Date: Tue, 18 May 1993 19:22:26 -0400 (EDT)
From: John C Klensin <KLENSIN@INFOODS.MIT.EDU>
Subject: Re: Comments on URN ID
In-Reply-To: <9305181951.AA13334@merit.edu>
To: clw@MERIT.EDU
Message-Id: <737767346.556103.KLENSIN@INFOODS.UNU.EDU>
Chris writes...
>I don't have a really good feel for how many older systems are out there....
>should the case-insensitive thing be dropped?
Chris,
It really isn't a matter of "older" in the computer sense -- Multics, a
case-sensitive system, was essentially present at the birth of the
ARPANET. The problem is with a much older, and harder to fix, system,
namely "people". The issue cycles back on whatever assumptions you are
going to make about human examination/typing/reading of the URNs or
whatever. If you are going to prohibit any of those activities and have
an effective way to do so--to hide URNs from people now and forever--
then, by all means, expand the character repertoire by making them
case-sensitive.
But I don't think that is going to be practical for any number of
reasons of which citations and debugging are only two. If you can't
keep URNs away from people, then the reality is that people will read
them out loud, write them down on the backs of envelopes, and carry out
a number of other short-term-memory-intensive applications. And there
is lots and lots of human factors experience that says that people who
aren't extremely sensitive to the peculiarities of computer systems will
regularly get case-funny spellings wrong.
Unless a strong, practical, argument can be made as to why
case-sensitivity will leverage improved information retrieval or
document access, let's not go looking for trouble. "Maybe there are few
enough old systems that we can get away with it" is really not the right
sort of reasoning for URNs (even though it might be for URLs, which are,
I think, conceptually a little closer to file system references).
john