search

ietf-tools / datatracker

The day-to-day front-end to the IETF database for people who work on IETF standards.
https://datatracker.ietf.org
BSD 3-Clause "New" or "Revised" License
624 stars 392 forks source link

RSAB voting: enough votes to pass #7860

Open elear opened 3 months ago

elear commented 3 months ago

Describe the issue

Currently the interface indicates that only one YES vote is needed to pass. That's not right.

For RSAB votes at this stage, and there are stages, any CONCERN would initially block passage, and a document is effectively sent back to the RSWG for further discussion.

If after that, there is STILL a concern, then 3 YES votes are needed to pass.

Code of Conduct

rjsparks commented 3 months ago

I think perhaps you looked at this when it said "Needs a YES". If so, that was stating a necessary but not sufficient condition. The IESG asked this to be called out specifically for ballots (and the ballots here inherited that) as it indicated a potential issue with the work the responsible AD had put into the document so far.

I'll look into modifying (or not presenting) that message when reworking the pass criteria calculations.

If you have it handy, point to where the 3 YES votes is documented (I'd like to leave a comment in the code pointing to it).

rjsparks commented 3 months ago

I found the reference needed: https://www.rfc-editor.org/rfc/rfc9280.html#section-3.2.2-2.12

rjsparks commented 3 months ago

I don't think the current code will prevent the RSAB from doing the right thing when the ballots are in an appropriate state.

elear commented 3 months ago

Hi Robert. The ballot flow is as follows:

A single CONCERN requires that the document be sent back to the RSWG for further discussion. If at some point, the matter is not resolved, then a new ballot is taken on the document, and the next time it takes 3 votes to pass.

So there are two work flows...

rjsparks commented 3 months ago

So, in terms of tooling, you anticipate that the first ballot will be closed and a second, fresh, ballot opened, rather than doing what the IESG typically does with holding a single ballot until all DISCUSS are cleared or a DISCUSS override vote is taken.

This can be accomplished with two ballot types with different rules. They'll need names.

rjsparks commented 3 months ago

I have to add: be careful with trying to put a lot of detail into the software model of a process you are just beginning to exercise. You will likely change that process along the way. Finding the balance of just enough software model is important, otherwise the built thing gets in the way hard when you try to change behavior.

elear commented 3 months ago

In most cases, I anticipate that only the first ballot will be needed.