Reorder the data part. Per data type all information:
- Supported formats
- Naming conventions
  - Datasets (what the user has to supply)
  - Datasources (the files, databases, tables, columns, ...), and file
    structure conventions

Raster data
- Supported formats
  - GDal
- Naming conventions
  - Datasets
    - Deprecated PCRaster / old Aguila conventions
    - Dal conventions
  - Datasources
    - Scenarios, timesteps, uncertainty

Feature data
- Supported formats
  - OGR
- Naming conventions
  - Datasets
    - Datasource/layer/attribute
    - How about credentials for attribute tables?
  - Datasources
    - Scenarios, timesteps, uncertainty
    - Simple attributes in feature table
    - Other attributes in separate tables
      - Column titles based on data space addresses

Constant data





Table layout
Apart from the attribute field there must be at least one of the folowing fields present:
- temporal data
  - field name: 'date'
  - field type: integer
- uncertain data
  - field name: 'quantile'
  - field type: floating point
- scenario data
  - field name: 'scenario'
  - field type: string

If the table is to be used as an external attribute table of a feature layer, than an additional field, called 'fid' of type integer, must be present.

Whatever the combination of dimensions that are present in a table, the combination of 'dimensional fields' (date, quantile, scenario) MUST be configured as the primary key of the table. This way, we are certain that any combination of a date and/or quantile and/or scenario maps to a single attribute value. Aguila checks whether the key is configured and will not open the table if it is not.

In case the feature id is part of the table, than it must also be part of the primary key.

The order of the fields does not matter.

The order of the records does not matter.

When creating tables, you will be writing <fid, scenario, date, quantile, value> to the table (in case all those fields are present).









