Message-Id: <199402172119.QAA13956@wilma.cs.utk.edu>
From: Keith Moore <moore@cs.utk.edu>
To: ccoprmm@oit.gatech.edu (Michael Mealling)
Subject: Re: caching
In-Reply-To: Your message of "Thu, 17 Feb 1994 09:41:17 EST."
<199402171441.AA06717@oit.gatech.edu>
Date: Thu, 17 Feb 1994 16:19:39 -0500
> > A stable object, or a way to get something
> > which may differ from request to request?
>
> Whether it's stable or not will depend on each occurrence and therefore
> can't be made a specification of the whole entity. The name itself
> is stable because it's immutable. The object itself can be mutabble
> subject to the whims of the author/publisher.
A name is useless without binding it to some object. The fact that the
name is stable isn't particularly useful unless the binding between
the name and the object it refers to is also stable.
If I had a (name, object) pair, I'd personally like to be able to find out if
the name referred to that object. One way to do this would be for me to ask the
publisher to give me an MD5 signature for the object. But this would only work
if there were only one valid representation of that object.
But without such a mechanism, how can we guard against the error caused by
conversions not done by the publisher?
Maybe it's okay if the publisher says "here is the official GIF version of this
object, and here is the official JPEG version of this object". But if an object
can exist in multiple formats, how do I tell a publisher-authorized GIF version
from one that someone else has converted from the JPEG version?
This concern may sound farfetched, but I've actually seen it happen -- someone
takes a GIF file from one archive, and converts it to JPEG because the JPEG
version takes up less space. Same thing for 8-bit mu-law audio to 4-bit
compressed audio. In either case the result can be a disaster.
Assuming we can deal with this somehow, and if we allow the publisher to have
multiple representations of an object with the same URN, haven't we now created
a need for names for the individual representations also? Isn't it useful for
me to be able to ask for a specific representation of an object, just like I can
specifically ask for the hard cover version of a book?
As I've said many times, I can see the need for a kind of name that refers to
any of several representations of an object. But the ability to name the
specific representations is *much* more important if you're trying to build a
system that manages replication of these objects over a network.
Keith