Plenary Talk today, which pulls together all of the conference attendees. Sunday's talk was based in databases, with Michael Stonebraker speaking on his Turing-award winning work. Monday's talk discussed interdisciplinary work, primarily centered in CS theory, and was given by Andrew Yao (a prior Turing Award winner). On Tuesday, Olivier Temam discussed neural networks in hardware, which focused on his work and efforts to better model or mimic the capabilities of the brain.
The F# Path to Relaxation -
There are opportunities to introduce new work toward relaxing and improving. Or perhaps create opposing camps. Thesis <-> Antithesis ==> synthesis. Or Functional <=> Interop. Back in 2003, functional languages were isolated, non-interoperable, using their own VMs. F# (along with Scala, Swift, ...) instead seeks to have an exosystem, being the external industry-standard runtimes. Another tension is between Enterprise and Openness. So F# is open and cross-platform. Tools are available for Android and iOS, as well as packages for Linux.
Functional <=> Objects
Thus embrace objects, without being object-oriented. Some cases in the cross-product of the expected features for objects and functions requires particular care for synthesis.
Circularities and Modularity in the Wild
Lambdas, generics, etc are clearly being embraced in modern language design. However, circular type dependencies are unfortunately also widely present. Languages need to enforce acyclicity.
Pattern Matching <=> Abstraction
How does the language support the functional concept of pattern matching, when you want to include type abstraction? Alas, the speaker skipped the solution quickly.
Code <=> Data
Most development is to providing tools for the information revolution. There is exponential growth in Open APIs for accessing data from the internet. This data then comes with dynamic types, where the types are only known once the data (or schema) has been accessed. The type creation can also enable blending code for other languages into the F# environment. For example, the support can allow opening csv or json files and having types for the data. This feature is, by far, the most exciting and interesting of the presentation. Not quite worth the price of admission, but clearly a great development.
Applied PL design comes from the synthesis at the heart of these contradictions. This tension also is part of the proliferation of languages.
No comments:
Post a Comment