iiinova — where your digital self keeps evolving

Reposted by Sjoerd Visscher

Sjoerd Visscher @sjoerdvisscher.w3future.com · 26/09/2026

If you have laws-as-code and you have code-as-diagrams, you get (you guessed it) laws-as-diagrams for free! github.com/sjoerdvissch...

The 6 laws of a traced monoidal category, rendered as diagrams.

042

Sjoerd Visscher @sjoerdvisscher.w3future.com · 26/09/2026

If you have laws-as-code and you have code-as-diagrams, you get (you guessed it) laws-as-diagrams for free! github.com/sjoerdvissch...

042

Sjoerd Visscher @sjoerdvisscher.w3future.com · 25/09/2026

Wij hebben de parasols al opgeborgen!

100

Reposted by Sjoerd Visscher

terrorjack.bsky.social @terrorjack.bsky.social · 19/09/2026

playground.terrorjack.workers.dev weekend toy project: haskell playground based on ghc-10.0, with multi-file editor, ghci repl and haskell language server; transitive deps of hls like aeson are also included. note that this is a vibe coded poc, if you have allergies to this kind of thing

playground.terrorjack.workers.dev

Haskell playground

232

Sjoerd Visscher @sjoerdvisscher.w3future.com · 16/09/2026

I would have thought it would be less work to handle this properly than to throw this error message.

000

Reposted by Sjoerd Visscher

Thomas Marsh @thomasmarsh.bsky.social · 14/09/2026

Apologies for the AI use, but I wanted to explore this in Oleg's visual style. Probably a lot of errors - I know spent too much time trying to get this even to this point. Script to generate it: gist.github.com/thomasmarsh/...

111

Sjoerd Visscher @sjoerdvisscher.w3future.com · 14/09/2026

It's in, but I had to drop the ascii again.

New optics diagram, now including Glass.

010

Sjoerd Visscher @sjoerdvisscher.w3future.com · 13/09/2026

Oh, they sit above lenses and grates, what a mess!

110

Sjoerd Visscher @sjoerdvisscher.w3future.com · 12/09/2026

Sorry that I didn’t reply, but it’s a bit hard to explain and somewhat subtle. You’re better off reading a blogpost about them I think.

110

Sjoerd Visscher @sjoerdvisscher.w3future.com · 11/09/2026

I drafted an article explaining it in more detail: github.com/sjoerdvissch...

120

Reposted by Sjoerd Visscher

Arnaud Spiwack @aspiwack.bsky.social · 09/09/2026

That is, for any object A, we have an arrow α_A : F A -> G A, such that for all R∈Rel(A,B), F(R) ⊆ G(R(α,α)) If F is a functor, parametric arrows should be natural, for instance. This should give us an algebraic framework to talk about what we've been doing in PLT. 6/7

101

Sjoerd Visscher @sjoerdvisscher.w3future.com · 08/09/2026

The main trick is to generalize strength/tambara from p a b -> p (Act m a) (Act m b) for some monoidal action to f ⊙ p ⊙ g -> p, for some subset of (f,g) profunctor pairs. If f ⊣ g then you get a lawful optic. But for example a getter requires g to be the terminal profunctor.

111

Sjoerd Visscher @sjoerdvisscher.w3future.com · 08/09/2026

This is part of proarrow. I will put it on hackage soon!

110

Sjoerd Visscher @sjoerdvisscher.w3future.com · 08/09/2026

Probably no single one is new. What might be new is that they are all together in a single library with automatic conversion between the connected ones. What’s definitely new is the way they are implemented.

111

Sjoerd Visscher @sjoerdvisscher.w3future.com · 07/09/2026

New optics just dropped! #haskell #category-theory

Optics diagram, showing:

-- > Iso <: Lens, MonoidalLens, Prism, Kaleidoscope
-- > Lens <: Getter, AffineTraversal
-- > MonoidalLens <: Getter, MonoidalTraversal
-- > Prism <: Review, AffineTraversal, MonoidalTraversal
-- > Kaleidoscope <: MonoidalTraversal, Grate
-- > Grate <: Setter
-- > Getter <: AffineFold
-- > AffineTraversal <: AffineFold, Traversal
-- > MonoidalTraversal <: Traversal
-- > AffineFold <: Fold
-- > Traversal <: Fold, Setter

350

Sjoerd Visscher @sjoerdvisscher.w3future.com · 02/09/2026

Pretty cool that (*) + (-) means loosely what is says "add the product to the difference", i.e. a function that adds the product of its arguments to the difference of its arguments.

010

Sjoerd Visscher @sjoerdvisscher.w3future.com · 31/08/2026

Zeker de moeite waard! Ik heb ze in 2013 in het Evoluon gezien.

100

Sjoerd Visscher @sjoerdvisscher.w3future.com · 27/08/2026

I’ve been working on proarrow for 3 years and only now I find out that all adjunctions between profunctors are trivial, as in they are always adjunctions between the two representables induced by a functor. At least it’s true for Hask enriched profunctors. Something to do with Cauchy completeness.

020

Reposted by Sjoerd Visscher

Haskell Interlude @haskellinterlude.bsky.social · 26/08/2026

In today’s episode, we’re joined by Brent Yorgey, Associate Professor of CS at Hendrix College and the creator of the Diagrams vector graphics library and Swarm, a 2D programming and resource gathering game. We talk about teaching, DSLs, and competitive coding. haskell.foundation/podcast/85/

haskell.foundation

Brent Yorgey

In today's episode, we're joined by Brent Yorgey, Associate Professor of Computer Science at Hendrix College and the creator, among other things, of the Diagrams vector graphics library and Swarm, a 2...

