Hi, it took some time to find out, why sometimes Lucene/Solr's tests failed with a very strange error message when trying to fork a process on MacOSX and Linux. The reason for this bug to appear is our randomized testing framework: As you know, we execute all tests in different settings like default charsets, default locales or default timezones, to find bugs caused by settings applied to the environment, that may affect the code in an incorrect way. For more information, see this (already famous) blog entry: http://blog.thetaphi.de/2012/07/default-locales-default-charsets-and.html
In our case the JDK issue was: https://bugs.openjdk.java.net/browse/JDK-8047340 The issue on our side: https://issues.apache.org/jira/browse/SOLR-6387 We just want to add, that this bug also applies to Java 7 - the same issue exists there (and possibly also on Java 6). So the fix (Locale.ENGLISH / Locale.ROOT) should be backported to Java 7 and Java 6 for customers with support contract, too. As I cannot modify the bug report, it would be good to add issues for Java 7 and Java 6, too. Uwe PS: We use the Forbidden-APIs checker (https://code.google.com/p/forbidden-apis/) to detect such bugs in Apache Lucene/Solr, Elasticsearch, Apache TIKA, and many other projects I don't know of yet. Maybe when building the rt.jar, you should do the same :-) Calling String#toLowerCase() without an explicit Locale is a bug, if you rely on locale-insensitive behavior. ----- Uwe Schindler H.-H.-Meier-Allee 63, D-28213 Bremen http://www.thetaphi.de eMail: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
