Mark S. Miller
Mark S. Miller
Still waiting for @dtribble to clarify what async tail recursion problem he's worried about. But until then, I'm going to argue that none of this is a problem. I'll start...
All looks good. But beware tagged template literals. The same transformation for them would not preserve meaning.
Yes. Except you should delay putting it into the memo until after the harden, so it doesn’t memo if it errors before then. Also protects against reentrancy in the important...
It leaves behind an unhardened and (depending on where the failure is) arbitrarily badly formed template that a second use will retrieve and then use without error. That's worse.
> I actually considered that, and came to the conclusion that identity-preserving store-once behavior is better than masking failures by overwriting a template rather than exposing an incomplete one. ¯\_(ツ)_/¯...
From reading this thread with dim understanding, my sense is that @turadg and @mhofman have not converged yet. I'm inclined to wait for that, and then to have one or...
> From reading this thread with dim understanding, my sense is that @turadg and @mhofman have not converged yet. > I think we're converging by doing both :) What is...
Also `test.after.always`. See https://github.com/Agoric/agoric-sdk/pull/2740/checks?check_run_id=2214861371#step:10:10
I'm listing methods in this issue as I encounter them in trying to switch to ses-ava. If it is easier to fix our code to not use these methods then...