Re: [PATCH v3 0/8] MR8184: Draft: mf: Implement thinning.

[Nikolay Sivov (＠nsivov)]

On Sat May 31 14:15:02 2025 +0000, Charlotte Pabst wrote:
> I don't see how it's unrelated to test that thinning actually does what
> it promises, i.e. deliver only keyframes (here done by checking the
> timestamps because the MFSampleExtension_CleanPoint attribute isn't
> actually reliable in winegstreamer yet - GST_BUFFER_FLAG_DELTA_UNIT gets
> stripped by the decoder). And there are already existing tests in this
> file (i.e. test_sample_grabber_orientation) that depend on source behavior.

Thinning at media session level doesn't promise anything, it's up to sources to do something with it, or ignore it. Source might even request thinning on their own, along with a different rate value.