Hi,

thanks for pointing that out. That is indeed an interesting use case.

However, I still think that some small debugging improvement should not stop you write more readable and consistent code.

The engine is evolving based on the language use. Multiple times we witnessed that previously slow (or inconvenient) syntax became faster in the next version. I refuse to use a structure or syntax that is wrong just because it is faster in the current engine. While it probably exists, I haven't came across a performance loss that would justify that.

So, saying that avoiding return await is ‘very wrong’ is, well, very wrong itself.

P. S. It seems that await prevents tail call optimisation, which could hurt the performance.