bsky.app/profile/aspi...

bsky.app/profile/aspi...

bsky.app/profile/aspi...

bsky.app/profile/aspi...

A quick thread-of-thread of my observation on my small sheaves-in-Haskell project from the other week.

Par curiosité, où était le collège à l'époque?

J'en profite pour enfin prendre le temps d'expliquer pourquoi
1. les frais d'inscriptions des étudiants étrangers concernent tout le monde
2. à partir du moment où on passe dans un système où l'étudiant (et les familles) supportent le coût, les frais d'inscriptions ne peuvent que s'envoler

…and I've yet to know the nuance.

So maybe there's an actual difference between unaccusative verbs (and ergative-alignment) and passive voice (besides one being the base form and the other being a derived form) that's worth making.

Japanese isn't listed in the ergative-aligned languages in Wikipedia. Maybe because it does have a passive-voice like construction (it also has a causative that goes the other way around though). When you have a choice between a passive voice and an unaccusative alternative, there's a right choice…

Japanese being a language with a lot of unaccusative verbs. I'm confident that it feels very alien, despite unaccusative verbs and passive voice existing in European language.

(hello random internet stranger, if you find that message, it was satire, I'm actually only metaphorically fuming)

*THAT'S* what your complaining about? When Pterodactyl is in the list and not even a dinosaur‽ I'm literally fuming right now!

Oui, la grosse innovation c'était Mitterrand et sa dissolution systématique. L'alignement des calendrier ne fait qu'en prendre la suite (et c'était clairement transpartisan, tous les politiciens voulaient ça pour eux…).

@gro-tsen.bsky.social You'll probably want to watch this one if you haven't already. It could just as well be custom-made content for you.

No one can force me to have a secure website!!! YouTube video by suckerpinch

It's this time of the year. Tom7 has a new project video.

Brilliant as always. Less delight than usual, but much more spite. www.youtube.com/watch?v=M1si...

But the important remark is that the `forall w` forces terms to be properly parametric, hence rules out exotic terms. Even when `v` is fixed. So I would argue that this is truer to the intent of PHOAS than the standard style of binders. 3/3

I used `Lam :: forall w. (v -> w) -> (w -> Term w) -> Term v` instead of the more common `Lam :: (v -> Term v) -> Term v`. I had to use this term so that I could give a more direct interpreter into (pre)sheaves. It came out of necessity. 2/3

Another thought on PHOAS, it's always been disturbing to me that while `forall v. Term v` only has terms, a `Term V` for a concrete `V` can have exotic terms. It struck me as too easy to misuse. But in the representation I had to use I don't have this problem. 1/3

It doesn't make actual formal sense, but in both case you have a two types instead of one, which requires a bit more care to use but allows for a little more expression. 2/2

I've had a thought, too that, PHOAS is a bit to HOAS what contextual CCCs are to CCCs. In that metaphor, contexts are terms and types are the (free) variables. 1/2

I should say: their goal is a place. They're just traversing land. And do random stuff on the way. And that's about it.

I've just finished watching Frieren's season 2. It's an improvement on the first season. In that nothing happens. No power-curve climbing, no tournament. Just steady progress toward their goal. And random shenanigans. It feels truer to its premise.

J'avais vu ce tweet tantôt. Et je m'étais convaincu, à l'époque, que c'était une réponse sarcastique (en ceci que l'auteur sait bien que c'est une propriété évidente) à un tweet pas bien malin.

Colloque en hommage à Gilles Dowek - Blog binaire Date: 19 juin 2026, 9:30-19:30 Lieu: Amphithéâtre Alain Aspect, ENS Paris Saclay, 4 avenue des Sciences, 91190 Gif-sur-Yvette, France Inscription: formulaire (inscription gratuite mais obligatoire avant le 31 mai) Gilles Dowek, directeur de recherche à l’INRIA et professeur à l’Ecole Polytechnique puis à l’Ecole Normale Supérieure de Paris-Saclay, est décédé le 21 juillet 2025 à […]

Colloque en l’honneur de Gilles Dowek à l’ENS Paris saclay

19 juin 2026, 9:30-19:30

blogbinaire.larecherche.fr/2026/04/10/c...

Ça permet de relativiser les annonces tonitruantes qu'on entend sur ce qui se passe dans les autres pays, je suppose.

The time it takes to write a joke, from conception to birth, is the jestation period.

Nintendo is currently deleting thousands of the most popular Mario Maker levels ever made. A couple of days ago Nintendo started to seemingly randomly delete popular levels off of Super Mario Maker’s servers. At first the community was flabbergasted because the deletions seemed to follow…

It would appear that Nintendo is currently mistakenly deleting a very large number of Mario Maker 2 levels (might even be hard deletes? which would be unreasonable). Another example of how the social internet needs positive rights for users too.
www.youtube.com/watch?v=Hb3N...

And this instantiate to our case, I believe: we take contexts to be the category of presheaves, and types to be sheaves. So yes, this does falls out of a general case. Just not the one we're usually taught. Yet one almost as simple. 8/8

You can interpret the simply typed λ-calculus in any contextual CCC in much the standard way. Types are interpreted to types of the category, and contexts to arbitrary contexts. 7/8

Here, you have a category (of contexts) and a distinguish full category of types. And if you have a context Γ and a type A, there's a product context Γ×A. And there's an exponential *type* A⇒B for a pair of types, appropriately adjoint to the product. 6/8