Re: [Sidrops] proposed, revised text for Section 6

[Job Snijders <[e-mail]>]

Dear Group,,

On Tue, May 5, 2020, at 21:52, Stephen Kent wrote:
> > I just want to note that while I think you made a logically consistent
> > choice, I do not agree with its intent, in as much as I believe cached
> > state could be used to construct the valid state which the manifest
> > and CRL declare.
> >
> > So, I think you drove to a harder place than I intended. But, I accept
> > this is a choice, and a strong choice.
> >
> > I suspect I am in a minority of one on this. I would not seek to
> > impede a group decision driving to a more narrow constrained
> > definition.
> 
> I believe Job made a good argument supporting your observation, and we 
> can revise the text to take advantage of the cached state. I will note, 
> in passing, that this can introduce variances in local processing based 
> on local cache loss and/or differences in cache refresh timing by 
> different RPs. But, if everyone is comfortable accepting the potential 
> variance, I'm happy to proceed.

About the next period of time: yes, Stephen, please proceed to write down your thoughts. I would like to read your messages (through this mailinglist). I can't compensate you for your efforts with a monetary currency, but I can commit my attention to what you write.

For what it is worth - as member (and dependent) of the community around OpenBSD rpki-client, I observe people are holding the various pieces of the puzzle up against the light in various direction and positions, trying to see how how to align (if possible) the implementation of our RPKI cache software with the spirit of George's observation. Test balloons in the form of POSIX.1 patch(1) files are shared within that community. And it appears it'll be a multi-week effort to resolve the puzzle in a way that all people involved would want to run in production on their EBGP routers.

As SIDROPS are not in an easy working space: the timezones differences are really hard, emails might not arrive, reliable distribution of audio from interim meetings seems challenging, and we all speak an approximation of a common language. Receiving messages in simple English helps improve my participation in this process.

Stephen, my request to you: share directly with the working group what you perceive as the space and "variances", and make strong strong suggestions about the specific single path/vector you perceive as most optimal through the space, and I look forward to comparing notes and observations.

Kind regards,

Job