Re: Resources and Identifiers

Daniel LaLiberte (liberte@ncsa.uiuc.edu)
Fri, 21 Feb 1997 16:16:15 -0600 (CST)


From: Daniel LaLiberte <liberte@ncsa.uiuc.edu>
Date: Fri, 21 Feb 1997 16:16:15 -0600 (CST)
Message-Id: <199702212216.QAA18376@void.ncsa.uiuc.edu>
To: Jonathan Rosenne <Jonathan_Rosenne@CompuServe.com>
Cc: URI List <uri@bunyip.com>
Subject: Re: Resources and Identifiers
In-Reply-To: <199702211616_MC2-1194-2C56@compuserve.com>

Daniel LaLiberte wrote:
 > >Same bits means same object, 

Jonathan Rosenne writes:
 > Same bits are not necessarily the same object.

Three people have pointed this out, so I better explain myself.
(I originally had a parenthetical qualification but too many
parentheses tends to get confusing).

If what you mean by "object" is only the remote object referred to by
the URL that resulted in the bits, then yes, getting the same bits
from two different results does not necessarily mean you got them from
the same remote object.

But if what you mean by "object" is whatever you have in hand, then if
you have two piles of bits that are identical down to the last bit,
then they are (copies of) the same object.  These are constant local
objects as opposed to remote resources.  In the programming analogy,
one number 1364 is identical to anther number 1364.  The HTTP spec
calls this an "entity".

If what you have in hand is two identifiers, you can compare the
identifiers to see if they are identical identifiers, and if so, then
the objects they refer to are obviously identical.  If they are not
identical identifiers, you may be able to do some operation on the
identifiers to find out if they are referring to the same object.

Identifiers themselves can be considered objects in this sense of some
bits that you operate on indirectly, just as an object can contain
within itself a reference to another object.  It's all very
consistent, though a bit unusual.

--
Daniel LaLiberte (liberte@ncsa.uiuc.edu)
National Center for Supercomputing Applications
http://union.ncsa.uiuc.edu/~liberte/