tomcat-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Mark Thomas <>
Subject Re: svn commit: r1358055 - in /tomcat/trunk: java/org/apache/catalina/connector/ java/org/apache/catalina/core/ java/org/apache/coyote/ java/org/apache/coyote/ajp/ java/org/apache/coyote/http11/ java/org/apache/coyote/spdy/ java/org/apache/tomcat/util/net
Date Sun, 08 Jul 2012 15:44:10 GMT
On 07/07/2012 22:06, Mark Thomas wrote:
> On 07/07/2012 14:58, Mark Thomas wrote:
>> On 06/07/2012 17:35, Filip Hanik (mailing lists) wrote:
>>> Turning on Java 7 does change the test landscape.
>>> Right now, I can get the test suite to run fine on Java 6, but lots of errors
on Java 7
>> I've tracked down the first of these issues. Calling
>> java.nio.channels.SocketChannel.socket().setTrafficClass()
>> after
>> socketChannel.connect()
>> has been called triggers a SocketException. Moving the setTrafficClass()
>> call to before connect() fixes the issue.
>> Tests on Java 6 show that the setTrafficClass() was having no effect if
>> called after connect(). It looks like Java 7 has started to throw an
>> exception when the setter has no effect.
>> On this basis, I' say that this particular issue is a long standing bug
>> that was masked by Java 6 and is now visible with Java 7.
>> I'll do a code review and move the setTrafficClass() calls as necessary.
> Having done all of this, some further research on the users list
> suggests that this is a Windows XP / Server 2003 specific issue.

Tests on Linux show that calls to setTrafficClass() after accept() have
no effect. On this basis, I believe removing the option to set the
traffic class in those cases was correct.

> That means the code I deleted in the accept() case almost certainly
> needs to be restored.

The deletion stands.

I currently see no test failures on Windows. Linux testing not yet complete.


To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message