Re: Uniform Resource Names - almost draft

Patrik Faltstrom <paf@nada.kth.se> Thu, 13 October 1994 20:18 UTC

Received: from ietf.nri.reston.va.us by IETF.CNRI.Reston.VA.US id aa07768; 13 Oct 94 16:18 EDT
Received: from [132.151.1.1] by IETF.CNRI.Reston.VA.US id aa07764; 13 Oct 94 16:18 EDT
Received: from [192.197.208.1] by CNRI.Reston.VA.US id aa16717; 13 Oct 94 16:18 EDT
Received: by mocha.bunyip.com (5.65a/IDA-1.4.2b/CC-Guru-2b) id AA01564 on Thu, 13 Oct 94 14:39:41 -0400
Received: from kth.se by mocha.bunyip.com with SMTP (5.65a/IDA-1.4.2b/CC-Guru-2b) id AA01556 (mail destined for /usr/lib/sendmail -odq -oi -furi-request uri-out) on Thu, 13 Oct 94 14:39:01 -0400
Received: from [130.237.225.201] (paf-mac.nada.kth.se [130.237.225.201]) by kth.se (8.6.7/8.6.9) with SMTP id TAA20669; Thu, 13 Oct 1994 19:38:23 +0100
X-Sender: paf@mail.nada.kth.se
Message-Id: <aac32be60a0210041f94@[130.237.225.201]>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Date: Thu, 13 Oct 1994 19:38:33 +0100
To: Martin Hamilton <martin@mrrl.lut.ac.uk>
Sender: ietf-archive-request@IETF.CNRI.Reston.VA.US
From: Patrik Faltstrom <paf@nada.kth.se>
Subject: Re: Uniform Resource Names - almost draft
Cc: Mitra <mitra@path.net>, uri@bunyip.com, clw@bunyip.com, schoultz@sunet.se

At 12.24 94-10-13, Martin Hamilton wrote:

>I was going by draft-ietf-wnils-whois-arch-01.txt, which
>says (p.17):

>     (Note: The name HANDLE can be replaced with the shortname '!')
>
>Is there a more recent version of the spec ?

Ooops...that is correct. I must wait for the editor, Richard Schoultz,
to comment this. I think the wnils working group said at last that
all shortnames should be removed. The '!' might have been saved, but
I have to wait for a comment from Richard or Chris Weider on this.

>So, my minimal URN lookup scenario (via 'handle') right now
>would be something like this...
>
>
>success:
>
>  % 220-whois++ server cuppat, LUT mini_whois++ 1.0
>  % 220-only understands handle and urn lookups!!
>  % 220 your mileage may vary
>  handle="urn:iana:rfc:1490"
>  % 200 command ok
>  # FULL 1
>  # URN cuppat:rfc1490
>   URN:IANA:rfc:1490
>   URL:ftp://ds.internic.net/rfc/rfc1490.txt
>   URL:ftp://nis.nsf.net/internet/documents/rfc/rfc1490.txt
>   URL:ftp://nisc.jvnc.net/RFC1490.TXT.1
>   URL:ftp://ftp.isi.edu/in-notes/rfc1490.txt
>   URL:ftp://wuarchive.wustl.edu/info/rfc/rfc1490.txt.Z
>   URL:ftp://src.doc.ic.ac.uk/rfc/rfc1490.txt.Z
>   URL:ftp://ftp.concert.net/rfc/rfc1490.txt
>   URL:ftp://ftp.sesqui.net/pub/rfc/rfc1490.txt
>  # END
>  % 226 transfer complete
>  % 203 toodle pip

I can see some problems with this:

1) You are not able to include the actual handle name (the URN) in
   a centroid, so you will not get all information you want, i.e.
   no servers-to-ask response.

2) The implementor of a Whois++ server is forced to use the URN as
   a handle, which is not the case every time. Me myself is using the
   handle as a direct index into the database, so the handle can change
   when the database is rebuilt.

The best thing is, I think, to store a "template URN" which have an
attribute named "URN" and one or more URL-attributes.

By doing this one does not have to explicitly handle the
template URN in the server or queries for URNs.

A query should look like:

  URN="IANA:rfc:1490"

The drawback (you have to search for a specific attribute and then
later get the handle) might not be more expensive. In my server there
is no difference in CPU cycles when searching for a certain attribute
and a handle, part from that the Handle might be a unique index which
in fact can make the search a little bit faster.

So, I vote for skipping the special handling of searches for URN's.

    paf