5.1 User-definable time-out

Resource : ATB*PR*Time_Out.
Value : positive integer.
Meaning : time-out in seconds.
Default value : 300 (seconds).

This option allows to modify the time-out value of the satellite provers PP (Predicate Prover) and ML (Mono Lemma Prover) in user-definable mode “User_Pass” (see chapter 9 page §) or replay mode “Replay”.

This option enables to test, when in User Pass, proof tactics that use massively the predicate prover. This opportunity of customization therefore enables to launch proofs with small time-outs (maximum time of computation allowed before the proof process stops) so one can quickly test such a tactic efficiency.

Alternatively one can also increase PP and ML time-outs when one is replaying (prove replay) a project on a slower machine: thus one can be confident that if some proofs succeed on fast machines, they will also succeed on slower ones.

Example :

If we have the following User_Pass theory at our disposal:


 
THEORY User_Pass IS  
    ff(0) & dd(0) & pp(rp.0)  
END  
 


and if we want to know whether this series of commands is efficient, we just have to set resource Time_Out to a small value, 10 seconds for instance, and start the automatic proof in User_Pass mode. Generally if a command run-time (for instance, the one of pp) lasts more than 10 seconds, it suggests that this command will never succeed. So if the series above succeeds in less than 10 seconds, it’s OK and the series is efficient, on the contrary, if one of its commands is still running after 10 seconds, the proof is stopped and the user may consider that the series is inefficient.