[dispatch] Web Service Discovery

[Phillip Hallam-Baker <[e-mail]>]

Following our last meeting, I have put together a draft describing the
approach I am using to perform Web Service Discovery which I would like to
discuss in Montreal

https://datatracker.ietf.org/doc/draft-hallambaker-web-service-discovery/

My approach is largely constrained by existing RFCs and 'common sense' but
these are not enough to remove all variability in approaches.

The goal here is to enable discovery of Web Services by means of a compact
account identifier in the form users are used to using. That is
[e-mail].


I am aware of previous work in this space. In particular the original
OpenID specification that tried to tell people http://example.com/alice was
an account identifier, XRI, etc. and a large number of 'open' protocols
with proprietary infrastructures. So there are now multiple services using
the same 'open protocol' but the users can't talk to each other because the
user names don't provide for federation.

In my view, none of those are as good as the approach established in rfc822
and descendants which users are already familiar with.

It will be noted that the approach does allow for negotiating across
presentation protocols. For example, http vs QUIC or COAP. I have not gone
down that rabbit hole at this time.


I would like to move forward with this document either as an independent
draft or as a Working Group. Which is appropriate really depends on whether
there is an interest in defining additional discovery description keys to
the three specified in the documentation. The keys I have defined are
narrowly focused and the scope for variation is limited to deciding whether
we should use abbreviations for the key names or not.