FLCOMMS data model

• Designed for real-time applications
• Collection of named data blocks
• Data blocks defined by name, size, type, and ``magic number''
• All processes accessing a block must agree on data block structure
• Data blocks typically contain several fields; contents are (mostly) transparent to FLCOMMS library
• Data blocks created when first process attaches
• Data blocks (normally) removed when last process detaches
• Single-writer, multiple-reader discipline enforced
• ``Best effort delivery''