FMI Message Formats

This section describes the message formats for the Function Management Interface (FMI). The message formats are presented in a language-independent notation. Details of the message format notation and key assumptions about the contents of the message formats are as follows:

• "Reserved" indicates that the field is set to zero (for a numeric field) or all nulls (for names) by the sender of the message.
• "Undefined" indicates that the value of the field is indeterminate. The field is not set by the sender and should not be examined by the receiver of the message.
• Fields that occupy two bytes—such as opresid in the Open(PLU) Request—are represented with the most arithmetically significant byte in the lowest byte address, irrespective of the normal orientation used by the processor on which the software executes. That is, the 2-byte value 0x1234 has the byte 0x12 in the lowest byte address. However, the following fields are exceptions:

  The srci and desti fields in buffer headers are stored in the local format of the application that assigns them, since only the assigning application needs to interpret these values.

  The startd and endd fields in elements are always stored in low-byte, high-byte orientation (the normal orientation of an Intel processor).

• Messages are composed of buffers consisting of a buffer header and zero or more buffer elements; see Messages for more information on buffer formats.
• Applications must assign unique index (I) values for every active LPI connection within the node. In particular, the Open(SSCP) Request must be different from the source index it sends in response to the Open(PLU). Additionally, zero should not be used as an I value. An I value of zero means that the sender of the message is inviting the recipient of the message to assign an I value.
• The startd field in each element gives the offset of the first byte of data in the element after the trpad field.

  For non-LUA applications, startd will either be 1 (data starts in the byte after the trpad field), 10 (nine bytes of padding are included between the trpad field and the start of the data), or 13 (12 bytes of padding are included between the trpad field and the start of the data).

  For LUA applications, startd is 4 (three bytes of padding between the trpad field and the start of the data) in the first element of a message and 13 (12 bytes of padding) in subsequent elements.

  The extra bytes are used by the local node for additional header information; this avoids having to copy data into a new buffer when adding this information.

• Because startd indicates the index into dataru starting from 1, not 0, the first byte of valid data will always be at dataru[startd–1].
• If startd is greater than endd, there is no valid data in the message.
• All fields within dataru are of type CHAR, except where the remarks indicate otherwise.
• Note that where a buffer element has a startd of 1, 10, or 13, this only applies to the initial element in the chain of elements, and subsequent elements in the chain have a startd of 1. Messages with two distinct linked element chains in the message formats (for example Open(PLU) Request and Open(PLU) OK Response), have the startd field in the elements at the start of the chains as the value (1, 10, or 13) given in the message format, and the startd fields in all other elements as 1.