regression: jdk1.7.0_04 java.net.InetAddress.getLocalHost().getHostName() return FQHN

[Wolfgang Hoschek]

Thanks. This bug is indeed fixed in java7u5 on Linux!

However, we ran into a new (related) regression when running on OSX: InetAddress.getLocalHost() throws UnknownHostException on java7u5 on OSX, which I have filed here: http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=7180557. 

Run the small test driver program attached to the bug database report on Oracle java7u5 on OSX. It throws an exception even though it should not. Note that the very same test driver program runs fine on java6u33 on the very same OSX host and config.

This regression prevents us from upgrading to java7.

Wolfgang.

On May 9, 2012, at 2:23 PM, Wolfgang Hoschek wrote:

> Awesome! Thanks!
> 
> Wolfgang.
> 
> On May 9, 2012, at 10:31 AM, Rob McKenna wrote:
> 
>> Hi Wolfgang,
>> 
>> I've pushed this into 7u5 which will GA on the 12th of next month. I'll push it into the current jdk7u repo once I get a (positive) response to the approval mail I'm about to send.
>> 
>>   -Rob
>> 
>> On 09/05/12 18:21, Wolfgang Hoschek wrote:
>>> Hi Alan,
>>> 
>>> Thanks for letting me know. I strongly encourage you all to revert the behavioral change on Linux and Mac not just in jdk8 but also in jdk7u as well. This critical java regression prevents us from upgrading to jdk1.7.0_04.
>>> 
>>> Best regards,
>>> Wolfgang.
>>> 
>>> On May 9, 2012, at 5:02 AM, Alan Bateman wrote:
>>> 
>>>> On 09/05/2012 10:15, Wolfgang Hoschek wrote:
>>>>> [7u6] In jdk1.7.0_04 GA java.net.InetAddress.getLocalHost().getHostName() return the fully qualified host name (e.g. "ip-10-172-17-93.us-west-1.compute.internal") instead of returning the hostname without the domain (e.g. "ip-10-172-17-93"). This is a regression wrt. all prior versions of jdk 1.7 and jdk 1.6. This breaks applications that expect the old behavior.
>>>>> 
>>>> This isn't really a bug, it's always been possible for getHostName to return the FQHN, it's just been very platform and configuration specific. However there was a behavior change on Linux and Mac in 7u4 and jdk8 that came via the addition of the Mac port. He has already been fixed in jdk8 as 7166687 [1] and I believe Rob intends to get it into jdk7u too.
>>>> 
>>>> -Alan
>>>> 
>>>> [1] http://hg.openjdk.java.net/jdk8/tl/jdk/rev/b26c04717735
>