ietf-nntp NNTP AUTH draft update

[Russ Allbery]

Chris Newman <chris.newman at INNOSOFT.COM> writes:

> (2) reserve x8x for authentication and create a new response code set
> for private extensions.  As a side effect, this retroactively declares
> all private extensions using x8x for purposes other than AUTHINFO as
> non-compliant -- retroactive non-compliance declarations are generally
> frowned upon in the IETF.  This choice would only be practical if a
> strong statement can be made that the retroactive impact will be
> extremely minor (e.g., are x8x error codes used _only_ for
> authentication?)

I don't have a particular constructive suggestion, just the observation
that this seems to be more evidence that reserving a chunk of namespace
for "private extensions" when the namespace is part of interoperability
concerns is a bad idea in a lot of situations.  Things like this seem to
always happen, where someone implements a good idea using private
namespace, all the clients are modified to use that private namespace, and
then we have all sorts of problems when we try to standardize it.

Happens all the time with X- headers too, and is happening in NNTP with
the various X* commands.

I wonder if anything really would have broke if we'd just called the
command OVER in the first place, rather than XOVER.

-- 
Russ Allbery (rra at stanford.edu)         <URL:http://www.eyrie.org/~eagle/>