Date: Sat, 22 May 93 22:23:08 +0200
From: aronsson@lysator.liu.se
Message-Id: <9305222023.AA23308@robert.lysator.liu.se>
To: hoymand@joe.uwex.edu
Subject: Re: Grainularity of URN
I see URNs as "electronic ISBN numbers". The existence of a URN would not
necessary mean that a corresponding URL would exist. I can publish a book
in 100 copies and sell to my friends, and a librarian would fail to
translate that book's ISBN to a physical address (library/department/shelf).
A URN-to-URL translator must be able to fail; to say "cannot resolve URN".
While an ISBN number refers to an entire book, I would assign separate URNs
to different parts of an electronic text. Say that I have registered the
URN prefix "runeberg" with IANA, and that I have "published" an electronic
copy of the Swedish translation of the Bible. If someone wants to get a
copy of the entire document, I will call this URN:runeberg::swbible:::,
and I will suggest FTP as a best means of transfer. If they only want
the gospel of S:t Matthew, that is URN:runeberg::swbible.matt:::,
the 3rd chapter is URN:runeberg::swbible.matt.3:::, the 7th verse is
URN:runeberg::swbible.matt.3.7:::. While a chapter might be a reasonable
chunk of information for Gopher or WWW, one verse is reasonable as a
"fortune cookie". For a book where chapters are called
"the stranger's sight", only the chapter number would be included in the URN.
This is to say that I don't want metainformation in URNs, at least not
too much, Keep URNs short, and in limited ASCII, as the draft says.
Today, you can translate URN:runeberg::swbible.matt.3::: to
"gopher://gopher.lysator.liu.se:70:11/project-runeberg/txt/bibeln/4003".
It does exist, but it has no URN assigned to it yet.
Lars Aronsson
Aronsson@Lysator.LiU.SE