Hi Patrick,
Looks good to me.
I realize we haven't specified what happens if the `laddr` is null.
Are you planning to fix that in a separate changeset or tag it to
this one?
Maybe we should unify the description of the `port` parameter in
the two constructors:
327 * @param port port to use.
357 * @param port local port to use
Maybe:
* @param port local port to use in the bind operation.
Nit: It's customary to break line before <p>:
343 * address. <p><a id="PortRange"></a>The local port must be
between 0 and
suggest:
343 * address.
* <p><a id="PortRange"></a>The local port must be
best regards,
-- daniel
On 28/04/2020 10:33, Patrick Concannon wrote:
Hi,
Could someone please review my fix for JDK-8243507 'DatagramSocket
constructors don’t always specify what happens when passed invalid
parameters'?
Currently, the DatagramSocket constructor `DatagramSocket(SocketAddress
bindaddr)` doesn't specify what happens if passed a SocketAddress
subclass not supported by this socket.
Also, there are two DatagramSocket constructors that accept a port
number, but neither constructor specifies what happens when passed port
0, or a port which falls outside of the valid range of port numbers i.e
between 0 and 65535 inclusive.
This fix updates the spec for each these constructors to inform the user
of what happens when passed an invalid argument. For the constructors
that take a port, the spec will now specify that an
IllegalArgumentException is thrown when passed a port outside of the
acceptable range, or, if zero is passed, that the system will choose an
appropriate port for them. For the constructor that takes a
SocketAddress, an IllegalArgumentException will be thrown if an invalid
SocketAddress subclass is passed as a parameter.
bug: https://bugs.openjdk.java.net/browse/JDK-8243507
csr: https://bugs.openjdk.java.net/browse/JDK-8243976
webrev: http://cr.openjdk.java.net/~pconcannon/8243507/webrevs/webrev.00/
Kind regards,
Patrick
