2014-10-24 1:36 GMT+04:00 Christopher Schultz <[email protected]>: > Mark, > > On 10/23/14 3:48 PM, [email protected] wrote: >> Author: markt >> Date: Thu Oct 23 19:48:08 2014 >> New Revision: 1633936 >> >> URL: http://svn.apache.org/r1633936 >> Log: >> Fix some intermittent test failures with NIO2 on OSX (problems with the >> test, not the code) >> >> Modified: >> tomcat/trunk/test/org/apache/catalina/connector/TestCoyoteAdapter.java >> >> Modified: >> tomcat/trunk/test/org/apache/catalina/connector/TestCoyoteAdapter.java >> URL: >> http://svn.apache.org/viewvc/tomcat/trunk/test/org/apache/catalina/connector/TestCoyoteAdapter.java?rev=1633936&r1=1633935&r2=1633936&view=diff >> ============================================================================== >> --- tomcat/trunk/test/org/apache/catalina/connector/TestCoyoteAdapter.java >> (original) >> +++ tomcat/trunk/test/org/apache/catalina/connector/TestCoyoteAdapter.java >> Thu Oct 23 19:48:08 2014 >> @@ -360,7 +360,13 @@ public class TestCoyoteAdapter extends T >> public void run() { >> for (int i = 0; i < 20; i++) { >> try { >> - os.write(BYTES_8K); >> + // Some tests depend on this write failing (e.g. >> + // because the client has gone away). In some >> cases >> + // there may be a large (ish) buffer to fill >> before >> + // the write fails. >> + for (int j = 0 ; j < 8; i++) { >> + os.write(BYTES_8K); >> + } > > I'm fairly sure you want j++ here instead of i++, otherwise your tests > are going to take a very long time to complete. >
Agreed. Luckily this code is used by one test only (testBug54928()) and in that test the client goes away. That makes os.write() to fail with an IOException, avoiding an infinite loop. Best regards, Konstantin Kolinko --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
