BFPG Meetup - September 2025 - Hole-driven development + Type-safe Gleam
Join the BFPG Discord: https://discord.gg/yYz2d8w7FY
Agenda
18:00: Welcome and setup
Presentation #1: Typed Holes and Hole-driven Development - George Wilson
Presentation #2: Gleam's Type-Safe OTP Concurrency: A New Era of Reliable Distributed Programming - Cameron Badman
20:00ish: Pack down, head to Criterion pub.
Typed Holes and Hole-driven Development
This talk will introduce typed holes in Haskell with examples. We will see how typed holes can help the programmer get assistance from their compiler. We will also discuss which aspects of Haskell make it well suited to hosting typed holes, and why this convenient feature is uncommon in other languages.
Gleam's Type-Safe OTP Concurrency: A New Era of Reliable Distributed Programming
Airbnb found that 38% of their bugs could have been prevented by type safety.
Message type mismatches. State corruption. Supervision tree configuration errors. Race conditions from uncoordinated process interactions. These preventable runtime failures plague even the most carefully written Erlang and Elixir systems. Your GenServer crashes at 3am because someone passed the wrong message format. Your supervisor fails to start because of a typo in the child specification. Your processes deadlock because message ordering wasn't enforced.
What if entire categories of OTP runtime bugs could be eliminated at compile time?
What if the compiler caught these bugs before they ever reached production? Gleam does exactly that, bringing Rust-level type safety to the battle-tested BEAM ecosystem. This isn't just theory—Gleam's HTTP servers now outperform Elixir (94% of Go's speed) while eliminating entire categories of production failures that have haunted OTP systems for decades.
This talk is a head-to-head comparison: Gleam vs Erlang vs Elixir across speed benchmarks, OTP patterns, and real-world reliability. You'll see identical concurrent systems implemented in all three languages, discover which runtime failures simply vanish with types, and learn when Gleam's compile-time guarantees are worth the trade-offs.
The BEAM's "let it crash" philosophy is brilliant—but why crash from bugs the compiler could have caught?
Target audience: Anyone building or maintaining concurrent systems on the BEAM.