On 16/02/2015 16:41, Rainer Jung wrote: > Am 16.02.2015 um 16:09 schrieb Mark Thomas: >> On 16/02/2015 14:49, Bill Barker wrote: >> >> <snip/> >> >>> test: >>> [concat] Testsuites with skipped tests: >>> [concat] TEST-org.apache.catalina.connector.TestRequest.NIO2.txt >>> [concat] >>> TEST-org.apache.tomcat.util.net.jsse.openssl.TestOpenSSLCipherConfigurationParser.NIO2.txt >>> >>> [concat] Testsuites with failed tests: >>> [concat] >>> TEST-org.apache.catalina.loader.TestWebappClassLoaderThreadLocalMemoryLeak.NIO2.txt >>> >> >> This is just plain strange. It is doing it consistently at least but >> only for Gump not for buildbot and not when I run the tests locally on >> Linux. I wonder if it is related to how Gump sets up the class path. >> I'll take a look. > > One of the failing openssl tests earlier today also contained a log line > about IBM JVM. Haven't checked though, what JVM Gump is using.
The tests keep a record of all the known Ciphers supported by IBM and Oracle JVMs and a list of ciphers implemented by OpenSSL that are known not to supported. The idea is when OpenSSL adds support for a new cipher either the name matches one of the known supported or unsupported ciphers and if not, the test fails so we realise we have to explicitly add the mapping or known unsupported entry. In this case it was a new cipher not supported by IBM. I believe I have fixed the OpenSSL Gump failures for trunk (OpenSSL 1.1.0) and 8.0.x (OpenSSL 1.0.2). There are still other failures to look at but it is getting late here so I'll leave that for another day. Mark --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org