Try Gleam

Gleam, a statically typed functional language targeting the BEAM VM, is drawing interest for its pleasant developer experience, strong type system, and ability to share safe code between backend and frontend—even when much of the code is generated by AI tools. Commenters weigh its design trade-offs, such as the lack of automatic currying, relatively small ecosystem, and interop story with Elixir, against benefits like better guarantees in distributed systems and good tooling for a young language. The project’s explicit progressive code of conduct and political messaging also emerge as a point of contention, with some appreciating the clear stance on inclusivity and others preferring to keep ideology separate from programming tools.

Role of languages in an AI-heavy world

  • Some argue that if LLMs generate most code, languages should be optimized for machines, not humans.
  • Many push back: humans still read and review code for safety, performance, compliance, and responsibility.
  • Languages with strong type systems and safe abstractions (Gleam, Haskell-like, Rust) can eliminate bug classes and act as guardrails for LLM-generated code.
  • Several note that LLMs struggle with parallel/distributed systems; runtimes like BEAM still matter, regardless of who writes the code.
  • “Languages for LLMs” are seen by some as premature; better to make models work well with clear, verifiable, human-readable code.

Determinism vs LLM behavior

  • Compilation and optimization (including PGO, JIT, LLVM) are discussed as largely deterministic in semantics, even if machine code varies.
  • This is contrasted with LLMs, whose outputs can change meaningfully and lack the same guarantees or specs.

Gleam’s design, currying, and ergonomics

  • Gleam is praised as delightful, typed, and good on BEAM, with a pleasant coding “craft” experience.
  • A major recurring complaint: no automatic currying. Some see this as crippling, especially for pipe-heavy FP code.
  • Others dislike currying (harder to read, trickier type errors, awkward argument order) and prefer Gleam’s explicit partial application using a single placeholder (_).
  • Single-hole partial application is seen by some as a practical compromise; more complex cases can use explicit anonymous functions.

Ecosystem, tooling, and full‑stack use

  • Tooling (LSP, general DX) is described as surprisingly good for a small project, though some rough edges exist (e.g., running a single test).
  • One user reports success with a full-stack Gleam app (frontend + backend), sharing types and relying on an LLM to generate most code; boilerplate is acceptable because the AI writes it.
  • Concerns are raised about limited ecosystem and especially about Elixir/BEAM interop.
  • Some find it problematic that seamless, first-class interop with Elixir libraries wasn’t a core design goal; others point to existing interop guides and tooling but the actual smoothness is left somewhat unclear.

LLMs, language popularity, and training data

  • One camp argues popular languages benefit from more training data, leading to better LLM output.
  • Others counter that modern models generalize across languages, can consult docs, and that popularity/network effects matter less than before.
  • Cited AI studies suggest mainstream languages yield somewhat better correctness, but differences are small enough that many recommend choosing the best tool/ecosystem for the job.

Community, code of conduct, and “activism”

  • A subset of commenters are put off by what they see as visible activism: explicit political/value statements on the homepage and a strong code of conduct.
  • Critics argue they don’t want ideology mixed with a programming language, and some worry (or claim) that the CoC effectively polices people’s beliefs or behavior outside project spaces.
  • Others respond that the CoC is standard anti-harassment language, mainly about behavior inside the community, and that explicit “no bigotry” stances help keep culture-war drama out.
  • Several note that you can freely use the language without engaging in the community and that, in practice, they rarely encounter politics while actually writing Gleam code.
  • Some readers appreciate early warnings about communities whose values or tone they might dislike; others see the controversy as overblown.