On 04/02/2016 22:33, Rainer Jung wrote: > Am 02.02.2016 um 01:20 schrieb Mark Thomas: >> The proposed Apache Tomcat 9.0.0.M3 release is now available for voting. >> >> This is a milestone release for the 9.0.x branch. It should be >> noted that, as a milestone release: >> - Servlet 4.0 is not finalised >> - The EGs have not started work on JSP 2.4, EL 3.1 or WebSocket 1.2/2.0 >> >> The major changes compared to the 9.0.0.M1 branch are: >> - Ability to use OpenSSL with JSSE configuration >> - Update to Tomcat-native 1.2.4 to pick up Windows binaries based on >> OpenSSL 1.0.2e >> - Allow HTTP redirects to be relative >> - Lots of bug fixes >> >> For full details, see the changelog: >> http://svn.us.apache.org/repos/asf/tomcat/trunk/webapps/docs/changelog.xml >> >> >> It can be obtained from: >> https://dist.apache.org/repos/dist/dev/tomcat/tomcat-9/v9.0.0.M3/ >> The Maven staging repo is: >> https://repository.apache.org/content/repositories/orgapachetomcat-1062/ >> The svn tag is: >> http://svn.apache.org/repos/asf/tomcat/tags/TOMCAT_9_0_0_M3/ >> >> The proposed 9.0.0.M3 release is: >> [ ] Broken - do not release >> [X] Alpha - go ahead and release as 9.0.0.M3 > > +1 to release as alpha.
Thanks for the review. > One interesting new observation is removed attributes "servlets" and > "deploymentDescriptor" from Webapp MBeans. Not sure this was > intentional. It was. Part of the removal of JSR77 support. > Unit test observations: > > - in the message > "org.apache.catalina.core.TestStandardContextResources.testResources > testResources(): orderedLibs:" the value after the message has changed > from "[classes, resources.jar, resources2.jar]" in M1 to > "[resources.jar, classes, resources2.jar]" in M2 and M3 Hmm. classes shouldn't be there at all. It looks like the classpath scan is picking that up when it shouldn't be. I'll get that fixed. > - in org.apache.catalina.valves.TestStuckThreadDetectionValve in test > testDetection, after the correct message from notifyStuckThreadDetected > and the following org.apache.coyote.AbstractProtocol.pause, I get a > notifyStuckThreadCompleted message, which normally oesn't show up. > Probably just a timing issue, that the Valve ran between the pause and > the complete shutdown. Sounds plausible. I won't worry about this for now. > - new WARNINGs since M2: > > 2 x > org.apache.tomcat.util.net.NioEndpoint$NioSocketWrapper.doClientAuth > http11processor.socket.sslreneg Fixed the broken i18n message. > 25 x org.apache.tomcat.util.net.SSLUtilBase.getEnabled Some of the > specified [protocols] are not supported by the SSL engine and have been > skipped: [[SSLv2Hello]] On one side we have excluded SSLv2Hello from the protocols a user can select. On the other, we included it in the defaults. We need to be consistent here. I'm going to implement: - allow SSLv2Hello to be selected - enable SSLv2Hello by default > 3 x org.apache.tomcat.util.scan.StandardJarScanner.scan Failed to scan > [http://felix.extensions:9/] from classloader hierarchy, plus Caused by: > java.lang.IllegalArgumentException: URI scheme is not "file" Expected. > 3 x org.apache.catalina.startup.ContextConfig.validateSecurityRoles > Security role name foo used in an <auth-constraint> without being > defined in a <security-role> I'm guessing some new tests with security constraints. > - Since M2 testConnectToServerEndpoint no longer logs > "org.apache.tomcat.websocket.WsSession.doClose Failed to flush batched > messages on session close" Good. > - message org.apache.tomcat.util.modeler.Registry.unregisterComponent > Error unregistering mbean appears in varying numbers. Should be fixed or at least reduced now. > - Test testHeaderLimits3 since M3 logs a stack in addition to the > previous message: OK. > - lots of new "Broken pipe" > - most of them in org.apache.tomcat.websocket.server.TestClose, but > that one fails, see below > - other known occurrences are now often logged twice: Those look OK. One for the read listener and one for the write listener. > - in org.apache.coyote.http2.TestHttp2Section_3_2 for NIO in test > testConnectionNoPreface: Haven't been able to repeat this one. > - the only unit test failures is > org.apache.tomcat.websocket.server.TestClose for NIO: <snip/> > Probably fixed by r1728137. Confirmed. --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org