RFR: 8022445: fix RMISocketFactory example to avoid using localhost

[Bernd Eckenfels]

All Stream Sockets where you specify an bind address (different from InAddrAny aka "::" or 0.0.0.0) are only bound to a single address family. You can't bind a single listening socket to a dual stack localhost IP Address. getLoopbackAddress is either ::1 or 127.0.0.1 (depending on the stack preferemces) and InetAddress.getByName("localhost") is also one of both (but not necesarily the same, unfortunately). This happened to me in the past, so I wanted to make sure that asumption does not show up in sample code of JavaDocs. Thanks for improving this!

Bernd

Am 22.08.2013 um 10:54 schrieb Alan Bateman <Alan.Bateman at oracle.com>:

> On 22/08/2013 01:58, Stuart Marks wrote:
>> Hi all,
>> 
>> Please review this small documentation change to the example I added to RMISocketFactory a couple weeks ago [1]. This change fixes a problem that Bernd Eckenfels pointed out [2] where the interface returned by getLoopbackAddress() might not match the interface that "localhost" resolves to.
> We don't have support for IPv4-only or IPv6-only listeners (at least not with stream oriented sockets, there is support for protocol specific datagram sockets but that doesn't apply here). So I'm curious about the original comments and whether there are configurations or loopback address plumbing where getLoopbackAddress doesn't do the right thing?
> 
> -Alan