@[email protected] — Jiří Beneš

PhD student at University of Tübingen. Interested in type systems and (co)effect systems.
You can find me at @[email protected]

[Jonathan Brachthäuser @__protected]

HOPE will take place tomorrow at ICFP (and in Paris)!

There are three talks by members (@jiribenes_) and students of the Tübingen PL Center.

Make sure to check out their great talks on Repeatedly Handling Rejection, Teaching Effect Handlers, and Graph Rewriting with Handlers!

[@[email protected] — Jiří Beneš @jiribenes_]

@HerrDreyer For errors, I don't think so, no. But it did for situations like not knowing at all what the definition of a graph is in a graph theory lecture in the third hour of the exam...

[@[email protected] — Jiří Beneš @jiribenes_]

@HerrDreyer But I agree that it's quite hard to gauge this in general. After years in Germany, I still have no intuitive idea how should a 1.3 performance be different from a 1.7 performance — I can really only judge them relatively to each other.

[@[email protected] — Jiří Beneš @jiribenes_]

@HerrDreyer The system of grading was quite helpful there as the scale was just 1, 2, 3, and Fail. So 1 was "can fix their own mistakes", 2 was "needs a little help", 3 was "needs a lot of help", and Fail was "needs too much help". (The joke was that it's the ratio of red vs black ink)

[@[email protected] — Jiří Beneš @jiribenes_]

@HerrDreyer That part was still iterative: the examiner points out an error and asks whether the student wants some time to think about it in peace.
I liked that part a lot both as a student (as I'm not quick-witted) and as an examiner (as I find it very awkward to just sit there and stare).

[@[email protected] — Jiří Beneš @jiribenes_]

@HerrDreyer But this style of exams was deemed to be fair by the students and insightful by the faculty thanks to the generous time limits and the iterative nature: the grade was in essence determined by how much the examiner had to help out in order to get the correct solution/defn/proof.

[@[email protected] — Jiří Beneš @jiribenes_]

@HerrDreyer The main disadvantages were mostly felt by the examiners: this required holding like 6 exams, each one taking half a day, often with multiple examiners (like TAs) at the same time, totalling up to many more person-hours than just making and grading a purely written exam.

[Felipe O. Carvalho @_Felipe]

@geofflangdale This recent OOPSLA paper shows a very elegant way of dealing with overloads and rendering nice error messages when there’s more than one possible call. The online demo shows you how it looks.

[Jonathan Brachthäuser @__protected]

The sixth and last talk of my group and collaborators will be by @jiribenes_ on the Simple Essence of Overload Resolution.

It'll be at 11:45 local time in Orchid East, so make sure to drop by.

[Jonathan Brachthäuser @__protected]

Extremely proud of Marvin Borner who won 1st price in the undergrad ICFP Student Research Competition.

He presented his work of his Bachelor thesis on bringing effects to optimal reduction via token passing.

Impressive work, well deserved!

[Jonathan Brachthäuser @__protected]

"The Simple Essence of Overloading"

In this paper we show that type-directed overload resolution can be made compatible with Hindley-Milner-like type inference (TBH it is algebraic subtyping) and doesn't require backtracking search; it turns out it can be extremely fast as well.

[Jonathan Brachthäuser @__protected]

I am more than proud of my PhD students, colleagues and collaborators!

Together we managed to have 5 papers at this year's OOPSLA'25 (R1 and R2) and 1 at the colocated ICFP'25.

🥳🎉

The papers cover a wide range of PL and compiler topics 🧵

[@[email protected] — Jiří Beneš @jiribenes_]

@dogecahedron @aramh @__protected There's both a JS backend and a LLVM backend (and a third, barely secret, Chez backend).

[@[email protected] — Jiří Beneš @jiribenes_]

@dogecahedron @aramh I'm a bit biased, but I'd recommend playing around with Effekt by @__protected et al. It has a nice online tutorial ("Tour") and it can be used from the browser without having to install stuff locally.
https://t.co/6cmvzEfP7E

[Aram Hăvărneanu @aramh]

My favorite paper of the year (and despite the title it's really about linear logic).

[Jonathan Brachthäuser @__protected]

@mattpocockuk If you are interested, here is a translation of an example from Effect TS to Effekt. This might make it a bit easier to see how the two relate.

https://t.co/PuznAcAHgN

Thanks @jiribenes_ !

[Jonathan Brachthäuser @__protected]

What a great Ξ Effekt Hackathon!

We merged 21 PRs and closed 24 issues!

Thanks to everyone who attended and contributed. See you in half a year :)

(some PRs and issues are in satellite repos)

[Jonathan Brachthäuser @__protected]

We are very happy to host "Effekt: Lexical Effect Handlers in Action" and are looking forward to welcoming many of you there!