Re: oleaut32: Add a VT_INT_PTR case for VARIANT_UserSize

[Alexandre Julliard]

Nikolay Sivov  writes:
> @@ -239,6 +239,8 @@ static unsigned int get_type_size(ULONG *pFlags, VARTYPE vt)
>      case VT_DISPATCH:
>      case VT_RECORD:
>          return 0;
> - case VT_INT_PTR:
> -
>    return sizeof(INT_PTR);

This won't do the right thing on 64-bit.

oleaut32: Add a VT_INT_PTR case for VARIANT_UserSize

[Nikolay Sivov]

On 3/2/2010 19:01, Alexandre Julliard wrote:
> Nikolay Sivov  writes:
> > @@ -239,6 +239,8 @@ static unsigned int get_type_size(ULONG *pFlags, VARTYPE vt)
> >       case VT_DISPATCH:
> >       case VT_RECORD:
> >           return 0;
> > - case VT_INT_PTR:
> > -
> >    return sizeof(INT_PTR);
>
> This won't do the right thing on 64-bit.

Unfortunately I don't have a 64bit to cross-build a test on it. Should 
it just be 4? or there's something more complicated?

oleaut32: Add a VT_INT_PTR case for VARIANT_UserSize

[Paul Vriens]

On 03/02/2010 05:16 PM, Nikolay Sivov wrote:
> On 3/2/2010 19:01, Alexandre Julliard wrote:
> > Nikolay Sivov writes:
> > > @@ -239,6 +239,8 @@ static unsigned int get_type_size(ULONG *pFlags,
> > > VARTYPE vt)
> > > case VT_DISPATCH:
> > > case VT_RECORD:
> > > return 0;
> > > - case VT_INT_PTR:
> > > - return sizeof(INT_PTR);
> >
> > This won't do the right thing on 64-bit.
>
> Unfortunately I don't have a 64bit to cross-build a test on it. Should
> it just be 4? or there's something more complicated?

Couldn't you just submit the patch to winetestbot and run the tests on 
64bit boxes only?

oleaut32: Add a VT_INT_PTR case for VARIANT_UserSize

[Nikolay Sivov]

On 3/2/2010 19:21, Paul Vriens wrote:
> On 03/02/2010 05:16 PM, Nikolay Sivov wrote:
> > On 3/2/2010 19:01, Alexandre Julliard wrote:
> > > Nikolay Sivov writes:
> > > > @@ -239,6 +239,8 @@ static unsigned int get_type_size(ULONG *pFlags,
> > > > VARTYPE vt)
> > > > case VT_DISPATCH:
> > > > case VT_RECORD:
> > > > return 0;
> > > > - case VT_INT_PTR:
> > > > - return sizeof(INT_PTR);
> > >
> > > This won't do the right thing on 64-bit.
> >
> > Unfortunately I don't have a 64bit to cross-build a test on it. Should
> > it just be 4? or there's something more complicated?
>
> Couldn't you just submit the patch to winetestbot and run the tests on 
> 64bit boxes only?

Will it build as 64bit executable?

oleaut32: Add a VT_INT_PTR case for VARIANT_UserSize

[Paul Vriens]

On 03/02/2010 05:26 PM, Nikolay Sivov wrote:
> On 3/2/2010 19:21, Paul Vriens wrote:
> > On 03/02/2010 05:16 PM, Nikolay Sivov wrote:
> > > On 3/2/2010 19:01, Alexandre Julliard wrote:
> > > > Nikolay Sivov writes:
> > > > > @@ -239,6 +239,8 @@ static unsigned int get_type_size(ULONG *pFlags,
> > > > > VARTYPE vt)
> > > > > case VT_DISPATCH:
> > > > > case VT_RECORD:
> > > > > return 0;
> > > > > - case VT_INT_PTR:
> > > > > - return sizeof(INT_PTR);
> > > >
> > > > This won't do the right thing on 64-bit.
> > >
> > > Unfortunately I don't have a 64bit to cross-build a test on it. Should
> > > it just be 4? or there's something more complicated?
> >
> > Couldn't you just submit the patch to winetestbot and run the tests on
> > 64bit boxes only?
>
> Will it build as 64bit executable?

I think it does both 32bit and 64bit.

oleaut32: Add a VT_INT_PTR case for VARIANT_UserSize

[Alexandre Julliard]

Nikolay Sivov  writes:
> Unfortunately I don't have a 64bit to cross-build a test on it. Should
> it just be 4? or there's something more complicated?

Most likely more complicated, test cases are probably in order.

oleaut32: Add a VT_INT_PTR case for VARIANT_UserSize

[Nikolay Sivov]

On 3/2/2010 19:31, Alexandre Julliard wrote:
> Nikolay Sivov  writes:
> > Unfortunately I don't have a 64bit to cross-build a test on it. Should
> > it just be 4? or there's something more complicated?
>
> Most likely more complicated, test cases are probably in order.

Ok, just tried with testbot to build 64bit tests as Paul suggested 
(thanks), return from VARIANT_UserSize is the same for 32/64. Let's 
delay this.