RFR[8237890]: 'DatagramPacket::getSocketAddress doesn't specify what happens if address or port are not set'

[Alan Bateman]

On 10/04/2020 11:16, Patrick Concannon wrote:
>
> Hi,
>
> Could someone please review my webrev and CSR for JDK-8237890 
> 'DatagramPacket::getSocketAddress doesn't specify what happens if 
> address or port are not set' ?
>
> DatagramPacket::getSocketAddress is misleading in that it can throw an 
> IllegalArgumentException even though it doesn't take any arguments. 
> This can occur if the port of a DatagramPacket is not set before a 
> call to getSocketAddress is made.
>
> In the recent fixJDK-8236940 
> <https://bugs.openjdk.java.net/browse/JDK-8236940>, additional checks 
> were addedto ensure DatagramPacket cannot besent to port 0. Following 
> on from this update,the current fix changes the default port of a 
> DatagramPacket from -1 to 0. An IllegalArgumentException will 
> therefore no longer be thrown when getSocketAddress with no port set. 
> Instead, an InetSocketAddress representing any local address with port 
> 0 is returned.
>
>
> bug: https://bugs.openjdk.java.net/browse/JDK-8237890
> csr: https://bugs.openjdk.java.net/browse/JDK-8242483
> webrev: http://cr.openjdk.java.net/~pconcannon/8237890/webrevs/webrev.00/
>
The default value for address and port are null and 0 (resp.) so you 
could just remove L93-94.

In the getSocketAddress() spec it might be better to say "Returns" 
rather than "Gets" so that its consistent with the getAddress and 
getPort methods.

Can you look at DatagramSocketAdaptor L202 where the comment is "throws 
IAE if port not set". I assume that comment should be removed.

Do we need to re-visit or withdraw the release note JDK-8237530 as IAE 
is there is no longer a "port out of range" condition?

-Alan




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.java.net/pipermail/net-dev/attachments/20200410/d0939fcd/attachment.htm>