[rfc-i] RFC citations committee I-D issued

RJ Atkinson rja.lists at gmail.com
Fri Feb 11 08:01:31 PST 2011

Earlier, Julien Reschke wrote:
> What we SHOULD recommend is that the ID name SHOULD be included, and 
> the full data (as opposed to Year/Month) MAY be included.

(NOTE:  From discussion elsewhere, and the broader context above,
I believe "full data" above is a typo, and that "full date" was 
what was actually meant. :-)


I agree that the full ID filename normally SHOULD be included
in the citation of any I-D.  

There are historical IDs (e.g. draft-odell-gse-*) that I've cited
in peer-reviewed papers this decade.  Frequently, readers of my papers
have asked me where one can find a copy of that ancient/expired I-D.  
So even in the research world, many folks like to actually read a copy 
of the cited paper/document.  We ought to be helpful to readers,
rather than needlessly obstructive.

There appears to be even more interest from the standards/products
world in reading historical documents.  Someone else mentioned
patent issues, which is a great example.  Other examples include
IDs published by some vendor that never were published as RFCs,
but contain insights useful for enhancing interoperability.


As near as I can tell, many RFCs are only published with a month
and year provided.  However, some other RFCs (e.g. humourous
ones published on 1st April) have included day, month, and year.

As near as I can tell, all modern I-Ds include day, month,
and year.  

So I believe the entire *available* date always MUST be included, 
both for I-Ds and RFCs.  This recognises that for some RFCs 
the "available date" will be day+month+year and for other RFCs 
that will be simply month+year.

This is subtly different from what Julian wrote above.  I think
the precise difference matters, especially for 1st April RFCs, 
even though there are credible reports that RFC-1149 has been deployed 
at least once.  <http://www.blug.linux.no/rfc1149/>



More information about the rfc-interest mailing list