javademos/docs/01-what-shipped.md

1. What shipped, and what to believe

Next: 2. G1 becomes the default

Java 27 reached General Availability on 2026-09-15. It is not an LTS: Oracle's support roadmap lists Premier support for 27 until March 2027, and names Java 29 (September 2027) as the next LTS after 25. Everything in this repository was run on real JDK binaries; where a claim comes from prose rather than from a run, the chapter says so.

The nine JEPs in 27

JEP	Title	Status in 27	Chapter
523	Make G1 the Default Garbage Collector in All Environments	final	2
534	Compact Object Headers by Default	final	3
527	Post-Quantum Hybrid Key Exchange for TLS 1.3	final	4
532	Primitive Types in Patterns, instanceof, and switch	fifth preview	5
531	Lazy Constants	third preview	6
533	Structured Concurrency	seventh preview	7
538	PEM Encodings of Cryptographic Objects	third preview	8
536	JFR: In-Process Data Redaction	final	9
537	Vector API	twelfth incubator	9

The ten JEPs in 26 (the release many teams skipped)

JEP	Title	Status in 26	Chapter
500	Prepare to Make Final Mean Final	final	11
504	Remove the Applet API	final	11
516	Ahead-of-Time Object Caching with Any GC	final	11
517	HTTP/3 for the HTTP Client API	final	11
522	G1 GC: Improve Throughput by Reducing Synchronization	final	11 (not measured here)
524	PEM Encodings of Cryptographic Objects	second preview	8
525	Structured Concurrency	sixth preview	7
526	Lazy Constants	second preview	6
529	Vector API	eleventh incubator	9
530	Primitive Types in Patterns, instanceof, and switch	fourth preview	5

How the list was verified

openjdk.org/jeps and jdk.java.net/27 returned HTTP 403 from the sandbox used to write this, and no workaround was attempted. The lists were cross-checked between the Inside.java release announcement, Oracle's release blog, InfoWorld's and foojay's write-ups, and then against the binaries: the JDK 27 and 26 builds used here were diffed class by class (51, 52), which is how two API names in a secondary write-up were corrected. Treat the JEP titles above as "as reported by four secondary sources"; treat every behaviour as "observed on Corretto 26.0.2.1 / 27+35".

What was not reproduced

• JDK-8377013 (a reported TimeZone.getDefault() regression) could not be reproduced and is left out.
• HTTP/3 was never spoken to a real HTTP/3 server: the sandbox has no UDP egress (11).
• JEP 522 (G1 synchronization) was not benchmarked.

Next: 2. G1 becomes the default