W3C home > Mailing lists > Public > uri@w3.org > October 2003

Re: Announcement: The "info" URI scheme

From: Roy T. Fielding <fielding@apache.org>
Date: Thu, 2 Oct 2003 01:29:36 -0700
Cc: "'Hammond, Tony (ELSLON)'" <T.Hammond@elsevier.com>, uri@w3.org
To: Larry Masinter <LMM@acm.org>
Message-Id: <8ECA9E0A-F4B2-11D7-99C4-000393753936@apache.org>

> I don't think this is a good rule of thumb, because is
> hard to evaluate and wouldn't apply retrospectively.

I don't follow your argument.  Just delete the scheme
and then check if the remainder, modulo a minor syntax
change like adding a ":", is sufficient to identify what
the scheme claims to identify with global scope.  If it
does and that holds true for all identifiers within that
scheme, then the scheme itself is meaningless and should
be discarded.

> My criteria for a good "rule of thumb" is that it's
> easy to tell whether it applies, even if the results
> are imperfect.

See, it's a good one.

> I also think your reasons for proposing it ("waste of bytes"
> and "political pain in the ass") are questionable; whether
> something is a "waste of bytes" depends on the application,

Outside of error-correction techniques, redundant bytes are
always a waste.

> and whether something is a "political pain in the ass" or
> a "system of enormous value" depends on whose side you're on.

No, many systems of enormous value are political pains in the ass,
they just take longer to standardize than they would have otherwise.

> Not that I don't think you have a good idea hiding in here
> somewhere, I just don't think you've hit the nail yet.
> Personally, I would have much less trouble with "info" as
> a URN namespace, named "us-niso", since it in effect sets
> up a delegated naming authority with no operational semantics.

I agree that would be consistent with the URN specs.  Note, however,
that had my rule of thumb been applied to "urn", we would have avoided
a significant pain in the ass.

> I've been pretty happy with the rule of thumb that says:
>   Namespaces whose definitions are: "there is an authority,
>   and the authority says what the name means"
>   should be URN namespaces.
>   Other namespaces where there is an operational definition
>   for how to interpret the URI (independent of how the URI
>   might get assigned) can be new URI schemes. The value of
>   a new URI scheme depends on the clarity and likelihood
>   of implementation of the operational interpretation.
> I wouldn't mind instituting a new rule-of-thumb that URI
> scheme proposals should come with examples of implementations
> of the operational definitions.
> What do you think of my proposed 'rules of thumb'?

I think meaning, authority, and operational definitions are
intertwined in the real world, so any effort to separate URIs
into one category or another will simply result in unexpected
complications as soon as they are actually used.  However,
within the context of a world in which the "urn" scheme is
still considered a good thing, I suppose they are correct.
There certainly isn't any reason to have both "urn" and "info".

Received on Thursday, 2 October 2003 05:10:18 UTC

This archive was generated by hypermail 2.4.0 : Sunday, 10 October 2021 22:17:44 UTC