Subject: Re: TCP_NODELAY in telnet (Re: CVS commit: src)
To: Jonathan Stone <jonathan@DSG.Stanford.EDU>
From: David Maxwell <david@crlf.net>
List: tech-net
Date: 06/18/2003 18:23:39
On Wed, Jun 18, 2003 at 03:07:34PM -0700, Jonathan Stone wrote:
> >I read it der Mouse's way.
>
> Then mabye you (and Mouse) SHOULD acquire better reading skills. :-/.
I may choose to ignore that SHOULD.
> Note that "SHOULD implement the Nagle algorithm", in this context,
> clearly means "SHOULD do the Nagle algorithm". If we default Nagle to
> disabled, we are in violation of the SHOULD, as written and as it's
> been interpreted by the TCP community for the last 14-odd years.
1. That's why standards have to be specific. If they meant 'do' they
need to say 'do', rather than 'implement'.
2. It's still only a violation of a SHOULD.
3. Neither Mouse, nor I, have _ever (TMK)_ said NetBSD SHOULD make this
change - only discussed the interpretation of the RFC text.
4. "as written and as it's been interpreted by the TCP community for the
last 14-odd years" would not mean that it should never be revisited, if
someone did wish to consider it.
> An implemntation MUST provide a way to not do Nagle. Consider a TCP
> implementation which fails the SHOOLD: it doesn't do Nagle by default,
> therefore the `way' (on this TCP implementation) which MUST be
> provided to disable Nagle, is just a no-op.
That's exactly what I wrote.
> If you really want to ask Bob Braden, go ahead. But please don't
> associate the name NetBSD with your request, as that may embarrass
> (some of) the rest of us.
This seems a bit more agressive of a response than I would have expected
to a simple logic discussion message. It seems to be an implicit insult
to my intelligence.
--
David Maxwell, david@vex.net|david@maxwell.net --> The only difference I see
between voodoo and marketing research is that voodoo sometimes works!
- Leonard Stern