[ACTION-897] best current practice regarding the withrawal of X- / partial

I am half-way through the action resolution. Here are the provisional results.

a) The registration of header fields at IANA follows a procedure specified by
the IETF (RFC 3864). Fields can be registered as provisional or permanent.
Provisional fields are relatively easy to register. Permanent ones require them
to be based on a published standard. 

b) Provisional fields can only be of type "provisional", whereas permanent ones
can be "historic", "obsoleted", "experimental". Fields are attached to a
protocol.

c) The IETF draws the attention towards avoiding synonyms with different 
semantics, and prohibits reusing existing MIME fields for a different purpose,
but is silent on having fields with different names but the same semantics.

d) So far, I could not find trace of a definition of or good practices about
X-* fields. It appears that this is an established practice, not a standard. 
If anybody is wiser, let me know (preferably with a reference).

e) It seems possible to rely upon the registration procedure of 3864 to retract
a field.

So for the moment, there is no information about
1. registering or not X-* fields;
2. having a registered nnn field deployed in parallel with a X-nnn field with
the same semantics;
3. deprecating fields, whether of the form X-* or not.

However, during my searches I discovered that a lively discussion took place
around just those topics (What are X-* fields exactly? How to obsolete fields?)
in the IETF mailing lists, in the context of the revision of RFC2822 to RFC5322.

I intend to look further and to ask the IETF about the above subject. If I get
an answer rapidly, I could present the final word in one week.


E.Casais

Received on Tuesday, 20 January 2009 14:21:16 UTC