ietf-nntp Response codes - debugging, authentication & unknown

[Russ Allbery]

Lee Kindness <lkindness at csl.co.uk> writes:

> !      9xx - Reserved for a possible future debugging extension.

[...]

> !      x9x - Private debugging output

So, um, given that absolutely no one has used the debugging codes so far
as we've been able to establish since NNTP was created, what's the
rational for reserving *yet another* set of codes for it?  If someone
plans on really using this, I suppose I can see this, but no one appears
to actually be *using* the debugging codes.

I'll bet there's been more thought expended on them on this working group
than has ever been expended on them by NNTP software developers.

>        x0x - Connection, setup, and miscellaneous messages
>        x1x - Newsgroup selection
>        x2x - Article selection
>        x3x - Distribution functions
>        x4x - Posting
> !      x7x - Nonstandard (private implementation) extensions
> !      x8x - Authentication

I'd rather see your x7x space merged with x9x rather than adding another
reserved space.

>        The exact response codes that can be returned in response to a
>        given command are detailed in the description of the keyword
> !      that is the first part of the command. If a client receives
> !      an unexpected response code from a command then it MUST make
> !      a 'best guess' on how to treat the response - typically the
> !      first digit is used for this purpose.

I would be very leery of specifying anything in the standard as a MUST
without fully specifying it.  I don't think it's kosher to say that you
MUST make a "best guess."  The vagueness of what the client is supposed to
do forces this down to at least a SHOULD unless we plan on fully
specifying what the client should do with each class of unexpected
response.

>        Response codes not specified in this standard MAY be used for
>        any installation-specific additional commands also not
> !      specified. These SHOULD be chosen to fit the pattern of x7x
> !      specified above.

Although if those extensions are specified in a standard, they shouldn't
use the x7x space, so if the intention is to develop an extension for
standardization, it might not be a good idea to start in that space.

I agree with Stan's suggestion of saying that x8x responses are intended
for extensions related to authentication and authorization.

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