So what is the use case for the getters? Any examples? Because whoever is
in charge of constructing of the PushBuilder has already determined and
intended what is being pushed. I don't see why they need to be queried.
Also will the builder methods throw ISE once the push() occurs?
Cheers,
Paul
On Sat, Nov 21, 2015 at 8:55 AM, Mark Thomas <markt_at_apache.org> wrote:
> On 21/11/2015 13:27, Greg Wilkins wrote:
> >
> > On 21 November 2015 at 23:57, Paul Benedict <pbenedict_at_apache.org
> > <mailto:pbenedict_at_apache.org>> wrote:
> >
> > If there was a concluding build() method (and from memory I don't
> > think there was), then I am completely on board with how the methods
> > have been designed.
> >
> >
> > There is a concluding push() method, which "produces" the actual push,
> > so this is really a builder.
> >
> >
> > Ed,
> >
> > I don't mind the noop approach either, but if others find it a problem,
> > at least the isPushSupported method makes it unnecessary.
>
> From my limited experimentation with the current API I think a
> isPushSupported() method is necessary. There is the potential for an
> application to do a lot of work to determine which resources to push and
> to construct the necessary pushes. All of that work is wasted if push is
> not supported. The application needs to be able to check if push is
> supported or not.
>
> Mark
>
>