Our main research interests are in declarative specifications for programming language syntax, semantics, and optimizing transformations. We are specifically interested in techniques that lead to a high degree of modularity in the composition of language specifications. We have designed a few unique tools that automatically compose and implement such specifications to create pre-processors, compilers and optimizers for the newly specified languages.
Specifically we are interested in tools and techniques that allow a programmer, who is not an expert in language or compiler construction, to easily import independently-developed (domain-specific) language features into their programming language. These features may add new syntax, semantic analyses, and optimizations to the language. This extended language raises the level of abstraction to that of the task at hand and, we conjecture, makes software development less time consuming and less error-prone.
There are many research efforts in extensible languages to support the highly modular design of programming languages. We are interested in a specific view of extensible languages and compilers in which the composition of language features is directed by a programmer that need not be an expert in language and compiler constructions.
This view leads to a few criteria that we believe must be satisfied for extensible languages or extensible compilers to have a long-term impact:
- Language extensions can be designed by independent parties.
- Language extensions can add new syntactic constructs.
- Language extensions can add new semantic analyses of these constructs and on constructs in the host language that is being extended.
- The composition of the extensions chosen by the non-expert programmer must succeed and form a working compiler or translator.
These criteria have some implications:
Criteria 2 and 3 require that the extensible language/compiler must solve the “Expression Problem”: that is, both new syntax and new semantics can be added without the modification of the host language specification of other language extensions.
By adding criterion 1, we need to solve a specific version of the expression problem, that one extension need not be aware of another.
Adding criterion 4 requires that the composition must be automatic and that no “glue code” be written to combine the language extensions.
To solve this problem, language and extension specifications must be easy to compose and some modular analyses, performed by the language extension writers, need to provide the assurance that the extension has the characteristics needed to compose with other extensions.
Much of our research is evaluated by writing software that realizes these ideas. Collectively, the tools and specifications described below satisfy the above criteria.
ableC is our primary vehicle for investigating extensible languages. This specification implements the C11 standard of the C programming language.
Silver is our attribute grammar system. The specifications for ableC and its extensions are written in the Silver AG language.
Uniquely, Silver supports a modular analysis that extension writers can use to certify that their extension will compose with other independently-developed, and similarly certified, extensions to form a well-defined attribute grammar. Essentially, this ensures that the composed attribute grammar will work.
Copper is a parser and scanner generator that generates integrated LR parsers and context-aware scanners from language specifications based on context-free grammars and regular expressions. The generated scanners use context (in the form of the current LR parse state) to be more discriminating in recognizing tokens. This is useful in extensible language settings and has the nice benefit of making LR parse table conflicts less likely.
Copper also has a modular analysis that can be used to ensure that certified language extension grammars will compose into a deterministic grammar.
Many of our software artifacts (at significant releases) are archived on the Data Repository of the University of Minnesota (DRUM) and can be found from the MELT group landing page on the DRUM.
Some Ph.D. dissertations from group members can also be found from the MELT landing page.
We are very grateful for funding from the National Science Foundation, the McKnight Foundation, the University of Minnesota, and IBM for funding different aspects of our research.