Design meeting, September 7, 2006.
Attendees: TE, KFC
First point of order was followup on last design meetings.
DOMS Content Model has been updated with the following:
- Relation for the four types FILE/IMAGE/VIDEO/AUDIO (has...File).
Seconds -> Milliseconds
- All content models are subtypes of the SB content model.
We then went on to discuss the DOMS architechture, and work pacakges this seemed to indicate. TE has an image of our whiteboard drawing. Please attach it here.
We ended up converging on a very OAIS-like model.
Components include:
- Fedora as metadata-storage
- Some sort of data-storage, first shot probably an apache we can access with URLs
- An ingest module, including transactions (DTV?) and validation (RIFF).
Access through API-A and REST-RELS to FedoraBrowser
- Access through OAI to Summa
A FedoraAdmin interface to edit the repository directly (the BIG stick)
- Some sort of batch integrity/validity-checker to be run at times.
- Project specific workflow modules
- Backup
- Auditing
Validation is expected done by getting content model definition from Fedora and checking against that. We need some bootstrapping before that is possible.
We need persistent IDs.
Next point was defining work packages and other stuff for iteration 1, and sketching iteration 2.
First iteration will be a "skeleton iteration", making the basic ingest module, working on a fedora repository and an apache server with files.
Work packages are:
- Fedora: Responsible KFC. Installation instructions for Fedora for DOMS.
- Ingest: Responsible TE. Basic Ingest module, doing no itegrity validation, but accepting FoxML files in a directory, ingesting them, and moving failed files to another directory.
- Apache: Responsible KFC. Installation instructions for Apache for DOMS.
- Release test: Responsible KFC. Release test of the first iteration.
Besides that basic development infrastructure needs to be built. This include Idea setup, CVS, Bugzilla, Wiki. Clover is under consideration. KFC responsible.
Notable decisions:
- Log4J used for logging
- JUnit used as much as possible
- A pseudo project module, containg Idea project files and project libraries will be made.
- Libs will be in CVS. Project depepndencies will be defined in Idea
- Ant is not used before a need is recognised
- Main language is English. Resource bundles will be used for end-user interfaces, for support of Danish.
- The Ingest module is a candidate for open sourcing.