--On Wednesday, February 02, 2011 02:42:40 PM -0600 Andrew Deason
<[email protected]> wrote:
I'm not entirely clear on if it's appropriate to have the registrar
maintain the flag values for the new RPC (as opposed to just specifying
them in the draft). To me, this seems similar to assigning code points,
but to my knowledge the registrar doesn't maintain such values for other
RPCs currently.
There are plenty of existing cases where the meanings of particular bits
and fields could be represented in registries but presently are not. For
example, I don't think we ever set up a registry of volume dump tags, but
that's clearly an appropriate subject for one.
Generally, a registry is unnecessary for things which can/will never
change, or for things which will change or be used only by producing the
one-and-only next version of a protocol. For example, when defining the
transport of Kerberos over TCP, we left the high bit of the message length
reserved/MBZ for use by a future protocol revision. Later we used that bit
to add an extension mechanism (which does have a registry of extensions).
I haven't read the document in question, so I can't comment on whether
creating a registry in this particular case is appropriate. But certainly,
don't reject it out of hand just because we don't have any other such.
-- Jeff
_______________________________________________
AFS3-standardization mailing list
[email protected]
http://lists.openafs.org/mailman/listinfo/afs3-standardization
