Request for approval: 8160174: java.net.NetworkInterface - fixes and improvements for network interface listing

[Langer, Christoph]

Hi Chris,

yes, the change for 8160174 would make the code mostly identical to the current JDK9 version, except for some calls to NET_ or JNU_ macros/functions wich are either not available in 8 or I didn't dare to touch.

Best
Christoph

> -----Original Message-----
> From: Chris Hegarty [mailto:chris.hegarty at oracle.com]
> Sent: Montag, 22. August 2016 16:04
> To: Langer, Christoph <christoph.langer at sap.com>
> Cc: jdk8u-dev at openjdk.java.net; Rob McKenna <rob.mckenna at oracle.com>
> Subject: Re: Request for approval: 8160174: java.net.NetworkInterface - fixes
> and improvements for network interface listing
> 
> Hi Christoph,
> 
> On 22/08/16 11:00, Langer, Christoph wrote:
> > Hi Chris,
> >
> > I understand your concerns regarding too much change here which could
> result in subtle differences that might not be wanted for a released version.
> >
> > The main motivation for me to integrate the change into JDK 8 is
> mergeability. In our SAP JVM 8 we had the need to do several fixes for
> problems on various of our supported platforms. And with the current coding
> layout it is very hard to do fixes, especially for AIX/Linux as all the #ifdefs make
> it a mess. So we already stepped to a version of code that merely matches the
> JDK9 version.
> 
> I understand, and can sympathize with this.
> 
> > But I agree that with my proposal
> (http://cr.openjdk.java.net/~clanger/webrevs/8160174.8udev/) I'm probably
> touching unnecessary places and make a review really hard.
> 
> Well after further thought, if we are going to make changes here,
> then maybe there is an argument for keeping the code consistent
> with 9, at least we end up with a single body of code.
> 
> 8160174 has been in JDK 9 for almost a month, and there have been
> no reported issues.
> 
> Is it the case that with your previous proposal that the 8u version
> of the file is identical to that of the 9 version?
> 
> -Chris.
> 
> > What about this proposal for downporting the fix to Bug 8158519:
> > http://cr.openjdk.java.net/~clanger/webrevs/8158519.8udev/
> >
> > Here I really only split the enumIPv*Interfaces methods to a clean structure
> and then make the necessary changes to eliminate getBroadcast() and
> getSubnet() functions in order to determine that information correctly in place
> before calling addif.
> >
> > Could you give a review for that?
> >
> > Thanks a lot
> > Christoph
> >
> >
> >
> >> -----Original Message-----
> >> From: Chris Hegarty [mailto:chris.hegarty at oracle.com]
> >> Sent: Donnerstag, 18. August 2016 17:49
> >> To: Langer, Christoph <christoph.langer at sap.com>
> >> Cc: jdk8u-dev at openjdk.java.net; Rob McKenna <rob.mckenna at oracle.com>
> >> Subject: Re: Request for approval: 8160174: java.net.NetworkInterface -
> fixes
> >> and improvements for network interface listing
> >>
> >> On 16 Aug 2016, at 15:41, Rob McKenna <rob.mckenna at oracle.com>
> wrote:
> >>>
> >>> Hi Christoph,
> >>>
> >>> If the patch has changed from 9 you will need a separate review.
> >>>
> >>> 	-Rob
> >>>
> >>> On 16/08/16 10:09, Langer, Christoph wrote:
> >>>> Hi,
> >>>>
> >>>> can I get approval for backporting the following fix:
> >>>>
> >>>>
> >>>> Original Bug: https://bugs.openjdk.java.net/browse/JDK-8160174
> >>>>
> >>>> Jdk9 change: http://hg.openjdk.java.net/jdk9/dev/jdk/rev/a8db670c7d12
> >>>>
> >>>> Jdk9 review thread: http://mail.openjdk.java.net/pipermail/net-dev/2016-
> >> July/010100.html
> >>>>
> >>>>
> >>>> I had to modify the jdk9 patch after unshuffling to get it to apply to 8udev.
> >> This is the new webrev:
> >> http://cr.openjdk.java.net/~clanger/webrevs/8160174.8udev/
> >>
> >> Wow, there are quite a lot of changes in this. I do remember reviewing this
> for
> >> 9 ( it
> >> took a long time ).  I do have a concern that this change may cause some
> subtle
> >> behavioural differences, since the underlying systems calls may be different.
> >> This
> >> may be acceptable for a major release, but not so for an update release.
> >>
> >> Is there a strong need for this to be backported?
> >>
> >> -Chris.