May 2024 - Wine-GitLab - list.winehq.org

Re: [PATCH v2 0/2] MR5734: mf: Implement audio capture device enumeration in MFEnumDeviceSources().
by Nikolay Sivov (＠nsivov) May 28, 2024

May 28, 2024

Nikolay Sivov (@nsivov) commented about dlls/mf/sac.c: > + > + hr = IMMDevice_OpenPropertyStore(device, STGM_READ, &ps); > + IMMDevice_Release(device); > + if (FAILED(hr)) > + goto done; > + hr = IPropertyStore_GetValue(ps, (const PROPERTYKEY*)&DEVPKEY_Device_FriendlyName, &pv); > + IPropertyStore_Release(ps); > + if (FAILED(hr)) > + goto done; > + if (pv.vt != VT_LPWSTR) > + { > + PropVariantClear(&pv); > + goto done; > + } > + IMFActivate_SetString(sources[i], &MF_DEVSOURCE_ATTRIBUTE_FRIENDLY_NAME, pv.pwszVal); > + PropVariantClear(&pv); I think cascading checks for hr are better here than multiple jumps. And if property store becomes useful for instance id, we could have a helper that checks value types. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/5734#note_71440

1 0

Re: [PATCH v2 0/2] MR5734: mf: Implement audio capture device enumeration in MFEnumDeviceSources().
by Nikolay Sivov (＠nsivov) May 28, 2024

May 28, 2024

Nikolay Sivov (@nsivov) commented about dlls/mf/main.c: > - FIXME("%p, %p, %p.\n", attributes, sources, count); > + GUID source_type; > + HRESULT hr; > + > + TRACE("%p, %p, %p.\n", attributes, sources, ret_count); > > - if (!attributes || !sources || !count) > + if (!attributes || !sources || !ret_count) > return E_INVALIDARG; > > - *count = 0; > + if (FAILED(hr = IMFAttributes_GetGUID(attributes, &MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE, &source_type))) > + return hr; > > - return S_OK; > + if (IsEqualIID(&source_type, &MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE_VIDCAP_GUID)) It makes no difference in practice, but IsEqualGUID() is a better fit. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/5734#note_71439

1 0

Re: [PATCH v2 0/2] MR5734: mf: Implement audio capture device enumeration in MFEnumDeviceSources().
by Nikolay Sivov (＠nsivov) May 28, 2024

May 28, 2024

Nikolay Sivov (@nsivov) commented about dlls/mf/main.c: > + HRESULT hr; > + > + TRACE("%p, %p, %p.\n", attributes, sources, ret_count); > > - if (!attributes || !sources || !count) > + if (!attributes || !sources || !ret_count) > return E_INVALIDARG; > > - *count = 0; > + if (FAILED(hr = IMFAttributes_GetGUID(attributes, &MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE, &source_type))) > + return hr; > > - return S_OK; > + if (IsEqualIID(&source_type, &MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE_VIDCAP_GUID)) > + { > + FIXME("Not implementted for video capture devices.\n"); Typo here. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/5734#note_71438

1 0

Re: [PATCH v2 0/2] MR5734: mf: Implement audio capture device enumeration in MFEnumDeviceSources().
by Nikolay Sivov (＠nsivov) May 28, 2024

May 28, 2024

Nikolay Sivov (@nsivov) commented about dlls/mf/Makefile.in: > mf.rc \ > quality.c \ > samplegrabber.c \ > + sac.c \ What does "sac" stand for? -- https://gitlab.winehq.org/wine/wine/-/merge_requests/5734#note_71437

1 0

Re: [PATCH v2 0/2] MR5734: mf: Implement audio capture device enumeration in MFEnumDeviceSources().
by Nikolay Sivov (＠nsivov) May 28, 2024

May 28, 2024

Nikolay Sivov (@nsivov) commented about dlls/mf/sac.c: > + if (FAILED(hr = IMMDeviceCollection_GetCount(devices, &count))) > + goto done; > + sources = CoTaskMemAlloc(count * sizeof(*sources)); > + for (i = 0; i < count; ++i) > + { > + WCHAR symbolic_link[256]; > + IPropertyStore *ps; > + IMMDevice *device; > + PROPVARIANT pv; > + WCHAR *str; > + > + if (FAILED(hr = create_activation_object(NULL, &audio_capture_activate_funcs, &sources[i]))) > + goto done; > + ++created_count; > + > + IMFActivate_SetGUID(sources[i], &MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE, &MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE_AUDCAP_GUID); To handle failures in individual Set methods we have things like mediatype_set_guid() in mfplat, that take &hr argument to set error code appropriately without introducing more cleanup conditions. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/5734#note_71436

