On Fri Dec 1 19:43:55 2023 +0000, Zebediah Figura wrote:
Would you prefer if I just tell the tests to explicitly rewind the
stream when changing compression mode? [As mentioned in wm_reader.c](https://gitlab.winehq.org/wine/wine/-/blob/master/dlls/winegstreamer/wm_read...), no plausible program reconfigures streams dynamically anyways. That's probably best. Alternatively we could do something like if (pts < last_pts) ++callback->rewind_count; todo_wine ok(!callback->rewind_count); which I think achieves the same goal while being simpler.
Told it to rewind, felt easier and less hacky.
Oddly enough, this seems to have changed some durations from 13930000 to 13920000. I'd guess native's rounding to nearest multiple of 10000 a few times more or less, and the accumulated error from these roundings make it spill over to another millisecond.