Anders Møller

Thanks for the suggestion. Just a quick comment: Random strings could be generated with many different probability distributions, and it is not clear why this one is particularly useful.

Thanks for the pointer! Please note that dk.brics.automaton already supports multiple minimization algorithms, including Valmari's (https://github.com/cs-au-dk/dk.brics.automaton/blob/master/src/dk/brics/automaton/MinimizationOperations.java). The paper you refer to unfortunately only contains an experimental comparison with the Hopcroft...

The complement operations involve 'determinize', so this may not be a good idea. Do you have some experimental evidence that your suggestion is in fact faster in general?

Provided that further experiments are convinving: Let's add a static field `intersection_de_morgan` (false by default) and a method `setIntersectionDeMorgan` (like `setMinimizeAlways`) to control which version to use.

The suggested change apparently doesn't support Unicode supplementary characters (surrogate pairs)?

The easiest is probably to add an automaton conversion operation (in SpecialOperations.java) that, for example, converts from lower-case to upper-case in all transitions, and then apply a similar transformation to...

We're working on it... :-)

Latest release has support for much of ES5, especially getters and setters

We shifted to IntelliJ IDEA long time ago. We'll look into the problem, but I suggest you use that or command-line compilation for now.

PRs with tests are very welcome!