From ImageWeb

Contents 1 Technical Meeting with Library Services 1.1 Goals 1.2 Agenda 1.3 Review changes to databank implementation 1.4 Review databank test cases 1.5 Review use of HTTP 201 created in response to POST 1.6 Review of state information 1.7 Other thoughts

Technical Meeting with Library Services

When: 27-Sep-2010, 11:15-15:00

Where: Zoology, room B-17

Present: AR, GK

Goals

Meeting with AR about databank progress

• Review databank changes implemented
• Get databank test script running

Agenda

• Review databank changes implemented
• Get databank test script running
• Review state data for inclusion in dataset metadata

Review changes to databank implementation

Changes to databank:

• no longer using /packages (which created dataset and uploaded ZIP, unpacked and populated dataset in a single operation). Instead, use /siloname/datasets to create new dataset, then /siloname/dataset/datasetname to add ZIP to dataset, then /siloname/items/datasetname to unpack ZIP to populate dataset - therefore can choose which ZIPs get unpacked.
• sameAs metadata merging problem addressed

Still working on state information at distinct URI.

For reference, notes from NeilJ about additions for V1 databank:

1. Proper RDF handling so existing RDF is merged rather than overwirtten
2. Separation of intinsic and extrinsic object properties...
   • Data returned from invoking the API with different formats should be consistent
   • Separate REST verbs for accessing intrinsic and extrinsic properties
3. Ensure that versioning works - via bulk object copying at this stage

No mention here of DOI allocation, but this will be needed.

No mention here of customized SOLR indexing per-project (within silo) - we believe this is still on the roadmap.

ACTION AR: remind NeilJ about these missing items (DOIs, configurable SOLR indexing)

Review databank test cases

Worked on test cases so far so that they all pass

Test cases TODO:

• update test
  • Submit new version of zip file to dataset (POST to dataset) - expect to see new ZIP file version in dataset, but no change to unpacked data
  • Unpack any version of a zip file to an existing target dataset, create a new version of the target dataset and update associated unpacked data in the target dataset. (Restriction: cannot unpack a different source zip file (not counting version differences) into an existing target dataset)
  • Uppack any version of a zip file to a new dataset: as previously, create new dataset and populate with ZIP file contents. (No need for separate test?)
• munging test

Changes noted:

• use dcterms:created / dcterms:modified consistently; eliminate dcterms:dateSubmitted use.
• serve state and metadata and different URIs, rarher than by content negotiation
• return some items currently in state info as part of metadata

ACTION GK: push modfied test suite to ADMIRAL code repository Done

ACTION AR: complete update and metadata munging test cases

ACTION AR: review use of dcterms created/modfied, and update test cases and databank accordingly

ACTION AR: return state/metadata at different URIs, rather than relying on content negotiation

Review use of HTTP 201 created in response to POST

Use of HTTP 201 response code when creating a resource via POST is correct, but the response SHOULD include a location header. Cf. http://www.ietf.org/rfc/rfc2616.txt section 9.l5.

ACTION AR: include Location information with HTTP 201 responses, for conformance with HTTP

Review of state information

The following fields are considered candidates for inclusion on the dataset metadata as well as in the state information:

• Submission identifier
• Created by
• Current version
• Embargoed?
• Embargo expiry date

ACTION AR: copy and/or move some information from state to metadata

Other thoughts

These were not raised during the meeting, but have been discussed previously:

• databank access performance - retrieving status and metadata information for a dataset, especially the metadata, is currently very slow.