Re: [Ice] RFC8445 triggered checks cancelation clarification

[Christer Holmberg <[e-mail]>]

Hi Sergey,

>There a couple of points to clarify about triggered checks.
>
>1. Since a triggered check can cancel current in-progress transaction it may cause a long loop of cancels and rechecks on both sides.
>
>It's a somewhat rare situation but still possible when both sides send triggered checks to each other simulatentously.
>To avoid this I think triggered checks should not be cancelled by other triggered checks on the same pair.
>
>2. The document states Cancel should wait for a response but avoid handling errors as well as don't do any retransmissions.
>
>From the document it's not clear where a success binding request from a cancelled transaction has to be handled as usually or not.
>If yes it also remedies 1 quite a bit.

The idea is that a success binding request is handled as usually. Otherwise, why would you wait and process it? I do agree that it could have been explicitly said.

Regards,

Christer