[OAI-implementers] Resumption Tokens

Simeon Warner simeon@cs.cornell.edu
Thu, 20 Feb 2003 09:20:40 -0500 (EST)

On Thu, 20 Feb 2003, Charles, Daphne wrote:
> Hi Folks,
> Can any one help with my understanding of how Resumption Tokens work?
> The resumption token may be expected in the response whenever the list of
> records which might be returned exceeds a maximum no. of returns set by the
> repository.  The original request may include other arguments.  However,
> since the resumption token is an exclusive argument, it cannot then be
> issued as part of a request with any other argument, so unless the value of
> the resumption token includes the original request, then the list of records
> returned has lost its context.

The resumptionToken must either encode all necessary context or be a key
for information cached by the repository. Both schemes are used, brief
discussion at
> In the v.2.0 documentation the examples show a ListIdentifiers request with
> from, set and metadataPrefix arguments.  A set of records is issued in
> response, together with a resumption token which includes attribute names
> and values for completeListSize and cursor.  So far so good.
> However, when this resumption token is subsequently reused in a new request
> it contains only a value of xxx45abttyz and has lost the completeListSize
> and cursor attributes as well as the other arguments used in the original
> request.  My question is, how can the repository reconstruct all this from a
> resumption token value of xxx45abttyz in order to provide a set of records
> which logically continues from the previous list?

This token is (likely -- it is opaque to the harvester) a key for cached
information about the result set. In this case the repository can store a
much state as it wants.


> Hoping for enlightenment!
> Daphne Charles
> _______________________________________________
> OAI-implementers mailing list
> OAI-implementers@oaisrv.nsdl.cornell.edu
> http://oaisrv.nsdl.cornell.edu/mailman/listinfo/oai-implementers