[standards-jig] Re: Versioned Namespaces

Evan Prodromou evan at prodromou.san-francisco.ca.us
Thu May 8 16:14:18 UTC 2003


>>>>> "JK" == Jacek Konieczny <jajcus at bnet.pl> writes:

    JK> The poit is, that experimental specs should not be used any
    JK> more, when the final spec is available. At least unitl the
    JK> final spec is implemented.

    JK> Namespace versioning seems a very nice solution to me. Every
    JK> jabber developer will notice when his software should be
    JK> updated. And if he doesn't do that the experimental
    JK> implementation will not break anything else.

+1

Thank you for such a clear summary.

I think most of the counterarguments against versioned namespaces are
misplaced emotional responses: frustration with some players' practice
of implementing experimental protocols, then expecting them to stay
stable, and complaining if they don't. I can definitely see why this
is undesirable.

But building support for change into the protocol definition process
is, IMHO, a Good Thing. The burden on the JEP author is _extremely_
low, and the advantage of specifying incompatibility is very high.

Jabber software is, fundamentally, communication software. That means
that there are many implementation of the protocol, and many
installation of those implementations. All of them happen to share the
same Jabber network (hooray for that), which means that if any
software uses older protocols, everyone on the network has to deal
with it.

It would be nice if, as a protocol changed, all implementors would
instantaneously implement the new version, and all users of the
implementations would instantaneously upgrade their
software. Overnight, all instances of the older version of the
protocol would disappear from the network.

But I don't think that the real world works like that, and I don't
think it's a good idea to base assumptions on that scenario.

~ESP

-- 
Evan Prodromou
evan at prodromou.san-francisco.ca.us






More information about the Standards mailing list