We are already playing with a testing modality (tests/equiv), where ADDA is run with command lines that must lead to exactly the same outcome. Right now there are only a couple of specific tests (one to be merged soon - #315), so it is less general (sensitive) than the ones in #6. However, it is much simpler and more robust with respect to different compiler versions/modes. So the two (system-)testing modalities can complement each other.

The goal of this issues is to make a general framework for such equivalent testing. A general test script (python or bash), which will read one or several files with pairs of equivalent command lines. Can be similar to (tests/2exec) but simpler, requiring less special workarounds.