Regression in DetermineEquivalenceTest::findSeparatingWord affecting AutomataLib 0.12.0 but not 0.11.0

[pfg666]

Bug Description
net.automatalib.util.automaton.equivalence.DeterministicEquivalenceTest::findSeparatingWord generates an exception when given certain non-equivalent DFAs . This exception only happens in AutomataLib 0.12.0 and not in AutomataLib 0.11.0.

To Reproduce
I uploaded a ZIP archive containing a Java Main class, two DOT automata, plus Maven boilerplate that can be used reproduce the bug. The Main class simply parses the DOT automata into CompactDFAs, which it supplies to findSeparatingWord. Instead of returning a separating word, the method throws the exception below.

Exception in thread "main" java.lang.ArrayIndexOutOfBoundsException: Index 99 out of bounds for length 11
	at net.automatalib.util.automaton.equivalence.DeterministicEquivalenceTest$ArrayRegistry.putPred(DeterministicEquivalenceTest.java:212)
	at net.automatalib.util.automaton.equivalence.DeterministicEquivalenceTest.findSeparatingWord(DeterministicEquivalenceTest.java:79)
	at findseparatingwordbug.Main.main(Main.java:30)

If you run this same code with AutomataLib 0.11.0 (after making the necessary adjustments to package paths), you will get a correct separating sequence.

Desktop (please complete the following information):

• OS: Windows, Ubuntu
• Java version: 17, 21
• AutomataLib version: 0.12.0

Context
We discovered this as we attempted to port StateMachineBugFinder to the latest version of AutomataLib. We make extensive use of findSeparatingWord.

[mtf90]

It's the same issue as #84 (or rather the incomplete fix that made it into the release) . Since you are the second person to run into this, I probably just prepare a 0.12.1 version.

[kostis]

Yes, we would appreciate it if there were a new version. Thanks @mtf90 !

[mtf90]

@pfg666 @kostis can you check with 0.12.1?

[kostis]

Works like a charm. See this PR. You can close this issue.

Thanks for the quick fix!

[mtf90]

Fixed with 5c72183.