David Richter

❤️ programming languages.
phd student at @stg_darmstadt.
DR are my initials not my title.

[STG Darmstadt @stg_darmstadt]

📄 A Direct-Style Effect Notation for Sequential and Parallel Programs by @drcicer, Timon Böhler, Pascal Weisenburger, and @m_mezini - https://t.co/VbiRM25nwE

[David Richter @drcicer]

@Soko2D @guidosalva @ECOOPconf @m_mezini @SebastFaust That's me!

[David Richter @drcicer]

@ProfMaxNew stuff, property, structure, thing, object, gadget, widget, container, category, ...

[David Richter @drcicer]

We also evaluate the performance of the compiler generated code to show that its similar to handwritten Solidity code. In contrast to other smart contract languages using linear, type-state or session types, Prisma is inspired by tierless and choreographic programming.

[David Richter @drcicer]

Prisma is a language / Scala library for writing distributed programs that interact via efficient and safe Ethereum smart contracts. @drcicer, Kretzler, Weisenburger, @guidosalva, @SebastFaust, @m_mezini #ecoop22
➡https://t.co/VmxEfPR0WT
➡️https://t.co/oIAq1gqfMi
Details below…

[David Richter @drcicer]

We formalise compilation&evaluation. Although compiler goes through heavy changes like CPS and defunctionalisation, we show that malicious client code 👿 cant distinguish contract code from before or after compilation (secure compilation), after inserting control-flow guards🛡️

[David Richter @drcicer]

@famontesi Cool!

[David Richter @drcicer]

@famontesi (Although its titled categorical logic, 95% of the pages contain only inference rules and no commutative diagrams 😅)

[David Richter @drcicer]

@famontesi The only book that contains inference rules and is called an introduction that i read recently is the following: https://t.co/aM38nHNXZP :D

[David Richter @drcicer]

@schrijvers_tom Ah! Initially I thought algebraic < scoped < latent, and the expected the I to be the effect. But I suppose the paper actually says to separate the latent part of the effect from the non-latent part. I probably need to reread it. The video is also nice😀

[David Richter @drcicer]

CPS translation gives every function another parameter (to be called with the result of the function) (\x.M := \x\k.M), and every application another argument (M N := M N K); and additionally all terms another argument to be called with the result of the term (... := \k.k ...).

[David Richter @drcicer]

Looking for nice equations. Today, continuation passing style translation: Via Griffin, POPL 1989. A formulae-as-type notion of control

[David Richter @drcicer]

Namely, the we must be able to pick a enclosing description (data) for any predicate (function), but we cant just try the function for all possible inputs to look for the best description (potentially infinite)! With a Kleisli GC, we return the element safely hidden in a monad.

[David Richter @drcicer]

Looks cool: Constructive Galois Connections, JFP'18, Darais, Horn. The problem is, how can we extract static analyses from mechanized soundness proofs for Galois Connections (GC)? The authors propose "Kleisli GC".

[R. Hill @gorhill]

Since Manifest V3's "performance" argument is back in the current news cycle, a reminder of what "Understanding the Performance Costs and Benefits of Privacy-focused Browser Extensions" concluded: https://t.co/uF4MBpwDGw

[Fabrizio Montesi ⬣ A➔B @famontesi]

The presentation of our @ECOOPconf distinguished paper on Multiparty Languages is online! 👀
https://t.co/MrFE7dYdfi

w/ @realThesave, @marcoperessotti, @drcicer, @guidosalva, and P. Weisenburger.

*Six* speakers in a 15 min 📽️!
Mini 🧵 on the paper in the 1st comment 👇.

[David Richter @drcicer]

@rob_rix Cool! But why not apply id to extract?

Oh, I see! Only if r is existential not universal. 😃
{-# LANGUAGE RankNTypes, TypeApplications #-}
type CPS a b = forall r. (b -> r) -> (a -> r)
pure :: (a -> b) -> CPS a b
pure f g = g . f
extract :: CPS a b -> (a -> b)
extract k = k id