Repository contents:

• Go to plugins to inspect the sources of the simulator and the test code.
• Go to raw-results to inspect the execution logs, *.dat files generated from them, and the plots generated with gnuplot.

This study aims to evaluate the possible distinct behavior of current in-memory solutions without dedicated temporal support (what we will call StandardEMF) and our temporal solution (TemporalEMF). Specifically, we aim to answer the following questions:

RQ1: Is there a significant difference of the time required for producing and manipulating temporal model elements?

RQ2: Is there a significant difference in the memory used and the storage size of temporal models?

RQ3: Is there a significant difference of restoring previous versions of temporal model elements? For an implementation of the queries used to evaluate this (Q1-Q4), please check plugins/edu.uoc.som.tll.simulation/src/test/SimulateXmi.java and plugins/edu.uoc.som.tll.simulation/src/test/SimulateTemf.java.

In all the experiments we have used a model with a single System and varying SimConfig parameters. The base model is available in plugins/edu.uoc.som.tll.simulation/test-instances/singlearea-fullhistory.xmi for StandardEMF and plugins/edu.uoc.som.tll.simulation/test-instances/singlearea-nohistory.xmi for TemporalEMF.

Based on the parameters on the table below, we created the instances available for the different experiments. These test instances are available in raw-results/Xmx2g.exectime/benchmark-instances, raw-results/Xmx2g.memory/benchmark-instances, and raw-results/Xmx512m.memory/benchmark-instances.

From the experiments listed above, a set of numerical data and graphs has been gathered and produced for an easy interpretation of results. For a better organization, these results have been included in their corresponding subfolders, together with the raw data:

• For RQ1 see raw-results/Xmx2g.exectime/README.md,
• For RQ2 and RQ3 see raw-results/Xmx2g.memory/README.md, and raw-results/Xmx512m.memory/README.md.