W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > September to December 2000

Re: Conformance Test for HTTP 1.1

From: Keith Hoffman <hoffmankeith@hotmail.com>
Date: Tue, 10 Oct 2000 12:04:09 -0500
To: "Mark Nottingham" <mnot@mnot.net>
Cc: "Carl Kugler/Boulder/IBM" <kugler@us.ibm.com>, "Miles Sabin" <msabin@cromwellmedia.co.uk>, <http-wg@cuckoo.hpl.hp.com>
Message-ID: <LAW-OE10K5LqN33HFnw00001349@hotmail.com>
The simplest answer as to why this isn't good is that it's outside the
charter of the IETF.  This organization is here to create standards.  Not to
validate/judge whether someone is compliant with them.


----- Original Message -----
From: "Mark Nottingham" <mnot@mnot.net>
To: "Caveman" <hoffmankeith@hotmail.com>
Cc: "Carl Kugler/Boulder/IBM" <kugler@us.ibm.com>; "Miles Sabin"
<msabin@cromwellmedia.co.uk>; <http-wg@cuckoo.hpl.hp.com>
Sent: Tuesday, October 10, 2000 1:55 PM
Subject: Re: Conformance Test for HTTP 1.1


>
>
> Err, could you give a more solid demonstration as to why this is not good?
>
>
> On Tue, Oct 10, 2000 at 09:23:10AM -0500, Caveman wrote:
> > Carl,
> >
> > Once we start doing any kind of compliancy checking we face the
proverbial
> > "slippery slope".  What comes next?  Seperate tests for things that MAY
be
> > done according to the specs?  Things that SHOULD be?
> >
> > I think the best thing to do is stay out of the compliancy checking
business
> > all together.
> >
> > Thanks,
> >
> > Keith
> >
> >
> >
> > ----- Original Message -----
> > From: "Carl Kugler/Boulder/IBM" <kugler@us.ibm.com>
> > To: "Caveman" <hoffmankeith@hotmail.com>
> > Cc: "Mark Nottingham" <mnot@mnot.net>; "Miles Sabin"
> > <msabin@cromwellmedia.co.uk>; <http-wg@cuckoo.hpl.hp.com>
> > Sent: Monday, October 09, 2000 2:39 PM
> > Subject: Re: Conformance Test for HTTP 1.1
> >
> >
> > >
> > > >
> > > >I just want to put my two cents into this conversation:
> > > >
> > > >I think the idea of doing compliancy testing is great.  And the idea
of
> > > >having one "check everything test" is also a good thought.  However,
how
> > > do
> > > >we guarantee that the test scenarios created are actually following
the
> > > >"specs"?
> > > >
> > >
> > > I was thinking along the lines of a script (or script fragment) for
each
> > > MUST in the spec.  MUSTs are supposed to be verifiable, right?  All
> > > compliant implementations, regardless of manufacturer/developer, must
do
> > > the MUSTs, right?   Using scripts makes it easy for people to inspect
a
> > > script and correct it if it isn't according to spec.
> > >
> > > >I think this is something better left to outside agencies to address.
> > The
> > > >testing game tends to get to be too industry biased.  Whether
> > intentionally
> > > >or not you will see tests similar to this proposed one done and get
> > totally
> > > >different results depending on who does it.
> > > >
> > > >I know this actually sounds like a good argument to create a
"standard
> > > >test", but in my opinion this leads the doorway too wide open to
start
> > > >skewing the tests in favor of one manufacturer/developer vs. another
one.
> > I
> > > >realize that there are currently many industry leaders involved in
this
> > > >organization and they provide valuable insights.  However, they are
just
> > > >involved in the CREATION of standards, not in judging the conformance
to
> > > >them.
> > > >
> > > >In short, while this is a good idea with the best interests of
everyone
> > in
> > > >mind, I think this is probably stepping outside of the charter of the
> > > >organization.
> > > >
> > > >-kh
> > > >
> > > >----- Original Message -----
> > > >From: "Mark Nottingham" <mnot@mnot.net>
> > > >To: "Miles Sabin" <msabin@cromwellmedia.co.uk>
> > > >Cc: <http-wg@cuckoo.hpl.hp.com>
> > > >Sent: Friday, October 06, 2000 11:30 AM
> > > >Subject: Re: Conformance Test for HTTP 1.1
> > > >
> > > >
> > > >>
> > > >>
> > > >> I think proxies are the biggest target, because they're so hard to
> > > >implement
> > > >> correctly, and so much more complex. In my experience, there's a
fairly
> > > >wide
> > > >> variance in how implementors choose to interpret the spec.
> > > >>
> > > >> Of course, once you do one for proxies, it's relatively easy to get
> > client
> > > >> and server test suites out of it.
> > > >>
> > > >>
> > > >>
> > > >> On Fri, Oct 06, 2000 at 10:24:14AM +0100, Miles Sabin wrote:
> > > >> > Mark Nottingham wrote,
> > > >> > > I've lately been considering starting discussion of
> > > >> > > development of something within the W3C, as it was involved
> > > >> > > in the development of the HTTP, and has an established
> > > >> > > history of developing similar tools (although I'm not sure if
> > > >> > > W3C can formally commit resources).
> > > >> > >
> > > >> > > If anyone has any thoughts about this, please share them,
> > > >> > > because I'd like to get this moving.
> > > >> >
> > > >> > This sounds like a fine idea (tho', as you say, it's an open
> > > >> > question whether or not the W3C would be able to commit
> > > >> > resources).
> > > >> >
> > > >> > Do you have any particular emphasis in mind: server, clients,
> > > >> > or proxies, or all equal weight on all?
> > > >> >
> > > >> > Cheers,
> > > >> >
> > > >> >
> > > >> > Miles
> > > >> >
> > > >> > --
> > > >> > Miles Sabin                       Cromwell Media
> > > >> > Internet Systems Architect        5/6 Glenthorne Mews
> > > >> > +44 (0)20 8817 4030               London, W6 0LJ, England
> > > >> > msabin@cromwellmedia.com          http://www.cromwellmedia.com/
> > > >> >
> > > >>
> > > >> --
> > > >> Mark Nottingham
> > > >> http://www.mnot.net/
> > > >>
> > > >>
> > > >
> > >
> > >
>
> --
> Mark Nottingham
> http://www.mnot.net/
>
>
Received on Tuesday, 10 October 2000 18:05:11 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 06:33:40 EDT