[Top] [All Lists]

Re: I-D ACTION:draft-shafranovich-feedback-report-01.txt

2005-05-27 09:52:10

On Fri, 27 May 2005, Charles Lindsey wrote:

 "william(at)" <william(_at_)elan(_dot_)net> writes:

On Wed, 25 May 2005, Yakov Shafranovich wrote:

Any particular reason why these fields were not included in the initial
header registry? Sounds rather strange to me.

3. If you don't want to register several dozen fields one-by-one then
   it'd have to be proposal to modify header field registration system
   and create special new subtype of header registry for "Prefix-x"
   (which I'd still like to do but can't quite get to it).

Ok I should have made clear that "x" does not mean in traditional smtp
sense "x-" header field. I meant that I'd like to be able to register
entire set of "Prefix-headerfieldname" fields all together without specifying each and instead make it based on that headerfieldname would already be registered and "prefix" changes its use in some pre-determined way.

I thought the whole idea of X-headers was that they would NOT be
registered, so that people can invent them ad hoc as needed (however, it
needs to be understood that they are primarily there for human

My understanding about "X-" is that they are for use in local-system
specific data (for use by other local systems or users) and for local experiments (local can extend to more then one organization) and that if they are for human consumption or machine consumption or both does
not matter. And when experiment wants to become standard then they
define full header field without "x".

So all the Original-* headers should really be X-Original-* headers.

Only if the data is meant entirely for use by local system and they
are not meant for reading by some human on remote network when looking
at email message info and what happened to it.

My view of it all is that Original-* header fields should really be defined as trace fields - that is really how they are used right now
if the data is meant to go outside the system (i.e. when using "X-"
may not be appropriate). They are also used at least in one instance
as non-header fields like what Yakov proposed (and that is why I said
to him on different list that perhaps Original- name should not be
used by feedback-report so as not to create confusion).

William Leibzon
Elan Networks

<Prev in Thread] Current Thread [Next in Thread>