096

Sjoerd Visscher @sjoerdvisscher.w3future.com · 24/08/2026

It seems like they read it as "stop using X for all official purposes of any entity" instead of "stop using X for all official purposes of the UK Government".

000

Sjoerd Visscher @sjoerdvisscher.w3future.com · 24/08/2026

Yes, but in those cases there's no feedback loop. With code, if the quality of the output goes down, the quality of the input goes down too and with that the usefulness of the tool. If the code is a mess, the AIs struggle too.

120

Sjoerd Visscher @sjoerdvisscher.w3future.com · 21/08/2026

A lot of people have a different association with KI, especially if you grew up among cow farmers like me.

010

Sjoerd Visscher @sjoerdvisscher.w3future.com · 21/08/2026

“Even though”, just like (same etymology even) the al in albeit.

010

Sjoerd Visscher @sjoerdvisscher.w3future.com · 20/08/2026

Dutch has this word "al", which frequently starts a sentence, and then it looks like "Al", indistinguishable from "AI" in most fonts, really annoying!

210

Reposted by Sjoerd Visscher

Conor Sheridan @qucat.bsky.social · 19/08/2026

Cool observation: one can build a bicategory whose objects are free word monoids over alphabets and whose morphisms are Turing transducers. A Turing machine then appears as a specialization of a guarded trace between a Moore machine (tape) and a Mealy machine (controller).

001

Sjoerd Visscher @sjoerdvisscher.w3future.com · 17/08/2026

It also includes a fix for a bug I reported , but I realized only last week that it only fixes the minimized case, not my original code!

010

Reposted by Sjoerd Visscher

Conor Sheridan @qucat.bsky.social · 14/08/2026

Looking for non-trivial examples of virtual double categories arising in TCS or logic, ideally many-object examples where the virtuality has a concrete computational/logical meaning.

011

Sjoerd Visscher @sjoerdvisscher.w3future.com · 11/08/2026

Yes this, but then say every hour a new developer. AIs are really bad at learning from what they did previously.

100

Sjoerd Visscher @sjoerdvisscher.w3future.com · 11/08/2026

Using AI to write your code is like hiring a consultant to write your code, but continuously hiring a different one. No matter how good you set it up, documenting everything etc., over time the result is terrible.

111

Reposted by Sjoerd Visscher

Zanzi Tangle @zanzi.bsky.social · 06/08/2026

after two years of development, I finally demoed Jermaine for the first time - a language with fully symmetric data and codata, and native call-cc.

5447

Sjoerd Visscher @sjoerdvisscher.w3future.com · 06/08/2026

Well it's a hobby so I don't consider discipline to be a factor. The question is: will it be more enjoyable if I can share the fun?

110

Sjoerd Visscher @sjoerdvisscher.w3future.com · 06/08/2026

Thanks! The downside though is that I'm having too much fun adding stuff instead of doing the things that makes this library actually useful for other people like writing documentation...

120

Sjoerd Visscher @sjoerdvisscher.w3future.com · 06/08/2026

I never thought the proarrow library would grow this big. I expected to hit a major roadblock at some point but that just didn't happen. sjoerdvisscher.github.io/proarrow/ #haskell

The contents of the proarrow library listing its 137 modules.

160

Sjoerd Visscher @sjoerdvisscher.w3future.com · 31/07/2026

What is the difference? David doesn’t say.

100

Reposted by Sjoerd Visscher

Arnaud Spiwack @aspiwack.bsky.social · 28/07/2026

I want to explain what's going on here. Because it's quite interesting. The type of monads over the category of presheaves is given by class PMonad f where return :: p i -> f i (>>=) :: f p i -> (forall j. p j -> f q j) -> f q i (just follow the textbooks) 1/11

175

Sjoerd Visscher @sjoerdvisscher.w3future.com · 24/07/2026

I was hoping to be able to write `class Seg (s :: Time) (t :: Time) | t -> s where seg :: s &> t`, then `leap` would simply be `withDict @(Seg s t)`. But this loops...

000

Sjoerd Visscher @sjoerdvisscher.w3future.com · 24/07/2026

Yeah, I rewrote the code without magic, and there's only one place (where it says `k lesson t`) where two available morphisms are composed automatically. But I guess with a compiler for a more complicated language there would be more opportunities.

100

Sjoerd Visscher @sjoerdvisscher.w3future.com · 24/07/2026

I played around with the code a bit to see if I could get it to work without unsafeCoerce, using GHC's withDict. But there's really no wiggle room at all, the compiler quickly goes into an infinite loop.

200

Sjoerd Visscher @sjoerdvisscher.w3future.com · 24/07/2026

What an amazing hack! This is a trick that works for any category, reifying any morphisms you have available into dictionaries, and then letting the compiler compose them together automatically to get the right morphism you need! #haskell

131

Sjoerd Visscher @sjoerdvisscher.w3future.com · 14/07/2026

I have an elasticube wooden puzzle at home, and yesterday I saw a version in a museumshop with balls instead of cubes. This allows for different shapes to be made, but the smooth surface of the balls caused the shapes to lack stiffness.

ochafik.com

OpenSCAD Playground

OpenSCAD Playground: edit .scad models and render them to STL from the browser (WebAssembly build OpenSCAD, with Monaco editor and STL Viewer)

110

Reposted by Sjoerd Visscher

ICFP Programming Contest 2026 @icfpcontest.bsky.social · 11/07/2026

Pre-registration for this year's ICFP programming contest is open! You may register your team at your leisure at icfpcontest2026.com/signup The contest begins on July 24th - just under two weeks to go!

icfpcontest2026.com

053