1 0

Re: [PATCH v2 0/2] MR5734: mf: Implement audio capture device enumeration in MFEnumDeviceSources().
by Nikolay Sivov (＠nsivov) May 28, 2024

May 28, 2024

Nikolay Sivov (@nsivov) commented about dlls/mf/sac.c: > +static void audio_capture_free_private(void *user_context) > +{ > + TRACE("%p.\n", user_context); > +} > + > +static const struct activate_funcs audio_capture_activate_funcs = > +{ > + audio_capture_create_object, > + audio_capture_shutdown_object, > + audio_capture_free_private, > +}; > + > +HRESULT enum_audio_capture_sources(IMFAttributes *attributes, IMFActivate ***ret_sources, UINT32 *ret_count) > +{ > + static const WCHAR devinterface_audio_capture_wstr[] = L"{2eef81be-33fa-4800-9670-1cd474972c3f}"; > + static const WCHAR mmdev_path_prefix[] = L"\\\\?\\SWD#MMDEVAPI"; It's unlikely those device strings need to be constructed in client library here, I'd expect those id to come from setupapi (DEVPKEY_Device_InstanceId?), if they are not present in mmdevice property store already. If application really depends on those attributes we could add them here as a temporary solution, but if it doesn't, I'd rather prefer a fixme. Device symlink is required for video sources I think, but for mmdevapi you'd use an endpoint id. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/5734#note_71435

1 0

Re: [PATCH v2 0/6] MR5724: widl: Various fixes.
by Elizabeth Figura (＠zfigura) May 28, 2024

May 28, 2024

On Mon May 27 09:51:12 2024 +0000, Rémi Bernon wrote: > What about decoupling this from the definition status? Something like `is_imported`? I'm not sure I see how? If a type is just declared in an import, and defined in the main IDL, then we still want to define it. I don't see how to distinguish those two cases. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/5724#note_71433

1 0

Re: [PATCH v4 0/1] MR5727: riched20: In EM_SETPARAFORMAT protect against out of bound cTabStop values
by Nikolay Sivov (＠nsivov) May 28, 2024

May 28, 2024

Looks fine to me. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/5727#note_71432

1 0

Re: [PATCH v2 0/4] MR5667: winegstreamer: Use GstCaps for wg_parser internal formats.
by Elizabeth Figura (＠zfigura) May 28, 2024

May 28, 2024

> I'm sorry but I've wasted enough time on this. I believe I'm still leaving more than enough time to give some feedback, either commenting on the code or even rejecting the changes for whatever reason you might find appropriate. I'm sorry for not reviewing I find it very difficult to want to review promptly when my review is invariably met with argument, insistence that there is only one correct way to do things, and inability or unwillingness to view things from my perspective. Despite this, I am sorry, and I will try to renew my efforts to review better. If you wish to have me removed as maintainer of winegstreamer, please propose that directly, rather than abusing GitLab to get your patches committed. As far as comment on this patch goes, I'm sorry if this is not quite as baked as I'd like it to be, but: * Renaming "preferred" to "current" is certainly true as far as the functionality of the code is, but it goes against its intent. The point of the "preferred" format is that it should be the one that does not require videoconvert, and therefore offers the best latency. (I know you come at things with a Media Foundation perspective, where the format is pretty much always specified by the application, but for DirectShow it's quite common to plug in a file and autoplug system components up to and including the renderer, so using GStreamer's preferred format is often perfectly feasible.) Hence I would rather see the code fixed so that the "preferred" caps remains what it is, and is not updated. * From what I've understood, getting rid of wg_format by itself does not save that many LoC. I'm still not convinced that it's a desirable change to make. I've already told you my intentions for wg_format, and despite the fact that you dismissed them as false, claiming it's an ad-hoc struct, I still believe that it's a desirable IR. I'm not convinced that the tradeoffs that you're trying to make are good ones. I'm not convinced they're *not*, either, which is part of why I hadn't said anything yet. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/5667#note_71430

1 0

Re: [PATCH 0/2] MR5734: mf: Implement audio capture device enumeration in MFEnumDeviceSources().
by Paul Gofman (＠gofman) May 28, 2024

May 28, 2024

Squad depends on that to properly fill audio capture device list for voice chat and handle capture device name change. The game uses mmdevapi objects for actual capturing so that fully works with this implementation (which is yet missing the actual media source implementation). -- https://gitlab.winehq.org/wine/wine/-/merge_requests/5734#note_71429

1 0