Re: #1047 Path field delimiters and syntax - status

[Russ Allbery <rra@xxxxxxxxxxxx>]

Charles Lindsey <chl@xxxxxxxxxxxxxxxx> writes:

> Does it? I see where it says that servers and clients MUST accept
> well-formed articles, but where does it say they MUST reject malformed
> ones? Surely they can use "garbage in, garbage out" if they don't
> actually fall over in the process.

3.6.  Articles

   [...]
   For the purposes of this specification, articles are required to
   conform to the rules in this section and clients and servers MUST
   correctly process any article received from the other that does so.

Don't let the MUST in the second half of the sentence blind you to the
first half of the sentence.  Articles sent to another server are required
to conform to the specification of what an article is in the NNTP standard
or the server is violating the NNTP protocol.  This means that a server
can accept malformed junk if it doesn't send it to another server, but at
some point it has to verify the syntactic correctness of the article if it
plans on ever doing anything else with it, including transferring it to
another server or serving it to clients.

Also note all of the other MUST and MUST NOT provisions in the rest of
that section.

> In any case, well-formed in the NNTP-sense means little more than that
> the header has a name, a ':' a SP and a CRLF at the end and uses the
> folding rules properly. No requirement to check detailed USEFOR syntax.

I never said that there was a requirement to check the detailed USEFOR
syntax.  I said that the server has to traverse the entire article headers
regardless in order to verify that the article is syntactically valid.
(For that matter, it has to traverse the body as well.)

> Note that the present USEPRO requirement for relaying agents was changed
> to "SHOULD check for the presence of all the mandatory headers" fairly
> recently as a result of discussions on this list. It used to be rather
> stronger.

Not particularly relevant to the discussion we were having.

-- 
Russ Allbery (rra@xxxxxxxxxxxx)             <http://www.eyrie.org/~eagle/>