Oliver Flatt

@0xalpharush @gakonst @VaivaswathaN Just logged in for the first time in a while, sounds cool but still busy with the PhD!

@gakonst Not sure if there’s interest but @oflatt did some interesting work on egraphs with Certora and has an interesting compiler project github.com/egraphs-good/e… Also @VaivaswathaN has done great work on Sway and has a cool project similar to MLIR

I am really glad we built Foundry. Would have been really awkward to be an EVM developer in 2025 without it. Now we really need to fix the Solidity compiler. $10K for successful referral on a world class compiler engineer to redefine the future of Ethereum development.

My friends and I made an amazing new game! youtube.com/watch?v=jTA5Wm…

The "winner stays" rule in sports is bad and here's why. oflatt.com/winner-stays.h… Graphs included!

Tune in tomorrow and ask questions at 10pm PT for the second episode of theforkjoin.com! Super excited to host @wise_jenna, who has been working on *Gradual Verification*: transforming dynamic checking into static, compile-time verification in a smooth way.

TheForkJoin is back for season 2!!! Join me and @notypes as we chat with new faculty in programming languages in systems, taking questions from the audience and generally having a great time. First up: @tonofcrates *Sep 7 at 10am PT*! Schedule: theforkjoin.com

Our paper “SpEQ: Translation of Sparse Codes using Equivalences” accepted to PLDI 2024 @PLDI. SpEQ uses equality saturation and equivalence checking to automatically translate input codes to library calls or domain-specific languages (DSLs). Paper: paramathic.com/wp-content/upl…

Join us on Thursday at 9am PT for the first EGRAPHS Community Meeting! egraphs.org/meeting/ For the first meeting, I'll be giving an overview of e-graphs, recent developments, and exciting open directions. Don't know much about e-graphs? This is a great starting point!

@hackedy @arntzenius Martin's work is super cool! It was done completely independently, but we came to quite similar ideas. Gives us confidence the ideas are good

DATALOG PAPER ALERT "Algebraic Semantics of Datalog with Equality" Martin Bidlingmaier shows how to give categorical semantics to Datalog + equality in conclusions, eg: X = Y :- edge(X,Y), edge(Y,X).

New blog post Follow the instructions to make cool dragon curves out of paper or code! uwplse.org/2024/01/22/Dra…

🌟 We're Hiring at the Lean FRO! 🌟 Interested in Lean, functional programming languages, and theorem provers? We have open positions that might be just right for you. For more details and to apply, please visit: lean-fro.org/jobs/

@FreyaHolmer Agreed

out of all the invasive privacy violations on the internet - the worst feeling one is youtube adding the videos you mouseover-previewed to your watch history

@FreyaHolmer Yep do this one

violently clickbait: "RECREating blENder IN just 3 months!!! 🤯😲" not that I'm making something even close to the scope of blender but that sure sounds like one of Those™ youtube videos right, isn't that how you're supposed to make money on youtube or something

what's a good title if I make a youtube video about my 3D modeling tool? standard route: "Making a 3D modeling tool in Unity" or, we can go the clickbait route: "i got covid and so I made a 3D editor" "i didn't want to learn blender so I made my own" "unity does WHAT??" [arrow]

@ben_kushigian Pls tell

@alpha_convert To the point where I have installed a command line tool that removes backgrounds rmbeg

@alpha_convert Me every single day it feels like

thanks me

@alpha_convert I want the following from egglog: given that I trust the rules I've written, how did I get a particular term or equality using them? It's a generalization of proof-producing congruence closure

egglog has its first release! crates.io/crates/egglog Egglog is a language that combines the power of egraphs and datalog, and it's ready to use! It's the culmination of a long line of egraph research across institutions and industry. I'm super excited to release it.

Moppet, the dog Tessa (my partner) walks is in trouble! They need a temporary home and it's not cheap. We started a GoFundMe, and anything helps! More info there gofund.me/b3447b59

@TaliaRinger Sounds like you are doing a great job as an advisor

Just sent a bunch of students an email that ended with "have a good weekend!" It's Tuesday. So anyways how's your day going

@yihongz_bot @xwangsd Don't worry Yihong, more is more

@xwangsd Oh I meant people think it’s powerful because it can represent infinite program spaces, but as it turns out it’s less powerful than people think since it can’t even represent some finite program spaces.

TIL e-graphs are surprisingly “weak”: an E-class representing program a and f(a) must also represent the entire infinite language f^*(a). One pitch for why E-graphs are so powerful was that it can represent infinite program spaces, but in fact going infinite is never a choice.