> On Feb 3, 2020, at 1:39 PM, Ben Schwartz <bemasc=40google....@dmarc.ietf.org>
> wrote:
>
> I thought in Case E the goal was that each delivery agent keeps its ticket as
> persistent per-process state, to avoid re-querying the global cache.
I don't recall discussing such a scenario. In Postfix the cache is shared.
There is not sufficient process affinity to particular destinations to
make such a cache design be viable, and the processes have limited
lifetimes, with new ones spawned as needed. So whatever E might be,
a per-process cache is not the use-case I'm working with.
I'll read the post more closely over the next few days and attempt to
summarize where I think we are, and propose a pull request to at least
clarify the issues that motivate potential tweaks to the design.
We're starting to lose focus, and I think something more concrete in
writing will help.
--
Viktor.
_______________________________________________
TLS mailing list
TLS@ietf.org
https://www.ietf.org/mailman/listinfo/tls
