On 28/05/2019 15:39, Chris Hegarty wrote:
Please review this small change to ServerSocket::toString so that it
correctly implements its specified behaviour ( to reveal the local
address if SecurityManager::checkConnect succeeds ). Looks like this was
a minor oversight in the implementation that was not caught by testing,
as there was no fine grained test for it. This review contains a
comprehensive test.
Additionally, the ServerSocket::toString specification would benefit
from a minor clarification that checkConnect is only invoked when the
socket is bound. ( A CSR will be filed for this )
https://cr.openjdk.java.net/~chegar/8224730/webrev.00/
This looks good. I just wonder if there is any merit is extending the
test to exercise the socket adaptors returned by
SocketChannel::socketand ServerSocketChannel::socket or maybe checking a
separate test. The socket adaptors are often forgotten and I'm only
bringing it up here while the issue with toString is fresh.
-Alan
