The Object Management Group made the unified modeling language (UML) more complex when they moved from UML1; we should have expected snags with UML2. This paper describes at least half-a-dozen traps with UML2 activity diagrams. Most stem from their novel, informal, and complex semantics. This is a familiar phenomenon in the history of computer languages [1]. For example, it is not clear what happens when objects flow through control nodes.

Schattkowsky and Forster point to many other troubling decisions in the standard. The authors claim that these will cause difficulties. I concur. In particular, they mention a potential difficulty that Fowler noted in 2003 [2]. Many people will not realize that when many edges leave an action, there is an implicit fork. There is an implicit join when many edges enter an action. Fowler recommends making the forks and joins explicit, and only having one edge into and one edge out of an action. The authors do not refer to Fowler or mention his proposal. However, the remaining traps have adequate citations. The paper is difficult to read, but the people revising the standard must study it. Meanwhile, trainers and practitioners need a clear roadmap to avoid these pitfalls.