https://bz.apache.org/bugzilla/show_bug.cgi?id=63022
Remy Maucherat <r...@apache.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |NEEDINFO
--- Comment #2 from Remy Maucherat <r...@apache.org> ---
During async there's no IO operation going on, so no way to detect a socket
disconnect. As a result, nothing will actually get closed magically. You need
to perform an IO operation to find it out, otherwise you can remain in async
mode forever.
So the test closes the socket, and then async completes. The problem is that
the wrapper isClosed() now returns true, which will prevent the async complete
processing (SocketProcessorBase.run calls isClosed()). This is not normal, but
closing using introspection is not either. So maybe there's a "problem" here,
but the test is not a representation of a bug which needs to be fixed. Also
maybe SocketWrapper.isClosed() shouldn't check the socket state and this could
solve this situation along with more legitimate ones, but this needs to be
carefully reviewed.
Set need info to investigate and review some more, just in case.
--
You are receiving this mail because:
You are the assignee for the bug.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org
For additional commands, e-mail: dev-h...@tomcat.apache.org
