[OAI-implementers] Re: [Dspace-tech] OAI validation - withdrawn
items, new dc types
hussein at cs.uct.ac.za
Wed Nov 9 11:28:04 EST 2005
well, for starters, i would say the repository explorer is wrong and
shouldnt make any ListMetaFormats requests for deleted ids ... so ive
fixed that! it now searches for the first non-deleted id in the ListIds
response or assumes no id exists if all ids are deleted.
the broader question of orthogonality of the deletion feature is of
course more difficult ... (and something we dredge up every now and then
with no resolution :))
hussein suleman ~ hussein at cs.uct.ac.za ~ http://www.husseinsspace.com
Tansley, Robert wrote:
> Hi Liam,
> Actually it's a weird corner case where it's not clear to me what the
> 'right' thing to do is. It could be that the repository explorer is
> wrong, or DSpace.
> adataPrefix=oai_dc asks for all records which have oai_dc metadata. You
> *have* to ask for a particular metadata prefix with ListIdentifiers.
> But when you actually ask for the oai_dc metadata on a deleted item,
> DSpace reports there is no oai_dc metadata for that item. Because there
> isn't. Because it's been deleted.
> However, if DSpace simply didn't include <record status="deleted"> for
> that item in the ListIdentifiers response, harvesters would never know
> the record had actually been deleted.
> DSpace could simply claim to have oai_dc metadata for deleted items in a
> ListMetadataFormats response, but this doesn't seem right.
> Alternatively, it could give an 'idDoesNotExist' error code, but this
> doesn't feel right either. ListMetadataFormats doesn't provide any
> means to indicate something like status="deleted".
> I've CC'd this to oai-implementers, in case anyone there can give us
> some pointers -- what's the appropriate behaviour?
> Robert TANSLEY / HP Labs / MIT Visiting Researcher
>>From: dspace-tech-admin at lists.sourceforge.net
>>[mailto:dspace-tech-admin at lists.sourceforge.net] On Behalf Of
>>Sent: 03 November 2005 11:26
>>To: Dspace-Tech (E-mail)
>>Subject: [Dspace-tech] OAI validation - withdrawn items, new dc types
>>Hi all -
>>Just testing out our OAI-PMH capabilities on a demo
>>repository using the OAI repository explorer (i.e. using this
>>http://re.cs.uct.ac.za/ ) and I have a couple of validation
>>errors. I've searched back in the lists and haven't found
>>anything obvious about these issues, so wondering if anyone
>>can help ....
>>One problem seems to relate to this feature of DSpace -
>>"DSpace's OAI service does support the exposing of deletion
>>information for withdrawn items, but not for items that are
>>'expunged' ( see above
>>If you look at this xml from a ListIdentifiers request,
>>you'll notice how the first item is deleted (i.e. withdrawn) -
>>i.e. this bit -
>>The OAI explorer test uses this particular item for a
>>ListMetadataFormats request - and it doesn't like what it gets back -
>>(22) Testing : ListMetadataFormats (identifier)
>>------ Start of XML Response ------
>><?xml version="1.0" encoding="UTF-8" ?><OAI-PMH
>>are no metadata formats available for the specified
>>------- End of XML Response -------
>>Test Result : FAIL!
>>**** [ERROR] Error tag found but not expected : noMetadataFormats
>>So what do you reckon - it doesn't understand what
>>status="deleted" means, and that if it did it wouldn't try
>>this test? Is this a problem with the xml DSpace produces or
>>more with the OAI explorer testing utility?
>>The next problem relates to these new DC types we've added -
>>basically, it doesn't like 'em (see below). And no there's
>>reason why it should, I guess - it's got it's own schema, it
>>knows what it likes. So as we've customised our repository to
>>add some dc types, we need to change the OAIDCCrosswalk class
>>to not put these in the oai_dc XML, right? That's fine, but
>>I guess it would be better if only dc fields that would be
>>accepted by the OAI_DC schema are actually put in generally -
>>i.e. if it only puts in elements that are in [title, creator,
>>subject, description, publisher, contributor, date, type,
>>format, identifier, source, relation, coverage, rights]. If
>>so no extra effort would be needed when new dc types are
>>added. Would it make sense to change OAIDCCrosswalk to do this?
>>Any thoughts much appreciated.
>>validation message -
>>(41) Testing : GetRecord (identifier, oai_dc)
>>------ Response from Xerces Schema Validation ------
>>[Error] re.0NiDLV:1:3495: cvc-complex-type.2.4.a: Invalid
>>content was found starting with element 'dc:entrez'. One of
>>"http://purl.org/dc/elements/1.1/":rights}' is expected.
>>/tmp/re.0NiDLV: 777;11;0 ms (35 elems, 10 attrs, 0 spaces, 2388 chars)
>>------- End of Xerces Schema Validation Report -------
>>This email has been scanned by Postini.
>>For more information please visit http://www.postini.com
>>SF.Net email is sponsored by:
>>Tame your development challenges with Apache's Geronimo App
>>it for free - -and be entered to win a 42" plasma tv or your very own
>>Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php
>>DSpace-tech mailing list
>>DSpace-tech at lists.sourceforge.net
> SF.Net email is sponsored by:
> Tame your development challenges with Apache's Geronimo App Server. Download
> it for free - -and be entered to win a 42" plasma tv or your very own
> Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php
> DSpace-tech mailing list
> DSpace-tech at lists.sourceforge.net
More information about the OAI-implementers