Is it a feature abuse to introduce default methods in new interfaces?

[Aaron Scott-Boddendijk]

We've used default methods in Spring-Data repositories where the obvious
implementation of a method is really just a preprocessing of arguments to
express the result in terms of the other interface methods (which are
dynamically generated) - eg two methods differing only by their
date-range-boundary exclusivity.

--
Aaron Scott-Boddendijk

On Fri, Feb 22, 2019 at 2:16 PM Joseph D. Darcy <joe.darcy at oracle.com>
wrote:

> Hello,
>
> FWIW, I wouldn't necessarily consider default methods on a new interface
> an abuse. The existence of default methods on interfaces changes the
> trade-offs between having an interface paired with an abstract helper
> implementation vs an interface with default methods.
>
> HTH,
>
> -Joe
>
>
> On 2/21/2019 6:54 AM, Daniel Fuchs wrote:
> > Hi Jaikiran,
> >
> > I don't think this is an abuse, and I'd say that the decision on
> > whether to provide a default implementation for a new interface
> > probably depends on how well you estimate that the default
> > implementation will match what concrete subclass would have to
> > implement. Even if that's not an authoritative answer ;-)
> >
> > I don't know if that's representative but Stream.Builder for
> > instance, came with a default add( ) method:
> >
> >
> https://docs.oracle.com/javase/8/docs/api/java/util/stream/Stream.Builder.html
> >
> >
> > best regards,
> >
> > -- daniel
> >
> > On 21/02/2019 14:35, Jaikiran Pai wrote:
> >> I have been trying to find some authoritative answer on when _not_ to
> >> use default methods on interfaces (a feature introduced in Java 8). The
> >> official documentation in Java, about this feature, talks about the
> >> motivation[1] behind introducing this feature and explains how
> >> _existing_ interfaces can be changed to add default methods without
> >> breaking binary compatibility:
> >>
> >> "Default methods enable you to add new functionality to the interfaces
> >> of your libraries and ensure binary compatibility with code written for
> >> older versions of those interfaces."
> >>
> >> This and some other docs don't mention whether it's an accepted practice
> >> or abuse of default methods feature, when dealing with new interfaces.
> >> Is it fine to introduce new interfaces with default methods? I have
> >> tried searching the mailing list archives to see if it's discussed
> >> previously but my search terms haven't found anything relevant. Any
> >> previous authoritative answer, similar to Stuart Mark's reply on
> >> stackoverflow (on a different topic) about Optional usage[2], is also
> >> good enough.
> >>
> >> [1]
> >> https://docs.oracle.com/javase/tutorial/java/IandI/defaultmethods.html
> >>
> >> [2]
> >>
> https://stackoverflow.com/questions/23454952/uses-for-optional/23464794#23464794
> >>
> >>
> >> -Jaikiran
> >>
> >>
> >
>
>