(void)
> > +{
> > - VARIANT v;
> > - V_UI8(&v) = (ULONGLONG)0x7ff80000<<32;
> > - NAN = V_R8(&v);
> >
> > +}
> > +#else
> > +static inline void set_jsnan(void)
> > +{
> > +}
> > +#endif
> > -
> >
> > DEFINE_GUID(GUID_NULL,0,0,0,0,0,0,0,0,0,0,0);
> >
> > static HINSTANCE jscript_hinstance;
> > @@ -110,6 +125,7 @@ BOOL WINAPI DllMain(HINSTANCE hInstDLL, DWORD fdwReason, LPVOID lpv)
> >          return FALSE;  /* prefer native version */
> >      case DLL_PROCESS_ATTACH:
> >          DisableThreadLibraryCalls(hInstDLL);
> > -
> >    set_jsnan();
> >    jscript_hinstance = hInstDLL;
> >    break;
>
> The previous implementation was better, having to explicitly initialize
> the constant on startup is ugly.

Some functions needs to return NAN. The other possibility of
implementing it is following:

static inline DOUBLE ret_nan(void)
{
    VARIANT v;
    num_set_nan(&v);
    return V_R8(&v);
}

or

#ifdef NAN
static inline DOUBLE ret_nan(void)
{
    return NAN;
}
#else
static inline DOUBLE ret_nan(void)
{
    VARIANT v;
    num_set_nan(&v);
    return V_R8(&v);
}
#endif

But both of these methods are probably even uglier. Do you have any
suggestions how should I implement it?

Piotr

jscript: Added NAN definition (1/10)

[Alexandre Julliard]

Piotr Caban  writes:
> Some functions needs to return NAN. The other possibility of
> implementing it is following:
>
> static inline DOUBLE ret_nan(void)
> {
>     VARIANT v;
>     num_set_nan(&v);
>     return V_R8(&v);
> }
>
> or
>
> #ifdef NAN
> static inline DOUBLE ret_nan(void)
> {
>     return NAN;
> }
> #else
> static inline DOUBLE ret_nan(void)
> {
>     VARIANT v;
>     num_set_nan(&v);
>     return V_R8(&v);
> }
> #endif
>
> But both of these methods are probably even uglier. Do you have any
> suggestions how should I implement it?

The first one is fine.