Java

java.com is a download page, not a language page. It splits its audience between end users who need a runtime to open a desktop application and developers who want the JDK, calls Java "the #1 programming language and development platform", says it "reduces costs, shortens development timeframes, drives innovation, and improves application services", and names Oracle as "the stewards of Java" (Oracle as a cloud vendor lives in ../cloud-wiki). It links OpenJDK early-access builds. No version number, no adoption figure, and not one sentence about the language.

T1 — first-party, and the emptiest source in the corpus. Recorded that way on purpose.

What the emptiness shows

Every other home page here argues a design position: go removes choices, rust proves memory safety, haskell commits to purity, scheme states a rule about what not to add. Java's public front door argues procurement — cost, timelines, enterprise fit — and routes anyone with a technical question elsewhere. That is what a language's marketing looks like when the language no longer has to be sold to programmers, and it is the cleanest evidence yet for the caveat this spoke repeats: a front page tells you who the language is being sold to, not what it is.

The technical material exists at dev.java, the JDK documentation and the JEP process. None of it is here, so the corpus still has no source describing Java's design or its evolution process, which is awkward given how much of the wiki depends on it.

The dependency the corpus already has on it

Java's absence is load-bearing. clojure is hosted on the JVM, kotlin targets it with Java interop as its founding feature, jank exists because someone wanted Clojure's semantics without Java interop, and rethinking-java-design-patterns rewrites the GoF catalog in functional Java. Four pages describe things relative to Java; none describes Java. See hosted-language — the host whose properties the guests keep inheriting is the one this wiki knows least about.

Growth edge, promoted to synthesis: the JEP process and a JDK release page would close both gaps at once — language-governance gains its best-documented corporate-plus-open-process case, and the JVM stops being an unexamined assumption.

The first source in which Java's own world moves (2026-09-07)

InfoQ's Java roundup for the week of 31 August 2026 does not close the gap this page describes — it contains no design argument and no JEP — but it is the first source here in which Java is the subject rather than the thing other languages are defined against.

What it establishes is cadence: JDK 27 sat at early-access build 35 while JDK 28 shipped early-access build 14 the same week, so two versions are in early access at once, both taking bug reports through the Java Bug Database. The column names no JEPs and no dates, so this is the shape of the release process and not its content.

One item does reach the runtime. TornadoVM 6.0.0 removed its dependency on JVMCI, the experimental JVM Compiler Interface, and its release notes give the reason InfoQ omits: JDK 27 compatibility. In the same release it replaced JNI with the Foreign Function & Memory API. A downstream project rewriting both of its interfaces to the JVM in one release, because the JVM moved, is the closest thing the corpus has to evidence about how the host evolves — see hosted-language, which has been carrying the JVM as its least-documented assumption since August.

The edge stands, narrowed: what is still missing is the JEP process read end to end and a JDK release page. A serial column is a poor substitute for either, but it is measurable if more weeks land.

Related

clojure · kotlin · jank · hosted-language · language-governance · rethinking-java-design-patterns · programming-language · synthesis