Compiler Construction

We present a compiler which can be used to automatically obtain efficient Java implementations of parsing algorithms from formal specifications expressed as parsing schemata. The system performs an analysis of the inference rules in the input schemata in order to determine the best data structures and indexes to use, and ensure that the generated implementations are efficient. The system described is general enough to be able to handle all kinds of schemata for different grammar formalisms, such as context-free grammars and tree-adjoining grammars, and it provides an extensibilitymechanismallowing the user to define custom notational elements. This compiler has proven very useful for analyzing, prototyping and comparing natural language parsers in real domains, as can be seen in the empirical examples provided at the end of the article.

This paper is based on research results achieved by a collaboration between Ericsson Hungary Ltd. and the Large Scale Testing Research Lab of Eötvös Loránd University, Budapest. We present design issues and empirical observations on extending an existing industrial toolset with a new intermediate language.
Context: The industry partner’s toolset is using C/C++ as an inter-mediate language, providing good execution performance, but “somewhat long” build times, offering a sub-optimal experience for users.
Objective: In cooperation with our industry partner our task was to perform  an  experiment  with  Java  as  a  different  intermediate  language and evaluate results, to see if this could improve build times.
Method: We extended the mentioned toolset to use Java as an inter-mediate language.
Results: Our measurements show that using Java as an intermediate language  improves  build  times  significantly.  We  also  found  that,  while the runtime performance of C/C++ is better in some situations, Java, at least in our testing scenarios, can be a viable alternative to improve developer productivity.
Our contribution is unique in the sense that both ways of building and execution can use the same source code as input, written in the same language, generate intermediate codes with the same high-level structure, compile into executables that are configured using the same files, run on the same machine, show the same behaviour and generate the same logs.
Conclusions: We created an alternative build pipeline that might enhance the productivity of our industry partner’s test developers by reducing the length of builds during their daily work.