From: mitra@path.net (Mitra)
Subject: On the internet draft
Date: Tue, 18 May 1993 16:57:03 GMT
Message-Id: <C78F34.105@pandora.sf.ca.us>
>INTERNET--DRAFT Chris Weider
>IETF URI Working Group Merit Network, Inc.
> Peter Deutsch
Nice paper Chris and Peter.
I like that it meets the (unstated) goal of being syntacticly equivalent
to a URL, so that software that is built to handle URL's can be extended
to handle URN's by just treating it as its own name-space with its own
rules.
Now .... the picky bits.
I'm not sure why the nameing authority scheme has been done this way, it
requires anyone assigning names to register with IANA, which defeats
your goal of having anyone be able to be a publisher? Wouldnt it be
usefull to be able to self-register by using your domain name, and then
be able to put a URN->URL mapping service on a well known port. The
alternative us that I can see many documents being given URL's instead
of URN's to avoid the hassle of registering.
This is the first time I've seen it said that URN's are NOT to be human
readable, if they are to be cited in (printed) documents they need to be
readable (e.g. an ISBN is readable if not intelligable).
Your xapha string is far too restrictive, if URN's are not to be human
readable they could be any value (i.e. a pure binary string). If they
are to be readable then they should at least include the rest of the
printable characters - for example why was $ left out?
Please have someone from a non-US character set perspective check this
over, I dont know the solution but it's bound to come up.
Why the ambiguity over case, either define it so a-z is equivalent to
A-Z, or is different. It can be coded either way, but cant be coded both
ways! (I'd recommend they are different, antique systems with UPPER CASE ONLY
can escape lower case characters!)
- Mitra