On Tue Sep 26 06:16:29 2023 +0000, Rémi Bernon wrote:
I meant the caller of `wg_transform_read_data`, which receives the format change error status. The quartz flavor, `wg_transform_read_quartz`, doesn't support it and prints the error message. It should be possible to support the format change in quartz_parser `transform_sink_receive`, keeping the current wg_format around and passing a wg_format pointer to that function, *then* comparing whether the two formats are actually compatible, and call `wg_transform_set_output_format`.
Ah, that makes more sense.
Okay, I can do that, but this is straying pretty far away from the purpose of this MR. And I don't know how well setting media types on samples works in practice - Microsoft's docs are pretty scary and vague [1]. Hopefully we won't find out. (If we do, that belongs in a separate MR.)
[1] "To make a significant change to the media type, the pins might need to reconnect and renegotiate the media type." https://learn.microsoft.com/en-us/windows/win32/api/strmif/nf-strmif-imedias... What counts as significant is, as far as I can determine, not defined anywhere.