Racket’s macros are fantastic for building DSLs that mix well with general-purpose code and inherit Racket’s extensibility. But they suffer when it comes time to equip your DSL with an optimizing compiler or static analyses. There, the very closeness with Racket that makes language mixing and extension so easy gets in the way. In this talk I’ll present my `syntax-spec` metalanguage. It extends Racket’s macro system to allow specifications of DSLs as fragments of a multi-language: DSL and Racket code mix, but only at carefully delineated boundaries. The result is that optimizing compilation, language mixing, and extension combine constructively rather than clash.

Consisting of more than 15k lines of code and documentation, written in spare time during the last two and a half years, the Frosthaven Manager would not exist without the Racket community. Let’s explore some of the community’s contributions, reflect on the benefits of building in the open, and dream about what comes next.

Mutation testing is the idea that we can assess the completeness of a test suite by updating (or ‘mutating’) a single location in the code under test, and checking to make sure at least one of the existing tests fails. Building on Lukas Lazarek’s mutation framework for Racket, we apply mutation testing in two very different settings—the classroom, and the open-source software project—to see what kinds of benefit mutation testing might offer above and beyond the commonly-used test case coverage metric.

When you add types to a portion of your partially-untyped code base in Typed Racket, the type soundness guarantees you gain will make you feel warm and fuzzy. Sometimes, however, doing so will cause your running time to skyrocket, turning your cute, fluffy type annotations into an unexpected tribulation. When such troubles occur, contract-profile can help you ease the runtime wrath of contract checking. In this talk, I’ll demonstrate how to use contract-profile to boldly reduce the overhead of type-checking in your program, without violating the following prime directive: once a module has been typed, it must stay typed forever.

Gathering with drinks and snacks, located close to a light rail station that is one stop away from the university.

Computation exists in service of data. Useful data structures, therefore, make the difference between what is possible and what is easy. And of the many forms of structured data, sorted data is arguably the most useful. In this talk, we’ll tour the sorted data structures in Rebellion, a second standard library for Racket I’ve developed over the years. We’ll collect data into sorted sets, sorted maps, and range sets. We’ll reduce and transduce data from one collection into another. We’ll order data according to composable comparators. We’ll build data up gradually, then all at once. We’ll query it, mutate it, persistently update it, and concurrently access it. And we’ll do it all with enough performance to wrap up by lunch.

In this talk I present a linguistic comparison of language choices in Racket and LISP vs. a number of other computer languages. LISP is by far the programming language with the simplest syntax: only using parenthesis as delineations. However, the Racket and LISP communities have over time added a great variety of semantics on this simple mechanism, some of which have proven great inventions, and some of which other languages may have better solutions to, and that the Racket and LISP communities may learn from. The talk will directly compare elements of yaml, SmallTalk, Ruby, Java, regular expressions, visual programming, and shell programming with Racket.

Please come with your big questions and discussion topics.