2008-06-20 DOMS Status Meeting
This status meeting was dedicated to get an overview over what we need to deliver to Mjølner before the DOMS team goes on vacation. Therfore the end of next week is the terminal deadline for giving Mjølner all the information they need to start their development.
The first delivery to Mjølner will take place tomorrow, tuesday 2008-07-01, where we will have a meeting and present the DOMS data model, which is very close to be finished. It was the plan also to provide a testbed, FoxML samples conforming to the DOMS data model and APIs by the end of next week, before the summer holidays kick in. However, BAM/KFC realised that the development of the testbed will take 10 man days, so Mjølner will have to choose what they think most important to get, as it will not be possible to provide everything before the summer holidays begin.
KFC had realised that it would only be realistic to implement "simple search" in the testbed, as using Summa would be impossible due to the lack of a Fedora 3.0 compliant OAI-provider for Summa indexing. Further he found Gsearch too unstable to use.
It was decided that JRG should work on some DOMS data model compliant FoxML samples, when the rest of the DOMS team are out for then summer holidays. The reason for this is that JRG will be able to send the samples to Mjølner, one by one, as they are finished.
A review of the data model was conducted on the meeting in order to get an overview over what is left to be described and identify any corrections which should be made before the meeting with Mjølner tomorrow.
During the meeting we came up with the following comments and corrections:
- Licence should be spelled with "s" and not "c", hence "License".
- Indirect circular references are allowed in the data model, however, it is not allowed that an object has references to itself.
- The Fedora DC data stream and the DOMS DC data stream seems to be redundant in many cases, hence, it was decided to make the DOMS DC data stream optional. Also, it would probably not always be convenient being forced to use qualified Dublin Core, as specified by the DOMS DC stream.
- The data model lacks a schema for the Rels-EXT data streams.
- By design, it is not possible to specify an XACML document per relation in a Policy data stream, however, we will do it anyway with the help of some front-end magic provided by Mjølner.
- If we really want to change the names of the STATE data stream, then it should be done before the meeting with Mjølner tomorrow, otherwise the state names will forever be DRAFT / INTERMEDIATE / PUBLISHED.
- Content Model Image: Image objects must have a relation to a Tiff file.
Content Model Text: Text objects must have a relation to an UTF-8 file and can have a relation to a PDF file. Any other file relations are optional. The DocFile data stream must be removed from the content model. Future collections are free to extend the Text content model to contain more data stream, e.g. OfficeOpenXML like removed the DocFile stream.
- Content Model License: The LICENCETEXT data stream must be removed, as the human-readable-license description shold be included either in the Fedora DC or DOMS DC stream.
- Content Model File: Specify that the contents of the ORIGIN data stream should be the output from PREMIS. Any alternative contents should have its own schema.
Base Content Model: The OAI:ItemID data stream must be removed - not everything is OAI harvested. Describe that the SCHEMABINDINGS data stream must contain bindings between schemas and data streams.
- The VIEW data stream lacks a schema.