On 02/10/2013 18:28, Mark Thomas wrote: > On 01/10/2013 19:39, Mark Thomas wrote:
All the other issues I am aware of have been resolved. >> The fairly frequent (5%-10%) failure rate of TestCoyoteOutputStream >> still needs to be addressed. > > This appears to impact all connectors so at this point it is unlikely a > tomcat native change would be required. I've tracked down the root cause of this. The logic that triggers the first onWritePossible() and dataAvailable() events after setting the non-blcoking write/read listeners does not take account of the fact that the listeners may be set on a non-container thread. Also, in the test case the application code writes / reads without waiting for these events. I have a partial fix for this but I want to wait until I have the whole fix. Mark --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org
