Bilingual

Counterexamples in type systems (2021)
a month ago
- The text lists 31 counterexamples in type systems, covering issues like polymorphic references, covariant containers, variance checking, and recursion.
- It includes topics such as object construction, Curry's paradox, runtime type misinformation, overloading, and subtyping vs. inheritance.
- Other points address selfishness, privacy violations, unstable expressions, avoidance problems, scope escape, and universal quantification.
- The collation is by Stephen Dolan, with acknowledgments to Andrej Bauer, Leo White, and Jeremy Yallop, and features a visual theme toggle.
Local Reasoning for Global Properties
25 days ago
- The author previously believed AI would not benefit from new programming languages, as AI can generate code in many existing languages.
- However, AI often generates good local code (e.g., functions) but struggles with global program understanding, leading to unnecessary defensive checks that can cause exponential state complexity.
- If AI's global reasoning weakness persists, programming language design may help by enforcing global properties through local reasoning, similar to Rust's data race prevention.
- Rust's ownership types and Send/Sync traits prevent data races statically, allowing local reasoning to ensure global data-race freedom without a new sublanguage.
- This demonstrates that programming languages can enforce surprising global properties (like data race freedom) with minimal burden on programmers.
- AI-generated code often includes excessive defensive checks, which are unnecessary, hurt performance, and mislead about program states, causing exponential state explosions.
- Future programming languages might incorporate features like effects systems to enforce more global properties, potentially benefiting AI code generation and human review.
- While not all global properties can be enforced, there is evidence that usable languages can provide more guarantees, changing incentives for language design in the AI era.
Odin, Wikipedia and Engagement Farming
22 days ago
- The Wikipedia article on the Odin programming language was deleted due to lack of notability and reliable sources, sparking controversy.
- GingerBill, Odin's creator, initially dismissed Wikipedia as an ideological playground gatekept by activists, but later expressed concern about its guidelines for programming languages.
- Casey Muratori argued that Wikipedia's sourcing rules are unfair to modern programming knowledge, which often lacks traditional academic coverage.
- Wikipedia's policies emphasize verifiability, reliable sources, and notability, with guidelines designed to maintain objectivity and avoid indiscriminate inclusion.
- GingerBill's contradictory behavior—claiming indifference while engaging extensively—suggests engagement farming and alignment with right-wing provocateurs he follows on Twitter.
- Analysis of Wikipedia editors involved in the deletion shows no evidence of ideological bias against Odin; they appear focused on general content moderation.
- The author criticizes GingerBill and Casey's clique for fostering a toxic, reactionary culture that prioritizes ego and spite over constructive discourse and software quality.
Returning to Zig
21 days ago
- The author was initially drawn to Zig in 2020 as a promising replacement for C, appreciating its compile-time code execution and simple abstractions.
- They left Zig due to instability, frequent breaking changes, and an immature ecosystem, switching to Rust for its stability and strong tooling.
- Concerns about Rust's governance, including corporate influence and slow response to LLM-generated code policies, led the author to reconsider Zig.
- In 2026, Zig remains unstable but familiar, with added features like a package manager, though its ecosystem is still limited compared to others.
- Zig prioritizes simplicity over Rust-like memory safety, using release modes and testing to ensure safety, which the author finds acceptable for certain use cases.
Rust is in TIOBE Index top 10
19 days ago
- Rust enters the TIOBE Index top 10 for the first time in July 2026.
- The TIOBE Index celebrates its 25th anniversary in July 2026.
- Python ranks first in the TIOBE Index for July 2026, followed by C and C++.
- TIOBE introduces a Programming Language Flowchart to help developers select languages for specific use cases.
- The TIOBE Index measures language popularity based on skilled engineers, courses, vendors, and web search data.
- Long-term trends show C, C++, and Java have remained top languages over 25 years.
- Feedback led to improvements in the index, including handling of dialects and addition of SQL.
- The index includes data on the top 50 languages and a hall of fame for annual award winners.
- Suggestions for index enhancements include adding non-English queries and tracking recently added pages.
- Complete historical data is available for a fee, covering over 150 languages from 2001 onward.
Scarf has moved away from Haskell
15 days ago
- Scarf moved from Haskell to Python due to AI changing development economics, making compile times and ecosystem friction significant bottlenecks.
- AI allows faster code generation, but Haskell's long compilation cycles hinder rapid feedback, especially in parallel AI-assisted workflows.
- The shift enabled faster development cycles, improved test coverage, and quicker hotfixes, with no noticeable loss from reduced type safety.
- Haskell faces stagnation risk if it doesn't adapt to AI by optimizing for fast builds, documentation, and agent-friendly tooling.
G#: A modern .NET language with Go, Kotlin, and Swift ergonomics
14 days ago
- G# is a modern .NET language that incorporates ergonomic features inspired by Go, Kotlin, and Swift.
- It includes features such as packages, functions, data classes, nullable handling with 'if let', and structured concurrency with scope.
