RE: Proposed text on reliability in the web services architecture

As much as I admire the intelligence and experience that is poured into this
thread, it has gone on for some weeks now and has rather lost touch with the
original subject -- actual text for the WSA document on the various meanings
of "reliability" and how to address it.

This leads me to suggest some points of netiquitte for this list:

- It is where the WS Architecture WG does its technical work.  It's not for
general discussions of the philosophical issues surrounding Web services.
Consider moving threads to [email protected] or [email protected] when they
diverge from the work of the WSA WG.  (I realize that www-ws looks like a
DAML-S mailing list, but that is not its intended purpose!).

- Remove everyone whom you know is on the mailing list from the To: and CC:
fields.  I suspect that we all get far more mail than we want already, and
duplicates of the same message are not appreciated.

- Please, PLEASE use the subject line appropriately.  The current subject of
this thread is something like "Implementing a reliable delivery system" or
"Why bother implementing a reliable delivery layer".  Most people "tune out"
of mailing list threads after a couple of days, so as a rule of thumb a
thread that has gone on for more than a few days has lost its readership.
Anyone interested in having their thoughts read should change the subject
line.

- Know when to "agree to disagree."  Remember that pertinent comments that
are made to www-wsa-comments pretty much MUST be tracked and formally
responded to before any document can advance in the W3C process.  You'll get
a fairer hearing on a point by recording a formal Issue than by beating it
to death on this list.

- Again, the more you orient messages "I think the WSA should say XXX about
YYY instead of | in addition to ZZZ", the more likely you are to actually
have an impact.  We just had a face to face meeting and will be updating the
documents very soon.  This is a VERY good time to suggest text on a) the
general subject of the definition and importance of "reliability"; b) the
various reliable messaging/delivery specs and what their common principles
might be; c) alternative ways to achieve reliable *applications* without a
reliable *infrastructure* (e.g., specific text and references about the
importance of idempotence, etc. ... but don't just say "see the Waldo
article" ... we've seen it!).

Thanks,

Mike Champion
[wearing my WSA WG co-chair hat and speaking at the request of the WG.  Dave
Hollander will be posting his own thoughts, I believe]