Re: the case of two bats

At 09:47 AM 5/18/00 -0500, Dan Connolly wrote:
>Er.. you'd fire anybody who used relative URIs to link between
>two documents in the same directory? Is that really what
>you mean to say? I'm not sure what you mean by "actual contents".

I clarified that in a following message:
SSL>Strike contents, replace with meanings.  In Dan's example, 
SSL>they overlapped significantly, but they don't necessarily.

>Anyway... if somebody uses relative URI references when they should
>have used an absolute URI reference, then yes, they should be fired.

When it changes the identy of the elements, I'd say they should have used
an absolute URI reference, and they should be canned.

>But does that motivate a special exception to make namespace
>identifier syntax different from identifiers for all
>other Web resources?

Yep.  Relative URIs are dangerous enough (i.e. cause errors that raise
warnings) in content-management practice, but their use in
meaning-management practice seems even more dangerous - and less likely to
raise warning flags.

>i.e. do we find all the 'cause for termination' misuses of Web
>technology and make exceptions in the specs to rule them out?
>What happened to the principle of minimal constraint?

What happened to the principle of interoperability?

>> What exactly is this 'Web Architecture' that keeps getting thrown around?
>
>In this case, as I said, it's my own personal understanding of the Web.
>It's based on my experience in the development of the URI, HTML, HTTP
>specs and libwww, and all the various implementations I've looked at.

Sorry, Dan - you should qualify that as 'my understanding' and not as 'Web
Architecture' with caps.  It'd generate a lot less friction and cause a lot
less confusion.

Simon St.Laurent
XML Elements of Style / XML: A Primer, 2nd Ed.
Building XML Applications
Inside XML DTDs: Scientific and Technical
Cookies / Sharing Bandwidth
http://www.simonstl.com

Received on Thursday, 18 May 2000 10:54:45 UTC