
Benchmark
results for static aheadoftime evaluation of tracematches 


stages 
shows which analysis stage succeeded in removing the last shadows 

PPFs after FI 
shows the number of potential points of failure after applying the
flowinsensitive stage (ECOOP 07) 

PPFs after FS 
shows the number of potential points of failure after also applying the
flowsensitive stage (this paper) 

PPFs removed 
shows the number of potential points of failure removed by the
flowsensitive stage (this paper) 

APFs 
shows the number of actual points of failure; as determined by runtime
monitoring and manual code inspection 

AFP ranks 
ranks assigned to the AFPs by our ranking algorithm 

delegate ranks 
ranks assigned to delegates by our ranking algorithm; we only show
delegates that actually match at runtime 

False positives 
shows the number of false positives, i.e. remaining PPFs minus APFs 

False positives removed % 
shows the quota of false positives removed, over 'PPFs after FI' 


summary times 
gives a summary of the various analysis times 

time cg phase 
time to compute the call graph and contextinsensitive pointsto sets 

time demandcs 
total time which it took the demanddriven refinementbased
contextsensitive analysis to answer its queries 

time flowins 
total time to execute the flowinsensitive analysis stage (ECOOP 07) 

time flowsens 
total time to execute the flowsensitive analysis stage (this paper) 

total compilation time 
total compilation time, including frontend, generation of the IR,
analysis and production of bytecode 


each analysis run is w.r.t. a single method and tracematch 

analysis runs 
total number of analysis runs 

aborted analyses 
number of aborted analysis runs 

jobs per stmt and loop 
number of analysis jobs per statement and loop 

max time analysis run 
maximal time it took to complete a single analysis run 

avg time analysis run 
average time it took to complete a single analysis run 


