Automatic Control

Component use is becoming more and more prevalent every day. Indeed advantages such as greater productivity represent interesting qualities for the creation of industrial applications. Important efforts are made to help engineers through the improvement of the design and the description of components and through the specification of contracts. However most of the approaches that associate components and contracts propose only run-time checking. In software architecture design, it would be useful to consider contracts when we check the validity of the architecture. Our work takes place in the context of the RM-ODP(Reference Model for Open Distributed Processing) and more precisely the DASIBAO methodology. This paper presents a componentbased model associated with several contracts and it describes some verifications that can be performed on them.

 Most measurement tools based on events  Some predefined change to system state  Definition depends on metric being measured  Memory reference  Disk access  Change in a register's state  Network message  Processor interrupt Copyright 2004 David J. Lilja 3 Event Classification  Count metrics  The number of times event X occurs  Number of cache misses  Number of I/O operations Copyright 2004 David J. Lilja 4 Event Classification  Secondary-event metrics  Record a value when triggered by some event  Record block size for each I/O operation  Count number of operations  Find average I/O transfer size  Record necessary information only when selected event occurs  Modify system to record event  Dump data when program terminates  May need intermediate dumps also  E.g. simple counter in page fault routine Copyright 2004 David J. Lilja 7 Event-Driven Strategies  System overhead  Only when the event of interest actually occurs  Infrequent events → little perturbation  Frequent events → high perturbation  No longer "typical" behavior?  Perturbation changes system being measured

The molecular basis of triosephosphate isomerase (TPI) deficiency was studied in 3 patients from three separate families. In all 3 patients, genomic DNA directly sequenced after amplification by the polymerase chain reaction exhibited the point mutation TPI315C amino acid 104 GluAsp. Although other mutations known to cause TPI deficiency have been restricted to single families, the amino acid 104 defect has now been described in nine apparently unrelated families throughout the world and is clearly the most frequently occurring form of the disorder. The basis of the repetitive occurrence of this mutation remains unexplained.