Chapter 2. Eclipse Temurin features
Eclipse Temurin does not contain structural changes from the upstream distribution of OpenJDK.
For the list of changes and security fixes that the latest OpenJDK 17 release of Eclipse Temurin includes, see OpenJDK 17.0.14 Released.
New features and enhancements
Eclipse Temurin 17.0.14 includes the following new features and enhancements.
Fix for invokedynamic string concatenation changing order of operations
The Indify String Concatenation feature that was added in OpenJDK 9 through JEP-280 introduced a regression in the order in which string concatenation expressions are evaluated. The Java Language Specification requires that operands are fully evaluated in left-to-right order. However, with the introduction of invokedynamic (indy) call generation by the javac compiler for evaluating string concatenation expressions, all operands were evaluated in order but not converted to strings. In this situation, each operand was converted to a string only later.
For example, consider the following code, where the third argument of the concatenation has the side effect of altering the value of builder to "goodbye":
StringBuilder builder = new StringBuilder("good");
return "" + builder + builder.append("bye");
Based on the preceding example, in earlier releases, the Indify String Contenation feature did not convert the second argument to a string until after the builder.append method altered the StringBuilder object. In this situation, the concatenation incorrectly became "" + "goodbye" + "goodbye", which produced "goodbyegoodbye" as the output.
In OpenJDK 17.0.14, string concatenation evaluates and eagerly converts each argument to a string in left-to-right order. In this situation, the concatenation correctly becomes "" + "good" + "goodbye", which produces "goodgoodbye" as the output.
The resolution of this issue has the same effect as using a version of the javac compiler that was available before OpenJDK 9 or running javac with the -XDstringConcat=inline command-line option.
See JDK-8273914 (JDK Bug System).
Option for jar command to avoid overwriting files when extracting an archive
In earlier OpenJDK releases, when the jar tool extracted files from an archive, the jar tool overwrote any existing files with the same name in the target directory.
OpenJDK 17.0.14 adds a new ‑k (or ‑‑keep-old-files) option that you can use to ensure that the jar tool does not overwrite existing files. You can specify this new option in either short or long format.
For example:
-
jar xkf myfile.jar -
jar --extract ‑‑keep-old-files ‑‑file myfile.jar
In OpenJDK 17.0.14, the jar tool retains the old behavior by default. If you do not explicitly specify the ‑k (or ‑‑keep-old-files) option, the jar tool automatically overwrites any existing files with the same name.
See JDK-8335912 (JDK Bug System) and JDK bug system reference ID: JDK-8337499.
IANA time zone database updated to version 2024b
In OpenJDK 17.0.14, the in-tree copy of the Internet Assigned Numbers Authority (IANA) time zone database is updated to version 2024b. This update is primarily concerned with improving historical data for Mexico, Mongolia, and Portugal.
This update to the IANA database also includes the following changes:
-
Asia/Choibalsanis an alias forAsia/Ulaanbaatar. - The Middle European Time (MET) time zone is equal to Central European Time (CET).
Some legacy time-zone IDs are mapped to geographical names rather than fixed offsets:
-
Eastern Standard Time (EST) is mapped to
America/Panamarather than-5:00. -
Mountain Standard Time (MST) is mapped to
America/Phoenixrather than-7:00. -
Hawaii Standard Time (HST) is mapped to
Pacific/Honolulurather than-10:00.
OpenJDK overrides the change in the legacy time-zone ID mappings by retaining the existing fixed-offset mapping.
-
Eastern Standard Time (EST) is mapped to
See JDK-8339637 (JDK Bug System).
Revised on 2025-01-30 11:17:24 UTC