On Tue, 10 May 2022 at 14:59, Gilles Sadowski <gillese...@gmail.com> wrote:

> Le mar. 10 mai 2022 à 09:53, Alex Herbert <alex.d.herb...@gmail.com> a
> écrit :
> >
> > On Tue, 10 May 2022 at 02:32, Matt Juntunen <matt.a.juntu...@gmail.com>
> > wrote:
> >
> > > Sounds reasonable to me. Are there any arguments against this change
> > > other than the fact that it is not a japicmp default setting?
> > >
> >
> > I do not know why the setting default is at the current value. Unlike the
> > documentation for revapi there is limited explanation of the default
> > settings in japicmp and why compatibility for binary or source will be
> > broken. It may be that the developer explicitly wished to be informed of
> > additions to interfaces.
> >
> > Since this is unlikely to affect much at all it may be fine left as is in
> > commons parent. The configuration can be added to the relevant POM in
> > Commons RNG.
>
> IIUC, it should be the other way around: If BC is not broken, the
> common Commons settings should not report otherwise, and if
> some specific component has additional requirements, let it modify
> its own POM.
>
>
The original change was added to commons parent by Gary Gregory, I presume
because a downstream project added a default method to an interface and
japicmp complained. This must be the first instance in commons where a
default implementation is added for an existing interface method. I can add
the change to CP when I know it is definitely correct.

I have finished the implementation in RNG. I am going through testing the
snapshot with code that was built against the previous version.

Alex

Reply via email to