By falling back to the default of 40% of track duration in such cases.
Also added a safety for the restart point exceeding acceptable bounds in
case of a non-zero offset.
Closes https://github.com/ppy/osu/issues/33308.
This is an effort to improve general performance at song select. At
least on the metal renderer, I can notice very high draw frame overheads
related to texture uploads.
By reducing the size of the texture uploads to roughly match what is
actually being displayed on screen (using a relatively inexpensive crop
operation), we can bastly reduce stuttering both during initial load and
carousel scroll.
You might ask if it's safe to disable mipmapping, but I've tested with
lower resolutions and bilinear filtering seems to handle just fine.
Bilinear without mipmaps only falls apart when you scale below 50% and
we're not going too far past that at minimum game scale, if at all.
Will still keep the override in `ClockBackedTestWorkingBeatmap` because
it still relies on a local track store and will fail the moment it uses
a non-virtual track.
A lot of tests are using test resources that populate the length field,
but do not populate hitobjects. The general expectation is that
components should be using the cached length in cases where hitobjects
are not relevant, but `GetVirtualTrack` was doing its own local
calculation.
This could cause tests to fail due to `MusicController` changing track
in the background.
After the recent changes introducing cancellation support to
`WorkingBeatmap`, it turned out that if the cancellation support was
used, `GetPlayableBeatmap()` would raise timeout exceptions rather than
the expected `OperationCanceledException`.
To that end, split off a separate overload for the typical usage, that
catches `OperationCanceledException` and converts them to beatmap load
timeout exceptions, and use normal `OperationCanceledException`s in the
overload that requires a cancellation token to work.