Await Syntax Discussion Summary

[Centril]

@gnzlbg

The document does not argue why should we block the whole async history on this: we can always add a syntax later, and deprecate the await! macro after that happens.

This knowingly adds technical debt, eventual churn, and obsoletes existing documentation. In particular, I don't expect things to change in 2 years time. The same arguments will be had again and again.

@gnzlbg

For try! that worked ok.

For try! it was socially costly and it still made things obsolete. One notable difference here is that try! is a user space macro whereas await! cannot be. It must be implemented in the compiler which makes the technical debt that much worse.

@gnzlbg

Sure, its not “as great as it could be”, but it would be uncontroversial, it would work, and it would let us do better in the future.

It is not uncontroversial. await!(expr) is a prefix syntax and moreover encourages parenthesis buildup. This same form of argument can be made for expr.await; the await!(expr) form just happens to be what exists today.

@burntsushi

Thanks, but that link isn’t particularly useful unless you’re willing to sift through the ~500 comments in that issue. Were there any interesting conclusions drawn from the experiments that can be summarized?

I don't think many shared conclusions were drawn but it did affirm my view that postfix is readable and hard to miss, that chaining is useful, and that prefix doesn't compose well and introduces temporaries. (I have read all of the ~500 comments but I don't have it in memory)