On Oct 25, 2021, at 11:34 PM, Carsten Bormann <[email protected]> wrote:
> No document here has to wait for any other document to be published.
Currently:
draft-gharris-opsawg-pcap-02 contains its own list of values for the
LinkType field in the file header;
draft-tuexen-opsawg-pcapng-02 points to
https://www.tcpdump.org/linktypes.html for the LinkTYpe field in the Interface
Description Block.
Note that those two sets must be *the exact same sets*, so, by the time these
become RFCs, either one should point to the other, or both should point to a
common location, such as an IETF registry.
If we go for "one should point to the other", e.g. the pcapng spec points to
the pcap spec, then the one that is pointed to must be published first.
If we go for "both should point to a common location", the common location must
be established first; if that involves publishing a spec that lists the
LINKTYPE_ values, e.g. by extracting that list from the pcap spec and putting
it into a separate spec, and having the registry point to that spec, the same
way that the registry for Snoop link types:
https://www.iana.org/assignments/snoop-datalink-types/snoop-datalink-types.xhtml#snoop-datalink-types-2
points to one or more RFCs for various link-layer types, then the initial spec
with the link-layer types would have to be published at the same time as, or
before, the pcap and pcapng specs are published.
I would vote for "both should point to a common location" so that neither the
pcap nor the pcapng spec says "there is *the* list of link-layer types" - it
points to a registry, and as more types are added to the registry, more specs
can be published define them.
_______________________________________________
OPSAWG mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/opsawg
