Programs in embedded domain-specific languages are realized as graphs of objects of the host language rather than as static input texts. This property enables dynamic meta-programming, but also makes it harder to attach location information to diagnostic messages that arise at a later stage, after the program graph construction. Thus, EDSL-generating expressions and algorithms can be difficult to debug. Here, we present a technique for transparently capturing and replaying location information about the origin of EDSL program objects. It has been implemented in the context of the LLJava-live EDSL-to-bytecode compiler framework on the JVM. The basic idea can be generalized to other contexts, and to any managed runtime environment with reified stack traces.

Units of measure with prefixes and conversion rules are given a formal semantic model in terms of categorial group theory. Basic structures and both natural and contingent semantic operations are defined. Conversion rules are represented as a class of ternary relations with both group-like and category-like properties. A hierarchy of subclasses is explored, each satisfying stronger useful algebraic properties than the preceding, culminating in a direct efficient conversion-by-rewriting algorithm.

This paper investigates the use of staged meta-programming techniques for the transparent acceleration of embedded domain-specific languages on the Java platform. LLJava-live, the staged API of the low-level JVM language LLJava, can be used to complement an interpreted EDSL with orthogonal and extensible compilation facilities. Compiled JVM bytecode becomes available immediately as an extension of the running host program. The approach is illustrated with a didactic structured imperative programming language, Whilst.

XSLT 1.0 is a standardized functional programming language and widely used for defining transformations on XML models and documents, in many areas of industry and publishing. The problem of XSLT type checking is to verify that a given transformation, when applied to an input which conforms to a given structure definition, e.g. an XML DTD, will always produce an output which adheres to a second structure definition. This problem is known to be undecidable for the full range of XSLT and document structure definition languages. Either one or both of them must be significantly restricted, or only approximations can be calculated. The algorithm presented here takes a different approach towards type correct XSLT transformations. It does not consider the type of the input document at all. Instead it parses the fragments of the result document contained verbatim in the transformation code and verifies that these can potentially appear in the result language, as defined by a given DTD. This is a kind of abstract interpretation, which can be executed on the fly and in linear time when parsing the XSLT program. Generated error messages are located accurately to a child subsequence of a single result element node. Apparently the method eliminates a considerable share of XSLT programming errors, on the same order of magnitude as a full fledged global control-flow analysis.

Models link empirical observations with formal reasoning. If the computer plays an essential role in their functioning, it seems only fair to consider theoretical computer science as well as the discipline of application. We review some perennial philosophical problems about the nature of state and behaviour in models from that perspective. A mathematical framework for their explication and the unbiased choice of solutions is proposed. Agent-based models play an increasing role in ecological modelling. They are taken here as an example for demonstrating critical assumptions and dilemmas in integrating criteria of methodological rigour with practical relevance of models. Some possible solutions and their practical implications are outlined based on this new theoretical perspective.

Reference-counting garbage collection is known to have problems with the collection of cyclically connected data. There are two historically significant styles of cycle-aware algorithms: The style of Brownbridge that maintains a subset of marked edges and the invariant that every cycle contains at least one marked edge, and the style of Martinez-Lins-Wachenchauzer (MLW) that involves local mark-and-scan procedures to detect cycles. The former is known to be difficult to design and implement correctly, and the latter to have pathological efficiency for a number of very typical situations. We present a novel algorithm that combines both approaches to obtain reasonably efficient local mark-and-scan phases with a marking invariant that is rather cheap to maintain. We demonstrate that the assumptions of this algorithm about mutator activity patterns make it well-suited, but not limited, to a functional programming technique for cyclic data. We evaluate the approach in comparison with simple and more sophisticated MLW algorithms using a simple benchmark based on that functional paradigm.