[standards-jig] Avatars

Evan Prodromou evan at prodromou.san-francisco.ca.us
Wed May 7 17:20:10 UTC 2003


>>>>> "RD" == Richard Dobson <richard at dobson-i.net> writes:

    RD> I was thinking exactly this just now while pondering this in
    RD> the bath, having the namespace as the jep page until it
    RD> reaches final in which case it reverts to the authors
    RD> preferred namespace.

Seems like there's a growing consensus around that.

    RD> Also the versioning idea is good but I think authors should
    RD> only bother to update the version number on the namespace if
    RD> there are incompatible or required changes between versions,
    RD> otherwise it could add an unnecessary burden on implementors
    RD> even tho by implementing an experimental protocol they have
    RD> undertaken the requirement to keep their implementation up to
    RD> date.

*I* think that 100% compatible JEP changes are the exception rather
than the rule, and that the benefits of having a consistent policy
about namespaces -- each new version of a JEP is a new namespace --
outweighs the very slight disadvantage of having to update clients
when a 100% compatible change happens.

In fact, it's hard to imagine a JEP change where there's _absolutely_
no change to the syntax or semantics of the protocol. Even spelling
corrections or clarifications of the text change the semantics, albeit
very slightly.

Revving the version number in a client implies that the implementor
has read and understood the protocol. That's a Good Thing.

~ESP

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






More information about the Standards mailing list