I'll admit I can't weigh on whether there's a plausible strategy for implementing backwards frame step in VLC without violating any of their sacred invariants. Maybe the existence of other players with that feature implies that the VLC devs just aren't seeing it. But without that knowledge, I'm left to trust the developer's presumably deep knowledge of their own architecture when they say that it just doesn't 'fit'. It seems more likely that the other players can do it because they're architected differently, or they make a different (smaller) set of guarantees about which formats they will play back in which scenarios.

>Never let the perfect be the enemy of the good. If people don't use your product, it doesn't matter how right and pure of vision you are.

This is great advice... for a business. VLC isn't a business, it's a useful thing that exists. It gives people fulfillment to donate their time to keep it useful and relevant. That fulfillment is going to be directly proportional to how elegant the design is and how much leverage they have, i.e. small amount of good code provides wide amount of functionality for many happy users. The more that maintenance of VLC degrades towards "brutal hand-crafted specialization of huge switch cases", the less well-maintained it's going to be. This is not an excuse, it's a statement about human nature, something akin to economics.

It's possible that mpv picked a better abstraction and they can provide a strictly better application for the same or less software maintenance cost. But I tend to suspect that the differences arise from levels of support, not better design. That's just a spider sense though.