> jdk-9+175 was supposed to be the first GA candidate so this is > why the pre-release identifier "ea" was dropped. Hopefully the > Hadoop maintainers will read JEP 223 to see how to parse version > strings going forward.
I noticed that at least two popular Maven plug-ins are affected, too: the Maven JavaDoc plug-in and the Enforcer plug-in. They both use rather old versions of Apache Commons Lang (2.x) whose utility method SystemUtils#getJavaVersionAsFloat() runs into a StringIndexOutOfBoundsException due to the changed version format as of b175. Perhaps some Apache committers are present here who could take a look at this one? 2017-06-26 10:00 GMT+02:00 Alan Bateman <alan.bate...@oracle.com>: > On 26/06/2017 08:35, Uwe Schindler wrote: > >> : >> >> So I just wanted to give you some feedback that translations are >> obviously missing. FYI, the Linux version in English returns the correct >> text. >> > The last translation drop went into jdk-9+172 so it's out of sync with the > CLI/usage messages in builds since then. In addition to --illegal-access, > there are a few other changes to, and at least one new, error message(s) > that also need updated translations. > > >> FYI, the new version string (just "9") again caused some serious problem >> in open source projects. This time Hadoop, see >> https://issues.apache.org/jira/browse/HADOOP-14586. Issue is the >> following static final constant, leading to NoClassDefFound error and a >> cascade of classes no longer initializing: >> private static boolean IS_JAVA7_OR_ABOVE = >> System.getProperty("java.version").substring(0, >> 3).compareTo("1.7") >= 0; >> >> Damn! [image: 😊] (the above code is IMHO just horrible wrong, never >> ever do something like this). In a ddition, it no even has a doPrivileged! >> >> jdk-9+175 was supposed to be the first GA candidate so this is why the > pre-release identifier "ea" was dropped. Hopefully the Hadoop maintainers > will read JEP 223 to see how to parse version strings going forward. > > -Alan. >
