Andy Powell, from UKOLN, visited OCLC last week and we had a lot of discussions with him. Some of them were about PURL's, a method of creating 'Persistent URLs' that we both think gets too little notice. Currently OCLC maintains the main PURL server at purl.org, although the software is available for others to run independent PURL servers. Andy thinks we should distribute responsibility for purl.org:
- Several organizations around the world to agree to replicate the PURL software and database
- We use DNS to direct users to the most appropriate PURL server
- OAI-PMH is used to keep the servers in sync, so that PURLs created or modified on one server are rapidly available everywhere
This would make the PURL service more robust, reduce its dependence on OCLC, and give PURLs even more reason to be used. Stu Weibel says that federation was considered from the very beginning of the PURL service. Maybe now is the time to try it.
--Th
Note: the image at the top was found on Sprott's Gateway at wisc.edu. Lorcan suggested 'string of PURLs' which I used even though I don't think a string is quite the right metaphor.
If you are interested in PURLs you might be interested in POIs.
Since I'm interested in digital preservation I've been contemplating the subject of creating permenant Permalinks using PURLs.
Two problems that I still can't find a way around are:
-Since most people copy and paste URI from the browsers' controls when linking, they will end up ignoring the PURL, which will have resolved by then, when linking to blog posts, for instance.
-Assuming that users do copy the Permalink designator which is given as a PURL, pingback mechanisms do cannot perform autodiscovery because of the absence of X-Pingback HTTP header in the response of PURL, and what seems to be an implementation problem on the part of pingback clients that refrain from following on the 302 redirect.
Do you have insights in this subject?
Posted by: Ahmad Gharbeia | January 07, 2007 at 11:01