[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: RFC Errata Delinquents



--On 11. september 2003 11:55 -0700 Bob Braden <braden@ISI.EDU> wrote:

  *>
  *> Bob,
  *>
  *> some thoughts here, none of which may be directly helpful:
  *>
  *> - in order to make those calls for verification, you seem to need to
  *> maintain an updated address archive for authors. Do you in fact
maintain    *> such an archive? If so, how do you get updates?
  *>

No we don't.  We have thought about it, but backed away, since it seems
like a bottomless pit.  However, I believe that in the cases I
forwarded, we DID NOT get email bounces.

in the case of 724, I'm sure you didn't, since all of the authors were listed as "address unknown" :-)


  *> - for "Status: UNKNOWN" RFCs like 724, it seems appropriate to ask
the    *> authors to comment on the technical validity of the errata. But
for a Draft    *> Standard like RFC 2046, it seems to me that if the
errata actually    *> introduces technical change, it needs community
verification, not just    *> author verification. Who defines (or should
define) the procedure for RFC    *> Errata vetting, and where is it
documented?
  *>

OUCH!  I must admit that had not occured to us. Well, the changes
involved are very local and SHOULD NOT alter the meaning, but clarify
it.  But I do see the issue you raise.  I HOPE we can construct a
procedure that is not too onerous.  How about asking the ADs to check
that the proposed fix does not alter the meaning?

that's how we hope it will remain :-) - but the experience with (for instance) the ISO/ITU "implementation reports" on the X.400 standards, where heavy modification was carried out using "errata", clearly shows that there's a lot of slipperiness on this slope.....

a more lightweight mechanism could be to send out "this month's errata" to some list of reviewers (which may or may not include the IESG), and depend on them catching anything "overly ambitious". Erroneous errata can, I assume, be easily removed from the RFC Editor's pages!

  *> - in the interest of the errata mechanism remaining lightweight, I
might    *> suggest something like saying "if the author cannot be
contacted, the    *> errata will be verified with at least one expert in
the field of the RFC in    *> question, and will be posted with a note
saying 'Note - not verified with    *> authors'" after a timeout on the
author request.
  *>

That seems reasonable.  We would be very happy to forward the actual
errata items to the IESG for their expert opinions... ;-))

the generic issue of finding someone whose expertise is relevant to RFC X, for any value of X, has come up in other contexts. No easy solutions (and the IESG doesn't know everything either)....

Appreciate the feedback.

Bob