Harvesting experiences

Ray Plante rplante at poplar.ncsa.uiuc.edu
Thu May 1 04:22:06 PDT 2008


Hi Guys,

> Linde, A.E. wrote:
>> I would have thought that deleted records should only be returned if they 
>> were recently deleted and the request is for an 'records updated since ...' 
>> but a request for ALL records should not bother with deleted records. 
>> That'd work,wouldn't it?

A quick note on this.  The Registry Interfaces standard makes no 
distinction between deleted and non-deleted records within the harvesting 
interface.  This is indeed what you want.  A harvester, when it encounters 
a deleted record, can choose to handle it in any way it desires.  For 
example, it can truely deleted all trace of it in its own database.  Since 
most harvests are incremental, the harvester should not be further 
bothered by a deleted record (unless it is later undeleted).  The cost is 
then only on the originating registry.  Such a registry may consider an 
alternative storage mechanism for deleted records to reduce internal 
overhead (e.g. of having deleted records cluttering up the searchable 
database).

Note, however, that we have said that the search interface does not return 
deleted records by default, although I'm not actually seeing any mention 
of this in the spec.  Am I remembering this correctly?  (Kevin?)

hope this helps,
Ray



More information about the registry mailing list