Sure! It's now better AND simpler. Be sure to see the beta API reference, which is up to date:
https://esotericsoftware.com/spine-api-reference-beta#Animation-apply
MixBlend is now boolean fromSetup for the Animation apply and Timeline apply methods. True means alpha transitions from setup pose to the animation pose. False means alpha transitions from the current pose to the animation pose.
MixDirection is new boolean out. True means mixing out, false means mixing in. This only matters for a few timeline types.
This makes for more booleans in the apply methods, which is harder to read, but it's already a low level API with many parameters so it wasn't exactly easy to before. The simpler concepts allow the logic in every timeline type to be simpler. We tried enums and the mix of enums and booleans in the method signature wasn't great.
Since MixBlend no longer exists, it's gone from TrackEntry.
Also the AnimationState "hold system" that prevents dipping is more robust. It handles all scenarios and you never need to manage it manually, so the TrackEntry holdPrevious property has been removed. In 4.2 and below, certain scenarios of mixing animations on various track could cause a pose to jump instead of mix smoothly, but that shouldn't happen in 4.3+.
A bigger change is how boolean add has improved. It now respects boolean fromSetup. When fromSetup and add are true, the animation is added to the setup pose. AnimationState now uses that so additive animations are never added to the pose from the last frame. If a property hasn't been set on a lower level, the additive animation will get the setup pose plus the animation. You no longer need to have a "base" animation that sets all properties that are keyed on higher tracks by additive animations.
Finally getting around to your actual question, you don't need to set MixBlend, and can't because it's gone. You can set animations on track 1 without an animation playing on track 0 and everything should work. If not, please let us know. AnimationState no longer has special behavior for track 0. It does use an optimization: when track 0 has 1 alpha, it applies the animation more efficiently. This is functionally identical, just more efficient for a very common case.
