Re: [PATCH 05/10] d3d11: Implement D3D11CreateDeviceAndSwapChain().

[Józef Kucia]

On Mon, Sep 14, 2015 at 1:48 PM, Henri Verbeet  wrote:
> On 14 September 2015 at 00:13, Józef Kucia  wrote:
> > - if (swapchain_out)
> > -
> >    *swapchain_out = swapchain;
> > - else
> > -
> >    IDXGISwapChain_Release(swapchain);
>
> Is "desc" validated or modified in this case? Otherwise you could
> return a lot earlier. The test doesn't really cover that.

You're right. The test should cover this case.
> > - hr = D3D11CreateDevice(adapter, driver_type, swrast, flags, feature_levels, levels, sdk_version,
> > -
> >        &device, NULL, NULL);
> > - if (FAILED(hr))
> > - {
> > -
> >    WARN("Failed to create a device, returning %#x.\n", hr);
> > -
> >    *device_out = NULL;
> > -
> >    return hr;
> > - }
>
> ...
> > -
> > - if (obtained_feature_level)
> > -
> >    *obtained_feature_level = ID3D11Device_GetFeatureLevel(device);
> > -
> > - if (immediate_context)
> > -
> >    ID3D11Device_GetImmediateContext(device, immediate_context);
>
> This seems odd. Why not just pass these to D3D11CreateDevice()?

The obtained_feature_level should definitely be passed to
D3D11CreateDevice. When it comes to the immediate context, the intent
was to get it only when the function succeeds so it doesn't have to be
released in every case in which the function fails.