Scottish Programming Languages Seminar (SPLS)

University of Strathclyde, 17th October 2018

Location: Royal College building

Talks: Room 641

Food/coffee: Room 513

The rooms are signposted from the main George Street entrance, and Montrose Street (James Weir building) entrance.

About SPLS

The Scottish Programming Languages Seminar is an informal meeting for the discussion of any aspect of programming languages. The next SPLS will take place on Wednesday 17th October at the University of Strathclyde.

Information and updates about the October edition of SPLS will be sent via the SPLS Mailing List.

Time and place

12:00–17:45, 17th October, University of Strathclyde. Royal College building, Room 641.

Registration

Please add your name to the Doodle poll to register. Any dietary requirements should be made as a comment on the doodle poll!

Programme

When What
12.00 Lunch (provided)
13.00 Context Constrained Computation - James Wood
13.30 Type-Driven Development of SoC Architectures - Jan de Muijnck-Hughes
14.00 Coffee
14.30 Industry experience of implementing 80+ DSLs on top of Xtext - Gregory Dyke
15.00 Notebook Programming Considered Harmful - Jeremy Singer
15.30 Where the linear lambdas go - Wen Kokke
16.00 Coffee
16.30 Generating Efficient Parallel Code for Irregular Shaped Arrays - Federico Pizzuti
17.00 Announcements?
17.15 Pub

Abstracts

Context Constrained Computation - James Wood

In normal typed λ-calculi, variables may be used multiple times, in multiple contexts, for multiple reasons, as long as the types agree. The disciplines of linear types and coeffects refine this by tracking how variables are used. For instance, we might track how many times a variable is used, or whether it is used covariantly, contravariantly, or invariantly. Such a discipline yields a general framework of “context constrained computing”, where constraints on variables in the context tell us something interesting about the computation being performed. We will present work in progress on capturing the “intensional” properties of programs via a family of Kripke indexed relational semantics that refines a simple set-theoretic semantics of programs. The value of our approach lies in its generality and the range of examples covered.

Type-Driven Development of SoC Architectures - Jan de Muijnck-Hughes

The protocols that describe the interactions between IP Cores on System-on-a-Chip (SoC) architectures are well-documented, describing not only the structural properties of the physical interfaces but also the behaviour of the emanating signals. However, there is a disconnect between the design of SoC architectures, their formal description, and the verification of their implementation in known hardware description languages. In this talk I will discuss my work in designing a modelling language with a substructural type-system that reasons about the structure of SoC Architectures.

Industry experience of implementing 80+ DSLs on top of Xtext - Gregory Dyke

Avaloq is an international leading Fintech company and the creator of the Avaloq Banking Suite. Domain Specific Languages (DSLs) are used to customize the business logic of the banking suite. Xtext is an Eclipse project that provides rich support for the implementation of DSLs. Xtext is very flexible and easily extensible allowing to apply it way beyond simple flowchart DSLs. At Avaloq we used Xtext to implement tools for over 80 DSLs out of which over 50 languages existed long before we learned about Xtext and hence were designed without Xtext implementation in mind. In this talk we introduce DSL Developer Kit, a family of DSLs built on top of Xtext to help with heavy customization of parsing, indexing, scoping, validation, and formatting.

Notebook Programming Considered Harmful - Jeremy Singer

Millennials write their code in Jupyter Notebooks (formerly known as IPython). Highly evangelized benefits include (1) familiar browser-based coding environment, (2) interactive feedback, (3) inline tutorial commentary and code blocks, and (4) tight integration with popular libraries e.g. NumPy, matplotlib. Many coders, particularly in the data science domain, use Jupyter as their development platform of choice. At Glasgow, we have switched our first year undergraduate coding labs to Jupyter. However Jupyter is not always a bringer of jollity. The notebook abstraction has several serious problems, which expose themselves (at best) as inconveniences to experienced hackers or (at worst) as conceptual misdirections to novice developers. In this talk, I will highlight some of these problems and propose potential fixups.

Where the linear lambdas go - Wen Kokke

Have you ever wondered where the lambdas live, and if you could just grab a random few to see if your compiler works, only to realise “WHOA, that’s kinda hard!” Turns out that linear lambdas live in an even more elusive land, where none of the existing solutions – as far as I know – really apply. A common problem with support for those pesky linear types, really. This talk will drag you though some rather naive solutions, some existing work, and then end rather abruptly with a cry for help. Spoilers.

Generating Efficient Parallel Code for Irregular Shaped Arrays - Federico Pizzuti

Data types in high-level programming languages capture semantic information about the program. Compilers for languages such as Accelerate, Futhark, Delite or Lift have shown how to exploit this information to produce high-performance code. While these approaches have the ability to handle multi-dimensional arrays, their type systems are currently limited to nested arrays of rectangular shapes. This talk presents our work on lifting some of these limitations and enable the representation of irregular, yet statically known shaped, multi-dimensional arrays at the type level. This will enable the Lift compiler to generate efficient code for applications operating on irregular-shaped structure, such as matrix-triangle multiplication or operations on compact representations of tree-like data structures. In particular, the type system is extended by allowing the array element type to depend, in a restricted way, on its position in the enclosing array. We show how existing high-level patterns together with a couple new primitives naturally express computations on such data structures. Finally, we show how classical low-level loop optimisations can be expressed in terms of operations on such irregular arrays, highlighting their utility even in applications that are expressible using more traditional methods.

Cancelled Talk - Folds, Unfolds, and Metaheuristics - Adam Barwell

Metaheuristics are families of algorithms that describe how to achieve acceptable solutions for hard optimisation problems. Current approaches to the design, development, and transformation of metaheuristic algorithms are predominantly ad hoc. In this talk, we present work in progress on a constructive approach to the definition of metaheuristic algorithms that supports mechanical reasoning about well-known algorithms from metaheuristics literature. This facilitates a type-driven rewriting system that is able to synthesise known metaheuristic algorithms. Using our approach, we intend for it to be possible to determine whether two algorithms are equivalent in terms of their results, and to choose an algorithm from a set of equivalent algorithms for a given problem, based on a given cost model, or the overall fitness of results. We conjecture that it is also possible to generate new metaheuristic algorithms using our approach.

Organizers

Acknowledgements

This meeting of SPLS will receive financial support from the Theory, Modelling and Computation theme of the Scottish Informatics and Computer Science Alliance.

SICSA logo