Re: [PATCH 4/5] wined3d: Remove the unused device type parameter from the fragment pipe's get_caps() function.

[Stefan Dösinger]

Am 03.02.2010 um 18:17 schrieb Henri Verbeet:
> On the subject of actually supporting software devices, we should probably
> implement those by creating a software or reference adapter with appropriate
> gl_info and pipe implementations and creating a device based on that.

Wouldn't that conflict with how adapters work in the d3d8/9 API? There an adapter identifies the display hardware, not necessarily the rendering device.

Or, to phrase it in API terms:
> static HRESULT WINAPI IWineD3DImpl_CreateDevice(IWineD3D *iface, UINT adapter_idx,
>         WINED3DDEVTYPE device_type, HWND focus_window, DWORD flags, IUnknown *parent,
>         IWineD3DDeviceParent *device_parent, IWineD3DDevice **device)

You'd want to remove device_type from here as well if we merge it into the adapter(and then deal with splitting up wined3d adapters into d3d8/9 adapters and device types in the client libs)

[PATCH 4/5] wined3d: Remove the unused device type parameter from the fragment pipe's get_caps() function.

[Henri Verbeet]

On 3 February 2010 19:06, Stefan Dösinger  wrote:
> Am 03.02.2010 um 18:17 schrieb Henri Verbeet:
> > On the subject of actually supporting software devices, we should probably
> > implement those by creating a software or reference adapter with appropriate
> > gl_info and pipe implementations and creating a device based on that.
>
> Wouldn't that conflict with how adapters work in the d3d8/9 API? There an adapter identifies the display hardware, not necessarily the rendering device.

A little, but the adapter as such is never exposed in those APIs, just
its index. For cases where it matters (e.g. CheckDeviceFormat()) you
also get a device type parameter.
> Or, to phrase it in API terms:
> > static HRESULT WINAPI IWineD3DImpl_CreateDevice(IWineD3D *iface, UINT adapter_idx,
> >         WINED3DDEVTYPE device_type, HWND focus_window, DWORD flags, IUnknown *parent,
> >         IWineD3DDeviceParent *device_parent, IWineD3DDevice **device)
>
> You'd want to remove device_type from here as well if we merge it into the adapter(and then deal with splitting up wined3d adapters into d3d8/9 adapters and device types in the client libs)

Exposing real wined3d adapters to the client libraries may not be a
bad idea, but it's not required here. You could use adapter_idx and
device_type together to pick an appropriate adapter to create the
device with.

[PATCH 4/5] wined3d: Remove the unused device type parameter from the fragment pipe's get_caps() function.

[Stefan Dösinger]

Am 03.02.2010 um 19:53 schrieb Henri Verbeet:
> On 3 February 2010 19:06, Stefan Dösinger  wrote:
> > Am 03.02.2010 um 18:17 schrieb Henri Verbeet:
> > > On the subject of actually supporting software devices, we should probably
> > > implement those by creating a software or reference adapter with appropriate
> > > gl_info and pipe implementations and creating a device based on that.
> >
> > Wouldn't that conflict with how adapters work in the d3d8/9 API? There an adapter identifies the display hardware, not necessarily the rendering device.
>
> A little, but the adapter as such is never exposed in those APIs, just
> its index. For cases where it matters (e.g. CheckDeviceFormat()) you
> also get a device type parameter.

So you want to separate d3d9 adapters from wined3d adapters? Sounds ok to me. We should just find a new name for one of them when(and if ever) we write a SW vertex pipeline.