@neuecc Is this the best fix? As the original issue states, the runtime was fine. It was just the analyzer. So do we really need a new concrete class to handle this type (for native AOT perhaps?) or is there a better way to suppress the analyzer (e.g. by recognizing that it implements IList<T> and fits some pattern than an existing formatter/resolver can handle)?

I'm pretty sure this just creates a redundant formatter, and still leaves a gaping hole between what the analyzer detects and what the DynamicGenericResolver is willing to do at runtime. I'm working on a real fix.