Re: Interoperability Event

Scott Lawrence (lawrence@agranat.com)
Fri, 29 Aug 1997 14:30:14 -0400 

>>>>> "LM" == Larry Masinter <masinter@parc.xerox.com> writes:

LM> I think we were really thinking more of an 'interoperability test'
LM> than an 'interoperability demonstration'. The goal is 'figure
LM> out whether people's HTTP/1.1 implementations work with each other,
LM> and if not, what's wrong'.

LM> A secondary goal is 'document interoperable use of each HTTP/1.1
LM> feature with independent implementations', in order for us to
LM> move to 'draft standard' status. That means that a checklist going
LM> down HTTP/1.1 headers and seeing if they're emitted, responded
LM> to correctly, etc., in various combinations would be a good idea.

  That being the case, I assume that we would like this to take place
  before the December IETF (which I was going to suggest as another
  possible venue).  Organizing a physical get-together that soon may
  be challenging unless someone steps up quickly (all you academics
  out there - this is your chance to become to HTTP/1.1 what UNH is to
  FDDI and Scott Bradner at Harvard is to Bridge/Router testing!).

LM> I think it's important that organizations be able to test
LM> implementations anonymously, so that there's no press leakage or story
LM> telling.

  As I noted in my last note, a few of us have publicized the
  locations of 1.1 origin servers that may be used for testing by
  anyone (we log the Server line from requests to our server, but we
  don't talk with anyone about the specifics of what we see).  We've
  all seen postings here resulting from various people doing testing
  with them.  I've asked for publicly usable 1.1 proxies and not
  gotten any response.

  We've also done some testing with the 1.1 clients that we've been
  able to get, and communicated issues to the authors where we found
  them, but as I said in my first note, an event at a fixed place and
  time tends to focus the attention of all concerned and produce more
  definitive results than the kind of ad hoc testing that is going on
  now.

  I think that there is some value in a publically visible event
  because for the good of the net as whole we would like to see 1.1
  displace 1.0 as quickly as possible; the only way that will happen
  is if the web users and especially server operators see it as
  something worth upgrading to get.  Perception is important.  That
  having been said, the public event could easily be separate and much
  later than the kind of test you suggest, Larry, and perhaps that is
  the better course to pursue.  We are eager to participate in either
  or both.

--
Scott Lawrence           EmWeb Embedded Server       <lawrence@agranat.com>
Agranat Systems, Inc.        Engineering            http://www.agranat.com/