4 jul 2015 kl. 16:29 skrev John C Klensin <john-ietf(_at_)jck(_dot_)com>:
--On Saturday, July 04, 2015 11:34 +0200 Patrik Fältström
<paf(_at_)frobbit(_dot_)se> wrote:
They're still opaque identifiers, so the format isn't
important. I don't know how to make that any clearer.
Because they have been published, we immediately have a
question about persistence. We can now, from my perspective,
not change the format as they have already been included in
the RFC Index. Its a persistence issue.
We COULD have changed the format, discussed it, or whatever,
but that point in time is passed.
This is one of the points I've been trying to make. The issue
for me is not, and hasn't been, whether we should assign DOIs to
the RFC Series or not -- I really don't care. And it isn't
"surprise" because, as Heather points out, it wasn't hard to
know that a decision had been taken to "do DOIs" and that _some_
DOI arrangements were underway. I am more concerned about the
decisions that lead to this "decide first, make irrevocable
decisions about details (or just let them happen), deploy, and
_then_ ask for review. And that leads exactly to Patrik's
comment about what we "COULD" have done above.
Perhaps the RSE interpreted the lack of feedback on those announcements on
rfc-interest as indication that nobody gave a hoot about the details. I
certainly would have.
Picking just one example, Heather's note pointed to the
September 2014 SOW
(http://iaoc.ietf.org/documents/Announce-RFC-DOI-SOW-3Sep14.pdf
if anyone wants to look at the action text rather than the
announcement) but it does not contain a word about the choice of
the particular format for the suffix. Similarly, not only was
that March 2014 call for review a call for comments to the
rfc-interest list (not even to the IAB) but there is nothing in
it that implies that it is a major decision point about the
details -- at that stage in the development of things, or even
at the time of the September SOW, it was reasonable (at least
IMO) to interpret the DOI format in the draft as a placeholder
to be discussed later.
Normally, "comment to a WG list" or "comment to other than the
IAB or IETF lists" implies a request for review and comments but
that the community will get another chance to review before the
final decision point. And normally we don't treat SOWs, even
SOWs that _do_ include the precise details, as decision
documents about issues that, once decide, have broad impact and
are irreversible. If the intention here was to make those
announcements of decision points, then they should have been,
again IMO, much more clear about it.
Sure, people could have noticed those lines and said "I
certainly hope you are not planning to use DOI: 10.NNNNN/rfcNNNN
as the suffix format" but should have been no expectation that
not doing so was support for a final decision, any more than
failure to study an I-D that lies largely outside one's major
area of interest and comment, to a list associated with it,
creates a binding decision that cannot be questioned or even
reversed on IETF Last Call.
Had someone asked the community the question "does DOI:
10.17487/RFCnnnn look like a satisfactory suffix format", we
COULD have had that review as Patrik suggests (and many others
of us have suggested in other ways).
I will defend a decision like that suffix format one as one that
the RFC Series Editor can make and announce to the community,
subject to whatever oversight the IAB (directly or via the RSOC)
chooses to apply. But that implies that the RSE needs to be
accountable for such decisions -- both their substance and the
details of the analysis that went into them. What I expected to
hear from Heather was "these are the tradeoffs we considered and
why the decision was made this way" not "this should not have
been a surprise".
While it is not yet irreversible, I think one can have a very
similar discussion about the messaging involved relative to
other work and identifiers. For example, I think some of this
discussion, including the RSOC and IAB openness to adding more
identifiers, ought to be in this document, making clear that
DOIs may be the first (actually second, given RFC 2648) in a
series rather than an IETF/ IAB/ RFC Editor commitment to one
particular system. That sort of issue is what calls for
comment are supposed to be about.
john