Re: Resources and Identifiers

Jonathan Rosenne (Jonathan_Rosenne@CompuServe.com)
Fri, 21 Feb 1997 16:16:23 -0500


Date: Fri, 21 Feb 1997 16:16:23 -0500
From: Jonathan Rosenne <Jonathan_Rosenne@CompuServe.com>
Subject: Re: Resources and Identifiers
To: URI List <uri@bunyip.com>
Message-Id: <199702211616_MC2-1194-2C56@compuserve.com>

Daniel LaLiberte writes:
>> But I don't think there's any universal concept of 'same
>> objects' other than 'objects referred to in exactly the
>> same way'.
>
>Same bits means same object, but for a complex object, determining
>that it has the same bits as another can be expensive (e.g. compare
>two collections).  

Same bits are not necessarily the same object. For example, the W3C carries
URL's for specific versions of documents and for the most current version.
At a certain moment in time, the latest release has two URL's that get
resolved to the same bits, but they are different because their meaning is
different. This is true even if one gets resolved into the other.

>One relatively easy work around for dealing with multiple identifiers
>is that the provider of a resource maintains a "preferred" identifier
>that it tells clients about.  You could then compare what two
>different identifiers refer to by comparing their preferred
>identifiers.

In the example above, both URLs are "preferred", but they have a different
meaning.

Jonathan