Versions Compared

Old Version 4

changes.mady.by.user Mark Diggory

Saved on

compared with

New Version Current

changes.mady.by.user Mark Diggory

Saved on

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.
Comment: Migrated to Confluence 4.0

...

The Chain of Responsiblity pattern is a popular pattern for easing the evaluation of which command in a set would be responsible for execution of a responsibility

Limitations in current Package Ingest framework...

  1. Packagers rely on being designated by NAMES from the command-line and or designated by SWORD.
    1. In the face of not knowing the appropriate packager, we would like the repository to offer an option to allow it to guess the packager to use.
  2. No graceful degradation exists if the package is just a file or simple zip package
    1. We would like to devise a gracefull degredation when specific "sub-type" can still be determined
  3. A series of default Packagers should be able to be constructed that take sane values and use a simple heuristic to determine which packagers can be applied:
    1. If the face of errors in a subtype packager, parent packagers could be set into a debug state, place the item in the submission workflow, allowing the user to review the result prior to to accepting/rejecting.
    2. Treats Submission Workflow as a "Staging tool" prior to pushing content into the repository.

...

  1. FileIngester: Takes any file and stores it in a bitstream, attempts to match BitstreamFormat where possible.
    1. Custom implementations can support format specific tasks (extract metadata of pdf or doc, acquire metadata from mpeg
  2. ZipIngester: Takes any Zip file and expands it into attached bitstreams, named by the directory path found in the archive.
    1. Custom implementations may handle different types of archives (ZIP, RAR, TAR, TAR.GZ)
  3. ManifestBasedIngester: Takes a Manifest and processes it to embellish the Item and Bitstreams from the manifest file. (Either placed inside a ZIP or standalone)
    1. Custom types to support websites and other types saved as zip packages

Challenges with current METS packager design.

  1. The problem is that the current packagers always assume you are processing a manifest file. I simply want to process the contents of the zip and name the item and contents solely on those feathers by default, then build customizaed packagers up based on this. It will make creating new packagers much easier because the basic functionality for getting the bitstreams created will be separated from the evaluation of the manifest.

...