Two models widely discussed are Abstract State Machines (ASM) and Event-B. There also some efforts on integrating domain knowledge (represented as ontologies) with state machines, which I find really nice. When designing and using LODM, I tried to use the conceptual model to ground the state machines and even to associate concepts to events (and, therefore, states transition).

Something we often struggle when defining the model is the specification. Currently we use UML, but there are alternatives. In the paper “Structuring the State and Behavior of ASMs: Introducing a Trait-Based Construct for Abstract State Machine Languages”, which was presented by Philipp Paulweber, offered an overview of options. Their proposal is CASM: https://github.com/casm-lang/casm

I tried it overnight for Google Java Collection (https://github.com/google/guava.git). A typical maven install takes 10 minutes. For tracing, I used strace -f –output-separately –output=guava-test.trace –decode-fds=path,socket,dev –string-limit=255 –timestamps=format:time,precision:us –syscall-times –decode-fds=all –summary mvn test –log-file guava-test.log –debug -o -Dorg.slf4j.simpleLogger.showDateTime=true -Dorg.slf4j.simpleLogger.dateTimeFormat=HH:mm:ss,SSS.

Basically I traced everything I could. I also recorded the results of Maven, with timestamps so I could later relate that with the trace results. I took one hour to finish running (one order of magnitude slower than a typical run). I haven't analyzed the data yet, but I already noticed that Java uses fmutex extensively (which may be worth addressing later).