ietf-nntp Re: DRAFT Basic-Format v0.2

[Stan Barber]

> In sob at academ.com (Stan Barber) wrote:
> 
> >> Is is really the case that your latest draft allows the response to an
> >> ARTICLE command to contain NULs and isolated CRs and LFs? And with no
> >> limit on the length of a "line"? And with no requirement to have a CRLF
> >> anywhere (even at the end of an article - that is the difficulat point, I
> >> guess)?
> 
> >If you would specifically quote any text in the current draft that supports
> >your assertions then we can look into fixing them. Obviously, this is
> >confusing you and is certainly not supposed to be in the draft.
> 
> I was asking questions, not making assertions.

Your questions imply that you have read the documents and are attempting to
clarify what you have read. Such clarifications must include an assertion of
your understanding what the document says . Otherwise, the questions have no
frame of reference.

The rest of your note seems to indicate that you know the answers to your
own questions. Were you attempting to be rhetorical?

> 
> >The specific line length limitations are on the length of commands sent to 
> >server by the client. 
> 
> AIUI a command sent to the client is limited to 512 octets (a bit on the
> shorside for NEWNEWS commands, but that is another matter).

No. The limit is on what can be sent to the server by a client, not on what
the client can receive from the server.

> As regards complete news articles coming from the server, all I can see in
> your draft is in section 4 where it speaks of multi-line responses. The
> first response line, terminated by CRLF, is presumably a response code.
> After that, there may be any number of lines terminated by CRLF, but there
> is no mention of any maximum line length, so presumably one line of a
> million bytes is in order. There must be a "last line" terminated by CRLF,
> followed by a line with a single "." on it (dot stuffing required as
> usual). There is no prohibition of NULs or of isolated CRs and LFs. So it
> would seems that an article in pure binary would be in order provided it
> has a CRLF at the end. Is that correct?

Yes.

> AFAICS (10.3.1, 10.3.2) an article sent TO the server uses exactly the
> same rules.

Right.

> 
> BTW, if there WERE to be a limit on line length, it ought to be at least
> 998 octets plus CRLF to conform with established usage (see DRUMS, for
> example).

We discussed this at IETF 40 and decided not to do that. Currently, there
is no limit to line length in server responses.

> Sent to usenet-format and ietf-nntp lists. Note Reply-To: usenet-format
> where this discussion originated in a suggestion from Brad that news ought
> to be "binary clean" (I personally disagree with that idea BTW).

I am copying my response to ietf-nntp as well since it may help others on
these issues.

-- 
Stan   | Academ Consulting Services        |internet: sob at academ.com
Olan   | For more info on academ, see this |uucp: {mcsun|amdahl}!academ!sob
Barber | URL- http://www.academ.com/academ |Opinions expressed are only mine.