Re: [PATCH 07/10] mf: Introduce topology_loader_connect_source() to handle connections of source nodes in _Load().

[Sergio Gómez Del Real]

On 22/06/20 6:53 a. m., Nikolay Sivov wrote:
> On 6/15/20 4:41 AM, Sergio Gómez Del Real wrote:
> > +    hr = IMFAttributes_GetUINT32(attrs_src, &MF_TOPONODE_CONNECT_METHOD, &method);
> > +    if (!enum_src_types || (hr == S_OK && !(method & MF_CONNECT_RESOLVE_INDEPENDENT_OUTPUTTYPES)))
> > +    {
> > +        for (method = MF_CONNECT_DIRECT; method < MF_CONNECT_RESOLVE_INDEPENDENT_OUTPUTTYPES; method++)
> > +        {
> > +            for (i = 0; i < num_media_types; i++)
> > +            {
> > +                IMFMediaTypeHandler_SetCurrentMediaType(mth_src, src_mediatypes[i]);
> > +                hr = S_OK;
> > +                goto out;
> > +            }
> > +        }
> > +    }
> > +    else
> > +    {
> > +        for (i = 0; i < num_media_types; i++)
> > +        {
> > +            for (method = MF_CONNECT_DIRECT; method < MF_CONNECT_RESOLVE_INDEPENDENT_OUTPUTTYPES; method++)
> > +            {
> > +                IMFMediaTypeHandler_SetCurrentMediaType(mth_src, src_mediatypes[i]);
> > +                hr = S_OK;
> > +                goto out;
> > +            }
> > +        }
> > +    }
> > I think this was carried from previous iteration of the patches? I think
> > same concern still apply, why specified method is essentially
> > ignored? Loops are reduced to 1 iteration until following patches, it's
> > better to avoid that.

Yes, this is indeed incorrect. Forgot to apply the modifications I had. 
Will send them in next iteration.