Report on the Aachen OCL Meeting

As a continuation of the OCL workshop during the MODELS 2013 conference in October 2013, a number of OCL experts decided to meet in November 2013 in Aachen for two days to discuss possible short term improvements of OCL for an upcoming OMG meeting and to envision possible future long-term developments of the language. This paper is a sort of \minutes of the meeting" and intended to quickly inform the OCL community about the discussion topics.

💡 Research Summary

The paper reports on a two‑day meeting of OCL experts held in Aachen in November 2013, convened as a follow‑up to the OCL workshop at the MODELS 2013 conference. The participants gathered to prepare short‑term improvements for the upcoming OMG standardization meeting and to outline a longer‑term vision for the Object Constraint Language. The discussion was organized into four main tracks.

The first track focused on refining the draft of OCL 2.5. Attendees identified ambiguities in the current OCL 2.4 specification—particularly in the type system, collection operations, and Boolean operator precedence—and agreed to introduce a formal type‑inference rule set, explicit definitions for collection literals, and clear precedence tables. Special attention was given to the semantics of “let” expressions and lifecycle functions such as oclIsNew, with the goal of eliminating the uncertainty that arises during model‑change validation.

The second track addressed tighter integration between OCL and UML 2.5. Because the mapping of OCL constraints onto UML elements is presently incomplete, the group proposed a profile‑based extension that directly aligns OCL’s “context” declarations with UML’s Constraint metaclasses. This profile would also prescribe a canonical serialization of OCL abstract syntax trees into UML XMI, thereby improving tool interoperability and enabling automated validation steps within model‑driven development pipelines.

The third track examined tooling and execution efficiency. Existing tools (Eclipse OCL, MagicDraw, IBM Rational, etc.) only partially support OCL, leading to fragmented user experiences. Participants recommended the definition of a standard OCL API and plug‑in interface to foster cross‑tool compatibility. Performance‑related enhancements were also discussed: a standardized lazy‑evaluation strategy, short‑circuit Boolean evaluation, and a streaming evaluation mode designed to reduce memory consumption on very large models. In addition, a prototype web‑service and containerized cloud execution environment for OCL were slated for development, aiming to make OCL validation accessible as a service.

The fourth track explored long‑term strategic directions and community building. The vision is to evolve OCL from a pure constraint language into a “semantic interface” that underpins broader model‑based system engineering activities. This includes coupling OCL with model‑to‑model transformation languages to express transformation rules, integrating OCL with simulation engines for executable scenario checking, and investigating machine‑learning techniques for automatic constraint extraction from model artefacts.

Community activation was identified as essential for sustaining these efforts. The group proposed a series of regular workshops, an online discussion forum, and multilingual educational material (tutorials, textbooks, video courses). Partnerships with universities and industry were encouraged to generate real‑world case studies, which would feed back into the standardization process.

The outcomes of the meeting were captured as concrete action items to be submitted to the OMG OCL working group. Responsibilities, timelines, and deliverables were documented, establishing a roadmap that simultaneously pursues short‑term standard refinements and long‑term ecosystem growth. The authors conclude that, by addressing the identified technical gaps and fostering a vibrant community, OCL can solidify its role as a cornerstone of model‑based development, providing robust, interoperable, and performant constraint validation across a wide range of modeling tools and domains.