Version 3 (modified by lawrence, 13 years ago) (diff)


Granule Handling

The assumption is that data can be described by the following input documents:

Granules are created by running either the command line granulite tool on one of the input documents, or by uploading one of these documents via the appropriate MILK granule edit page.

The idea is that each granule can form part of multiple data entities.

There are currently two possibilities at upload:

  • the granule input data has an identifier within it, in which case the MOLES granule uses that identifier as the localID, or
  • the granule input data does not include an identifier, in which case the MOLES granule is given a new identifier.

There are currently then two logical possibilities:

  • that granule id is already present, in which case
    • the command line tool should fail with a sensible error, or a command line argument should be used to indicate that the granule is being replaced, or it should override with a new identifier.
    • the web interface should provide a button to indicate which option to use.
  • the granule id is not present, in which case this is clearly a new granule.

The former option is to be expected in the case of datasets being extended, or where metadata is being corrected. In this case, the granule document in MOLES should be copied into the backup repository, and a new version created.

It is thus possible for two identical granules to be created, which only differ by their ID. In this case, the command line and web tools offer a removal option, which can only remove an identifier when passed with the duplicate identifier which is to be kept. All use of the incorrect identifier in other MOLES objects will then be replaced with the correct one.


  • cdml.pdf Download (184.4 KB) - added by lawrence 13 years ago. Climate Data Markup Language