4 Jul
2022
4 Jul
'22
6:58 p.m.
On Mon Jul 4 18:39:42 2022 +0000, Nikolay Sivov wrote:
I see, thanks for checking that. Could you make it work this way? It's probably having two paths to get the handler, maybe even with E_NOINTERFACE if none of the two interfaces are implemented. The reason I think it's better to follow it here is first because sink is not required to expose the handler this way, thus explicit method. And second because in topology context you don't normally set something that doesn't support IMFStreamSink as an output node object.
I don't think there's really any need to make it support that case after all, it was mostly to make the tests easier to write, and I thought it was correct, but I'll write the tests with a fake IMFStreamSink instead.