Registers and records

The registers indicate a stream of records.

A record in a register is distinguished from a table of records as each record in a register or record stream may have a different number of named attributes and metadata in it.

A record is passed through a pipeline and acted upon by commands.

All processing uses the underlying record object model, so readers read from a data source and convert data into a record, while writers take records and convert it into the desired output form.

The following diagram shows an illustration of registers and records:

The attribute can be empty or contain multiple values.

Metadata is distinguished from attributes here as being something descriptive about the record rather than content of the record. For example, when a CSV writer writes out a record, only attributes and not metadata are written out.

Note: The model is still flexible. For example, an object with 10 attributes can be stored as 10 records with the same metadata ID, one attribute per record. These records can be merged into a single object or separated again when needed.

• Any attribute (not just associations) may have 0-n values, which is required to process general engineering data.

• Managed processing does not consider associations as EIWM does, only attributes. An association may be represented by an attribute and later mapped to an association. In EIWM terms, is referenced by ISOMETRIC-00001 of class P&ID is considered to have lost information, whereas attribute P&ID with value ISOMETRIC-00001 may be interpreted as explicitly specifying a P&ID of ISOMETRIC-00001 (which can be checked and asserted as incorrect).