- Source code compiles directly to managed assemblies, allowing execution on the .NET runtime.
- An example provided demonstrates a simple package with a greeting function that prints "Hello, world!".
Python Is So Slow. Can Julia Solve the Two-Language Problem?
11 days ago
- Award acceptance lectures by Turing Award-winning computer scientists often stand out as influential exceptions to the banality of the genre.
- Examples include John Backus's 1977 lecture inspiring functional programming, Ken Thompson's 1984 warning about compiler backdoors, and Edsger Dijkstra's 1972 call for humility.
- Kenneth Iverson's 1979 lecture 'Notation as a Tool of Thought' argued that mathematical notation fosters discovery, exemplified by APL, a programming language designed to fuse mathematical and programming languages.
- The two-language problem in scientific computing today involves prototyping in slow Python and rewriting performance-critical parts in faster languages like C++ or Rust.
- Julia, created in 2012, aims to be a single language as ergonomic as Python and as fast as C, addressing this problem.
- Julia has gained a sober, academic community and offers significant speedups (e.g., 10x to 1000x faster than Python) but remains niche due to Python's robust ecosystem and lack of Big Tech adoption.
- Julia is used in high-performance fields like drug discovery and machine learning at institutions like ASML, CERN, and NASA.
- The two-language problem persists across domains (e.g., gaming with C++/Lua, backends with Python/Go), and it's uncertain if any language can fully solve it.
- The piece is part of a series on AI-enabling languages, reflecting on the potential for future innovations akin to Iverson's insights.
SendLang: A DSL for Email Automation
10 days ago
- SendLang comprises SendQL for segment-definition queries and SendFlow for drip-workflow management.
- Languages enable version control, pre-send error checking, and compatibility with coding agents.
- SendQL integrates raw event streams, contact attributes, and deliverability rules for precise targeting.
- SendFlow offers structured control flow without goto or unbounded loops, mapping to flowcharts.
- Coding agents can write and iterate on campaigns using machine-readable docs and type checkers.
- Design focuses on a golden-pinned grammar, precise diagnostics, and lossless text-canvas conversion.
A Road to Lisp: Which Lisp
8 days ago
- Lisp is a family of programming languages with many dialects, unlike single-language evolutions like Python or Java.
- Beginners should focus on learning Lisp's programming concepts rather than worrying about which dialect to choose, as switching later is relatively easy.
- Common Lisp (CL) is mature, standardized since 1994, with multiple implementations like SBCL, offering high performance, a powerful REPL, and comprehensive features including CLOS for OOP.
- Clojure is a modern Lisp on the JVM, emphasizing functional programming with immutable data structures, and is widely used in production by companies like Nubank and Netflix.
- Racket is a language-oriented dialect ideal for creating new languages, with features like hygienic macros and Typed Racket, and comes with the DrRacket IDE.
- Elisp is specialized for customizing Emacs and extends editor functionality in real time.
- Choosing a dialect depends on needs: Clojure for modern syntax and JVM integration, Common Lisp for native performance and interactivity, Racket for language design and GUIs, and Elisp for Emacs customization.
Why care about programming languages
2 days ago
- Programming languages are valuable for the ideas they embody, not just syntax or tooling.
- Concepts like Rust's borrow checker, Haskell's typeclasses, and SQL's relational theory expand your mental toolkit for problem-solving.
- Understanding these ideas helps evaluate AI-generated code, choose appropriate abstractions, and design maintainable systems.
- In the age of AI, deep knowledge of programming concepts distinguishes true engineers from those who only prompt.
Programming language file extensions that match ISO 3166-1 alpha-2 country code
a day ago
- ISO 3166-1 alpha-2 provides two-letter codes for countries (e.g., ME for Montenegro) and non-country entities (e.g., EU for European Union).
- These codes appear in top-level domains and emoji flags (e.g., :flag_fm: for Micronesia).
- Many programming languages share their file extensions with country codes (e.g., .rs for Rust and Serbia).
- The article lists examples like AM (Armenia/Automake), ES (Spain/ECMAScript), PY (Paraguay/Python), and RS (Serbia/Rust).
- The author invites readers to suggest additional matches not included in the list.
Online Historical Encyclopaedia of Programming Languages
a day ago
- The online roster lists 8,945 programming languages from the 18th century to the present.
- It includes 7,800 influence links and over 11,000 citations.
- The genealogical tree of HOPL is maintained by librarians Diarmuid J Pigott and Bruce M Axtens, with quartermaster Steve Poulson.
- The content is copyrighted by Diarmuid Pigott from 1995-2020, with a prohibition on copying or reproduction.
- Errata and crash notifications can be sent to hopl.acquisitions@gmail.com.

programming languages

Counterexamples in type systems (2021)

Local Reasoning for Global Properties

Odin, Wikipedia and Engagement Farming

Returning to Zig

Rust is in TIOBE Index top 10

Scarf has moved away from Haskell

G#: A modern .NET language with Go, Kotlin, and Swift ergonomics

Python Is So Slow. Can Julia Solve the Two-Language Problem?

SendLang: A DSL for Email Automation

A Road to Lisp: Which Lisp

Why care about programming languages

Programming language file extensions that match ISO 3166-1 alpha-2 country code

Online Historical Encyclopaedia of Programming Languages