GRaDS talk: Semantic Preservation in Extensible Languages

by Dawn Michaelson on 2020-10-02



Extensible languages, composed of a host and independently-developed extensions, allow users to select the language features they wish to use, which simplifies programming. However, there is currently no strong guarantee that independent extensions will interoperate correctly, so some of the features introduced in one extension may not work correctly when other extensions are included. We introduce a method for proving that independent extensions will preserve language properties when composed. Our method allows both the host and extensions to introduce language properties defining correct behavior of the semantics introduced. The host and extensions have modular proof obligations for their properties, which will provide a full proof of their properties even when other, unknown extensions are added to the language.

About the speaker

Dawn Michaelson is a fourth-year graduate student officially advised by Eric Van Wyk and unofficially advised by Gopalan Nadathur. She is working on proving metatheoretic properties of extensible languages and using monads implicitly in attribute grammars. She is originally from Wisconsin.