วีดีโอ

Engineers these days are getting younger and younger. Faith in humanity partially restored.

Emiliano Crescent

Theories and models go back to Godel's completeness theorem for FOL (1929) or Birkhoff's universal algebra (1935). Dating them to Landin in 1966 is much too late.

This excellent history omits Eugenio Moggi's "Abstract View of Programming Languages" notes, which he wrote in Spring 1989, while visiting John Mitchell at Stanford. These notes include a complete and detailed discussion of monad transformers, which were not understood by the rest of the community until January 1995, even though Phil Wadler cited Moggi's notes in his "Comprehending Monads" paper (1990) and "Essence of FP" paper (1992). Another important reference is Jon Beck's paper on Distributive Laws from the 1966/67 Seminar on Triples [ triples are monads !] and Categorical Homology Theory, which is available as a TAC reprint. Monads were the new hotness in 1967. 🙂

This is exactly what I was looking for, great!

Funny 61:56 says it does not want high-level language semantics then introduce junks like wasip2 with component model. WebAssembly is dead.

Nice. I like the mvs approach. Seems like it's simpler than dealing with reference lifetimes. But one syntax question. If I understand correctly, the ampersand (&) in an expression like &n += 1 is just a marker to indicate that n is mutated in place. But why is the marker required? Isn't it just visual noise? I'd imagine you'd have to write & all over the place in larger programs.

It's all great, but where is implementation? Is it published?

Can we profile the heatbeat scheduler to determine a near-optimal grain size, and then recompile the algorithm with the determined value? Perhaps call it heartbeat autotuner? What are the disadvantages of using such an approach? Thanks for both the papers.

;da in niss heal

This is the perfect video to show how we shouldn't explain things.

Interesting

no audio :(

899 Eula Flat

Amazing! Looking forward to seeing this implementation in popular RegEx matchers.

Video is for another talk

I want to thank ACM and ICFP for streaming these talks. Thank you for not putting this behind a paywall (I am an ACM subscriber). This allow me to to share and discuss with people that would never be able to make an ICFP event.

Very interesting. Excellent presentation!

Thanks for the video & its bookmarks 🙏

39:31 On the design and implementation of Modular Explicits - Samuel Vivien 1:02:30 Flambda2 Validator - Irene Yoon 1:22:34 A Non-allocating Option - Richard A. Eisenberg 1:44:59 Mixed Blocks: Storing More Fields Flat - Nick Roberts 2:38:09 Structured diagnostics for the OCaml compiler - Florian Angeletti 3:02:23 Project-wide occurrences, a progress report - Ulysse Gerard 3:24:23 Mica: Automated Differential Testing for OCaml Modules - Ernest Ng 3:45:27 First-Class Windows: Building a Roadmap for OCaml on Windows - Sudha Parimala 5:35:31 Picos: Interoperable Effects Based Concurrency for OCaml - Vesa Karvonen 6:02:20 Distributed Actors in OCaml - Wenke Du 6:20:49 Priodomainslib: Prioritized Fine-grained Parallelism for Multicore OCaml - Stefan K Muller 6:46:37 Saturn: a library for verified concurrent data structures for OCaml 5 - Carine Morel & Clement Allain 7:37:01 opam 2.2 and beyond - David Allsopp 8:01:34 Recursion schemes in OCaml: An experience report - Tim Williams 8:22:52 ChorCaml: Functional Choreographic Programming in OCaml - Rokas Urbonas 8:43:25 B.o.B A universal peer to peer file-transfer in OCaml - Romain Calascibetta

Shame they chose Guile instead of Chez

Oh, I remember Martin Odersky taught Coursera's Functional Programming in Scala - one of the most relatable and engaging courses on the platform I've encountered

36:00 MicroHs - Lennart Augustsson 1:06:47 Higher Order Patterns for Rewrite Rules - Jaro Reinders 1:33:05 Welcome to the Parti(tioning) (Functional Pearl) - Robert Krook, Samuel Hammersberg 2:36:50 Analysing uninstrumented Haskell processes using ghc-debug - Zubin Duggal 2:56:22 A zero-copy interface to compact regions - Thomas BAGREL 3:21:18 Building Haskell with Buck2 - Andreas Herrmann 3:41:30 The JavaScript FFI feature in GHC Wasm backend - Cheng Shao 5:35:50 Calculating Compilers Effectively - Zac Garby, Graham Hutton, Patrick Bahr 6:11:00 Cloaca: A Concurrent Hardware Garbage Collector for Non-Strict Functional Languages - Craig Ramsay, Rob Stewart 6:38:25 Functional Reactive Programming, Rearranged - Finnbar Keating <not timestamped> Lightning Talks 8:08:10 17th Haskell Symposium HIW '24 Chairs' Report - J. Garrett Morris

Za Eridan Domoratsky!

This talk made me install scala again. Thank you Martin. I trust your taste. Scala 3 is a masterpiece of readable statically succinct code.

presentation starts at 34:57

50K

Sad that such a great presentations get only few hundreds of view and not millions.

Good presentation ruined by terrible audio

A true joy!

Extremely interesting, thanks Joe!

quality of the video is extremely bad. please correct or provide a better link. thanks

现在对Cubical Agda里面应用Encode-Decode method有什么说法么?

Good work!

Very interesting and clear presentation! Thank you

Pi calculus is underrepresented in industry. Very nice talk.

Great talk!!!

11:30 Does the order matter here? I don’t think the template operator is commutative, is it?

Syntax seems like a mix of Python & Nim. But it is good.

It's weird how Haskell is known as a specified language, but actually, as GHC took over and accrued under-documented extensions, stayed as ad-hoc as it started out before the spec was written. Look at how many implicit language revisions there were before Haskell98, by looking at the GHC source. GHC always required extensions. Cabal-install is just a continuation of this ad-hoc philosophy. What this talk doesn't mention, that I am curious about: Does typing somehow make it more difficult to have a stable API? Does Haskell's language extensions make it more difficult? Dynamic programming makes interfaces implicit. The interface is often just an invariant specified in a comment. If you use the interface as intended, everything works out great. But I played with the Scribble API for generating Idris2 docs, and it seems like there are parts of it that are very ad-hoc too. So how did core Racket manage to develop this stable API?

The sound is terrible

不错 不错

first!1!

Incredible

What if there was APL-inspired lang written on top of TinyGrad?

Thanks for sharing this video! It sparked a great conversation with my friends......

let's descend into the madness of reimplantation!

The mic went out of battery in the middle of the session but thanks to the volunteers, it was fixed very quickly!

prushdown automaton are real

Very cool. First I've seen of interleaving LLM with programming language static analysis. I'd be interested in seeing this applied to a compiler, rather than a post-compilation static analysis pass.