Re: [01/18] windowscodecs: Do not assume that vtable is the first element of the object.

[Michael Stefaniuc]

Hello Dmitry,

On 07/24/2012 08:01 AM, Dmitry Timoshkov wrote:
> The patches in this series do not depend on each other, numeration is just
> for a convenience. Patches do basically the same job for different objects,
> so I decided to not invent a new subject for every separate kind of object.
>
> dlls/windowscodecs/bmpdecode.c | 9 ++++++---
>  1 file changed, 6 insertions(+), 3 deletions(-)
>
> diff --git a/dlls/windowscodecs/bmpdecode.c b/dlls/windowscodecs/bmpdecode.c
> index 6da6a3c..1a1ee0a 100644
> --- a/dlls/windowscodecs/bmpdecode.c
> +++ b/dlls/windowscodecs/bmpdecode.c
> @@ -94,6 +94,8 @@ static inline BmpDecoder *impl_from_IWICBitmapFrameDecode(IWICBitmapFrameDecode
>  static HRESULT WINAPI BmpFrameDecode_QueryInterface(IWICBitmapFrameDecode *iface, REFIID iid,
>      void **ppv)
>  {
> -
>
> BmpDecoder *This = impl_from_IWICBitmapFrameDecode(iface);
> -
>
> TRACE("(%p,%s,%p)\n", iface, debugstr_guid(iid), ppv);
>
> if (!ppv) return E_INVALIDARG;
>
> @@ -102,7 +104,7 @@ static HRESULT WINAPI BmpFrameDecode_QueryInterface(IWICBitmapFrameDecode *iface
>          IsEqualIID(&IID_IWICBitmapSource, iid) ||
>          IsEqualIID(&IID_IWICBitmapFrameDecode, iid))
>      {
> -
>    *ppv = iface;
> -
>    *ppv = &This->IWICBitmapFrameDecode_iface;
> }
>    else
>    {

this part of the change is gratuitous. As long as there is only one
interface implementation in the object there is no need for impl_from_Foo().

bye
    michael

[01/18] windowscodecs: Do not assume that vtable is the first element of the object.

[Dmitry Timoshkov]

Michael Stefaniuc  wrote:
> > @@ -102,7 +104,7 @@ static HRESULT WINAPI BmpFrameDecode_QueryInterface(IWICBitmapFrameDecode *iface
> >          IsEqualIID(&IID_IWICBitmapSource, iid) ||
> >          IsEqualIID(&IID_IWICBitmapFrameDecode, iid))
> >      {
> > -
> >    *ppv = iface;
> > -
> >    *ppv = &This->IWICBitmapFrameDecode_iface;
> > }
> >    else
> >    {
>
> this part of the change is gratuitous. As long as there is only one
> interface implementation in the object there is no need for impl_from_Foo().

That's for consistency with other places, so if/when a new interface is
going to be added it's obvious for an implementror what to do.

Thanks for the review.