[7u10] Request for approval: 7112670: Inet4AddressImpl should use getaddrinfo/getnameinfo

[Shi Jun Zhang]

On 9/18/2012 12:21 AM, Phil Race wrote:
> In addition to Alan's point, but not in any way specifically about 
> this fix,
> I'd like to ask why backport requests are unaccompanied by any kind
> of risk assessment and justification ? Do we lack a guideline on this ?
> Is the release manager expected to be able to deduce the risk etc ?
>
> 'Stabilisation' fixes, Escalated bugs, support for new versions of 
> platforms,
> key features that can't wait for JDK 8 are all potential reasons, and its
> really not hard to provide such reasons in the backport request.
>
> As Ihe "N" in 7uN increases, the number of fixes going in needs to 
> decrease,
> and the bar for the fixes, needs to be increased. The whole update 
> process
> needs to be move into something more like a 'rampdown' phase.
>
> All my opinion of course, and I'm sure there are other opinions out 
> there.
>
> -phil.
>
> On 9/17/12 2:06 AM, Alan Bateman wrote:
>>
>> For the record, I have concerns about back-porting this to 7u. The 
>> reason is that switching to getnameinfo and getaddrinfo can result in 
>> subtle behavioral changes that cause problems for applications and 
>> developers. A recent one in this area was:
>>
>> http://bugs.sun.com/view_bug.do?bug_id=7166687
>>
>> where the Mac port changed 7u4 to use getnameinfo, it was switched 
>> back in 7u5 to restore previous behavior. I realize the changes you 
>> are looking to back port are for the IPv4 case and maybe there are 
>> people running with IPv6 disabled but I think we need to fully 
>> understand any behavior changes before this is considered for a 7u 
>> release.
>>
>> -Alan.
>>
>>
>> On 17/09/2012 09:14, Shi Jun Zhang wrote:
>>> Hi all,
>>>
>>> I'd like to request for approval to push the following change into 
>>> 7u10.
>>>
>>> Changeset in jdk8
>>> http://hg.openjdk.java.net/jdk8/tl/jdk/rev/81987765cb81
>>>
>>> Webrev
>>> http://cr.openjdk.java.net/~zhangshj/jdk7u/7112670/webrev.00/
>>>
>>> Reviewed by chegar, alanb, mduigou, ngmr
>>>
>>> Review thread
>>> http://mail.openjdk.java.net/pipermail/net-dev/2011-November/003702.html 
>>>
>>>
>>
>
I'm also curious whether there is a guideline to tell us which should be 
backported and which should not.

-- 
Regards,

Shi Jun Zhang