From the subsequent mails from Robert and Anil, there is a proposal that the
change will be enhanced to fix equals() to be backward compatible. +1 to that.
Thanks,
Faseela
From: Abhijit Kumbhare [mailto:abhijitk...@gmail.com]
Sent: Monday, April 23, 2018 3:05 AM
To: Luis Gomez
Cc: Faseela K ; R
On Sun, Apr 22, 2018 at 11:22 AM, Robert Varga wrote:
> On 22/04/18 10:01, Anil Vishnoi wrote:
> >
> > If we can make Ipv4Address.equals() final, we can define it in way,
> > which would prevent this sort of breakage happening again -- but
> > that is:
> > 1) a restriction on how
On Sun, Apr 22, 2018 at 11:18 AM, Robert Varga wrote:
> On 22/04/18 10:01, Anil Vishnoi wrote:
> >
> > To drive the argument to conclusion, though, is the analysis why
> things
> > broke -- which is because Ipv4Address.equals() uses getClass() to
> > determine equality. Ipv4AddressNoZ
+100. Couldn’t have said it better. Folks - please confirm the answers to
Luis’s questions.
On Sun, Apr 22, 2018 at 1:16 PM Luis Gomez wrote:
> I hope as a community we can learn some lessons from this breakage:
>
> - Upstream projects: please do not push changes breaking or changing API
> behav
I hope as a community we can learn some lessons from this breakage:
- Upstream projects: please do not push changes breaking or changing API
behavior without notice or weather item. If the impact is unintentional please
revert quick to minimize downtime and confusion.
- Downstream projects: plea
On 22/04/18 10:01, Anil Vishnoi wrote:
>
> If we can make Ipv4Address.equals() final, we can define it in way,
> which would prevent this sort of breakage happening again -- but
> that is:
> 1) a restriction on how generated code can be used
> 2) a relatively high-risk change (
On 22/04/18 10:01, Anil Vishnoi wrote:
>
> To drive the argument to conclusion, though, is the analysis why things
> broke -- which is because Ipv4Address.equals() uses getClass() to
> determine equality. Ipv4AddressNoZone inherits that -- it's not what
> Ipv4AddressNoZone type or
Here are the revert patches in lispflowmapping and bgpcep.
https://git.opendaylight.org/gerrit/#/c/71187/
https://git.opendaylight.org/gerrit/#/c/71193/
However, FYI that Anil has removed his -1 on ofp patch, and a weather is
already in place. Also genius and netvirt CSITs are passing with the o
On Fri, Apr 20, 2018 at 12:37 AM, Robert Varga wrote:
> On 19/04/18 21:49, Anil Vishnoi wrote:
> > It is not a matter of 'it could be supported', anything not defying
> laws
> > of Physics can be supported. The question is: does it make sense? can
> > you show an example when it would