[tm-pubsubj-comment] RE: [topicmaps-comment] referring to a topic fromoutside a TM -- PURL

[Mary Nishikawa <nisikawa@fuchinobe.oilfield.slb.com>]

At 11:36 AM 3/14/2002 +0100, Bandholtz, Thomas wrote:

> > -----Original Message-----
> > From: Mary Nishikawa 
> [<mailto:nisikawa@fuchinobe.oilfield.slb.com>mailto:nisikawa@fuchinobe.oilfield.slb.com] 
>
> > Sent: Thursday, March 07, 2002 1:09 AM
>[...]
> > Do you think that it would be possible to use a PURL URL
> > (Partial redirect)
> > or in the future, a URN with a query string?
>...
> > I registered my domain in Purl as 
> <http://purl.org/mnishikawa>http://purl.org/mnishikawa.
> > This will
> > redirect you to my published subjects that are in the
> > tm-pubsubj area of
> > OASIS.
>
>There are some problems:
>I tried:
><http://purl.org/mnishikawa/psdoc71.htm>http://purl.org/mnishikawa/psdoc71.htm 
>
>and was redirected to
><http://www.oasis-open.org/committees/tm-pubsubj/docs/mni/psdoc71.htm>http://www.oasis-open.org/committees/tm-pubsubj/docs/mni/psdoc71.htm 
>
>(BTW, very interesting content in this document - but this is another 
>story ----- )

Yes, Thomas, this is how the PURL works. It is a partial redirect. It is 
intentionally set up by the user this way. If I wish to change the location 
of my published subjects I can reassign the redirect of my PURL to the new 
URL. This is one way to recommend persistence of a URL for published 
subjects. I am not convinced of the stability of URLs. I do not know that 
this is the best way, but we need to address this issue.


>But it works somehow strange using a fragment identifier:
>In your doc, there is a link to
><http://www.oasis-open.org/committees/tm-pubsubj/docs/mni/psi-unspsc-71.htm#010295>http://www.oasis-open.org/committees/tm-pubsubj/docs/mni/psi-unspsc-71.htm#010295 
>
>When you try
><http://purl.org/mnishikawa/psi-unspsc-71.htm#010295>http://purl.org/mnishikawa/psi-unspsc-71.htm#010295 
>
>instead. the positioning works, but the browser shows only
><http://www.oasis-open.org/committees/tm-pubsubj/docs/mni/psi-unspsc-71.htm>http://www.oasis-open.org/committees/tm-pubsubj/docs/mni/psi-unspsc-71.htm 
>
>in the URL field (fragment identifier removed). If you bookmark in this 
>state, the position will be lost.

What you point out is exactly what is described by the OCLC Purl service. 
See http://purl.oclc.org/OCLC/PURL/FAQ#toc5.7 We get the same problem when 
using a fragment identifier appended to an xml file, don't we? I am only 
taken to  the anchors in html files, when redirects are not used.
  I have encountered the same problem with RDF or XTM. The question is, do 
we really need to be taken  directly to the anchor in the file? Isn't it 
good enough to be taken to the file containing the published subject? In 
this case, it may be a good idea to have the metadata near the top of the 
file, to make it a little easier for the person looking for this information.


>And, BTW, the bookmark will not know that 
><http://www.oasis-open.org/committees/tm-pubsubj/docs/mni>http://www.oasis-open.org/committees/tm-pubsubj/docs/mni 
>is not ment to be persistent itself
>
>So this looks quite unsufficient for a PSI, imo.

No, PURLs cannot be bookmarked, but do we need a bookmarking capability for 
published subjects?

The most important requirement for published subjects I think is this, to 
have a URL that is persistent so that someone who is using my PSIs two 
years from now will be able to merge them with their own. It would be very 
nice if we could resolve the URL to see who the publisher is, and find out 
other pertinent information. If a company or organization can guarantee the 
persistence of their URL, then they would not need to register with PURL. 
Others who cannot, may want to use PURL.

Cheers,
Mary

**************************************
Mary Nishikawa,  EDMS Technical Advisor
XML Evangelist

Technique
Schlumberger K. K.
2-2-1 Fuchinobe
Sagamihara, Kanagawa 229-0006

Tel:  +81-42-759-5376
Fax: +81-42-759-3563
**************************************