--On Thursday, December 19, 2024 09:46 -0600 Jean Mahoney <jmaho...@amsl.com> wrote:
> Hi all, > > On 12/18/24 9:06 PM, Stephen Farrell wrote: >> >> Hiya, >> >> On 19/12/2024 02:24, Brian E Carpenter wrote: >>> >>> I also don't think that the current errata system really >>> creates a >>> big workload. We have a big backlog, but that seems to >>> be due to >>> low prioritization (and I think that's a collective error >>> ). >> >> I think that's an erroneous diagnosis of the problem. The current >> system is so bad, for all concerned, it causes most people to >> de-prioritise errata processing would be my take. > > [JM] I have heard from many verifiers that it is difficult to build > the context to evaluate a report for an RFC that the verifier is > not familiar with. This task gets harder with older RFCs that no > longer have active working groups or reachable authors. > > What can a new system do to help a verifier process these reports? > > Should the system not allow reports on Legacy / Historic / Obsolete > RFCs? That would reduce verifier burden and could be turned into a > policy statement. [With my RPC hat on: We should attempt to treat > RFCs as uniformly as possible. Note that obsolete RFCs can still be > widely implemented. Switching to my personal hat: We should > encourage readers to read the newer RFCs. We should reduce > attractive nuisances. For example, Legacy RFCs receive far more > junk reports.] > > There seems to be an unwritten policy that if the verifier can't > verify a report, they have to keep the report open until somebody > (e.g., the next AD) figures out. The current final statuses > (Verified, Rejected, Hold For Document Update) may encourage > keeping reports open. That is, a verifier may not feel comfortable > rejecting a report if they can't say with certainty that the report > was incorrect. > > A new way to close a report might help: "Closed - insufficient > information". The verifier asked for assistance, but didn't receive > any. The request and non-answer would need to be visible in some > way (a link to a mail or issue thread in the report's notes). A > policy statement here would be along the lines of "If the verifier > cannot verify the report after asking for assistance, the verifier > may close the report." > > To deal with the backlog, should we just close reports that have > been open for a long time? The definition of "long time" would need > to be determined. Or should the RPC host a hackathon at an upcoming > IETF meeting where verifiers work through as many open reports as > possible? We could bring pizza :-) Jean, The more I think about this, the more I find myself aligned with Christian's view -- the main focus of the system should be less about "error" identification and correction than about providing input for future revisions of a document and maybe at least a clue about when such a revision is needed. That perspective would largely eliminate my concern about verification (and possibly some rejections) without the same level of consensus that was associated with the original approval and publication. More important, it would reduce the workload expectations on the IESG and maybe that on the RPC. We might even get the "errata" terminology out of our vocabulary and start talking about "request for update or revision". That adjustment in focus would have the useful side-effect of providing an easy and immediate answer to several of your questions. For example, we don't accept requests to update Historic RFCs because the community (presumably) has decided they are no longer of interest. For even more obvious reasons, we also don't accept requests to change documents that have already been obsoleted. Even for a widely implemented obsoleted RFC, if a fix is needed, the right way to apply it would be a clarifying update to the obsoleting document (any change to the obsolete one would almost certainly create a confusing fork and require even more work later). Update requests for documents that were associated with active WGs would go direct to the WGs. For individual submissions, the authors would be notified. Otherwise, I'd hope we could keep an easily-accessed table that would show how many requests/ comments have been received on a given RFC, when the earliest and most recent ones, were, etc., but that the IESG was not required to evaluate that list until the usual process for proposing a update or replacement for and RFC was actually initiated through community discussion and, typically, an I-D. The notion of "closed" would yield, at the extremes, to "active discussion underway" and "generally ignored" but neither of those would be an official category requiring any action. The tracker should show "discussion initiated" but not "errata verified and applied". As another side effect, the idea of people finding errors and submitting errata reports to get credit would vanish in favor of "started (or tried to start) discussion about perceived problems" -- probably less useful to them and more useful to the community. And, fwiw, I see the above outline as consistent with many of the ideas in draft-farrell-errata, including "chuck it out and replace it entirely", even though the change in focus would require some revision to specific text. best, john -- rswg mailing list -- rswg@rfc-editor.org To unsubscribe send an email to rswg-le...@rfc-editor.org
