IndexOutOfBoundsException in IncrementalMealyDAGBuilder when input word is longer than output word

[BlobbyBob]

IndexOutOfBoundsException in IncrementalMealyDAGBuilder when input word is longer than output word

[mtf90]

Dear @BlobbyBob, the issue is that Mealy machines typically map inputs of length n to outputs of length n and the implementation assumes that this invariant holds. Now, inserting a trace A A that should map to A trips the implementation and the exception occcurs. This can probably be reported by a more meaningful error message, but it is primarily a usage error in my eyes.

[BlobbyBob]

Thanks for your response. So it is probably best to open the issue in the learnlib repo instead?

Nonetheless, a more meaningful error message or an assertion that both word length match would be very helpful instead of throwing a "generic" exception.

[mtf90]

Hm, you wrote that the issue occurred when using a SULCache. Typically, the SUL interface also suggests one-for-one semantics, so I wonder how you managed to construct traces with unequal length in the first place.

[BlobbyBob]

That is indeed an interesting question. I assumed the SULCache was at fault as the issue did not occur when I commented it out. I have made some other changes to the setup in the meantime, but I will try to find a minimal working example again.

[BlobbyBob]

As this might take some time, I will close the issue until then

[mtf90]

Sure, feel free to open it again if it turns out to be an actual issue with the cache. You may also enable assertions (-ea) to activate additional checks that may help you tracking down the issue.

If you are unsure about certain concepts in AutomataLib/LearnLib, don't hesitate to open a discussion thread either.