Common Lisp
lisp-lang.org sells Common Lisp on one property: extensibility. The language absorbs new abstractions instead of waiting for a revision to add them, and the site quotes paul-graham to the effect that a language like that does not go out of style. Everything else on the page is evidence for that claim — a macro-defined SQL DSL (SxQL), first-class functions and composition, class hierarchies through CLOS (the Common Lisp Object System), and interactive development through the REPL and the SLIME environment, which the site frames as working without a compile-and-debug cycle at all. Web server benchmarks (Woo) carry the performance argument. Designed by Fernando Borretti; copyright 2015–2025.
T3 — a community site about a language, not a first-party one, which is the same grade c‘s c-language.org gets and for the same reason: no standards body or vendor stands behind it. It is also the corpus’s only source with a visible end date on its copyright line, which matters for a page whose benchmarks and library links are its evidence.
Extensibility as the axis
The four axes on programming-language ask what a language decided about memory, concurrency, runtime and governance. This page answers none of them and argues a fifth: who gets to add to the language. Macros make the answer “the person writing the program”, so the distance between a library and a language feature collapses — SxQL is presented as SQL syntax, not as a function that builds strings.
clojure is the same wager one generation later, and the pair is useful. Clojure took code-as-data and macros and then removed choices around them: immutable data by default, STM for coordinated change, one hosted runtime family. Common Lisp adds instead — CLOS is an object system layered onto a language that did not start with one, which is the extensibility claim demonstrating itself. Both are Lisps; only one treats “the language can be extended” as the headline.
What the site does and does not settle
The page names the ANSI Common Lisp specification and presents the standard as a stability feature, a “rock-solid foundation” — the opposite framing to c‘s, where the same committee arrangement is described through its revision cadence (C99, C11, C17, C23). It does not date the specification, name its revisions, or say who could revise it, so language-governance gains a fifth arrangement in outline only: a standard held still on purpose. Allegro CL is mentioned; the site does not lay out the implementation landscape, which for a language defined by a document is exactly where the interesting differences live.
Typing, memory and concurrency go unmentioned. On type-system‘s scale it sits with clojure — dynamic, checked as it runs — but that comes from the corpus, not from this page.
The success stories, and what they are worth
Three named users: Grammarly (a grammar engine handling “more than a thousand sentences per second”), ITA Software, acquired by Google, for the Matrix flight search engine, and SISCOG for European rail crew scheduling. These are the corpus’s first industrial deployment claims of any kind, and they are unusually specific about domain — search, scheduling, natural language, all algorithm-heavy back ends rather than the web and mobile front ends the other sources target. Undated and uncited, so treat them as the site’s argument rather than as evidence.
Related
programming-language · clojure · paul-graham · language-governance · type-system · design-patterns · c · synthesis