Bob Braden wrote:
Ray,
I realize you have suggested this before, but unfortunately it is a bad
idea. When document A is held up by a normative reference to B, A and
B must be published together. It is not at all unusual for some
inconsistency between A and B to show up at the very last minute.
Therefore, it is unreasonable and undesirable to say that A must be
fully ready to publish before B is done.
Furthermore, it seems to be forcing a particular resource allocation on
the pub service with no gain, but rather a loss of flexibility and
therefore efficiency. If you accept that the pub service has the same
goal that the IAD/IASA/IETF has -- publishing documents as quickly and
efficiently as possible -- then imposing arbitrary scheduling rules
like this cannot be a useful thing to do.
I suspect we may be over-designing a bit here, but I'd observe
that holding documents for missing references at the entry to
the queue (which I believe is current practice) may not be ideal
either - although it may be a perfectly sensible pragmatic
choice when the workload is high. So while validation of
references is a requirement, I'm not sure we should
write down *when* it has to be done. When I buy a Model T Ford,
I don't specify where in the production line the wheels
have to be attached.
Brian
_______________________________________________
Techspec mailing list
[email protected]
https://www1.ietf.org/mailman/listinfo/techspec
