ietf-nntp Feedback on the 9/3 nntpext.
Brian Hernacki
bhern at netscape.com
Thu Sep 4 13:36:13 PDT 1997
Larry Osterman (Exchange) wrote:
>
> ----------
> From: Brian Hernacki [SMTP:bhern at netscape.com]
> Sent: Thursday, September 04, 1997 10:13 AM
> To: ietf-nntp at academ.com
> Subject: ietf-nntp Feedback on the 9/3 nntpext.
>
> Here is some feedback on the 9/3 draft:
>
> 4.
> o The miminum timeout is required to be 10 minutes. I
> don't
> think the RFC should specify a required minimum.
> *******
> I disagree - the RFC SHOULD specify a required minimum to allow
> clients to know how often they should poll the server to keep their
> sessions alive. This is consistant with the what is done for POP3 and
> IMAP4.
>
> In particular, it's totally acceptable for a server to NOT
> implement timeouts, however for a server that DOES implement timeouts,
> clients need to know a minimum value below which they won't get > dinged.
For most servers, I suspect not implementing tineouts isn't an option.
Given that, I'd hate to admin a server where I was forced to allow idle
users to consume resources for 10 minutes. If we must have a timeout,
then we should make the minimum sufficiently low. I'd settle for three
minutes.
> 10.1
> o This section specifies that the server should use the
> "first
> unused number" for arriving articles. We may want to clairfy
> this to say
> "next sequential unused number".
>
> ****
> If we change this wording, I'm concerned that clients may expect
> that article numbers are monotonicly increasing, instead of strictly
> increasing. This becomes relevant if a message arrives, and is then
> canceled, a client may be confused when it sees a hole in the article
> id's.
But "first unused" could be read so as to include numbers available
lower than the high water mark.
--brian
More information about the ietf-nntp
mailing list