Exactly.

2006/8/3, Oleg Khaschansky <[EMAIL PROTECTED]>:
According to "15.9.3 Choosing the Constructor and its Arguments":

Once the actual arguments have been determined, they are used to
select a constructor of C, using the same rules as for method
invocations (§15.12).

On 8/3/06, Mikhail Loenko <[EMAIL PROTECTED]> wrote:
> Well constructor is not a method :)
> so RI and Harmony are not necessary violate the spec
>
> I'd probably stick to the current behavior
>
> Thanks,
> Mikhail
>
> 2006/8/3, Alexei Zakharov <[EMAIL PROTECTED]>:
> > As far as I understand RI simply take the first declared method. In
> > other words, if we swap constructors in the above example like this
> >
> > public static class MyBean {
> >       static String calledM = null;
> >
> >       public MyBean(Integer arg) {
> >           calledM = "new2";
> >       }
> >
> >       public MyBean(Object arg) {
> >           calledM = "new1";
> >       }
> >
> >   }
> >
> > the result will be "PASS".
> > :-/
> >
> >
> > Thanks,
> >
> > 2006/8/3, Mikhail Loenko <[EMAIL PROTECTED]>:
> > > How does RI behave if there are three methods? does it alway selects less
> > > specific?
> > >
> > > Thanks,
> > > Mikhail
> > >
> > > 2006/8/3, Alexei Zakharov <[EMAIL PROTECTED]>:
> > > > Hi community,
> > > >
> > > > I'd like to attract everyone's attention to another RI inconsistence.
> > > > It seems RI has a bug in the implementation of execute() method of the
> > > > java.beans.Statement class. The spec states:
> > > >
> > > > 1. "When the target's class defines many methods with the given name
> > > > the implementation should choose the most specific method using the
> > > > algorithm specified in the Java Language Specification (15.11)."
> > > > 2. "The reserved method name "new" may be used to call a class's
> > > > constructor as if all classes defined static "new" methods."
> > > >
> > > > But the following test shows that RI does not follow these rules –
> > > > does not choose the most specific method:
> > > >
> > > > import java.beans.*;
> > > >
> > > > public class StatementTest {
> > > >
> > > >    public static class MyBean {
> > > >        static String calledM = null;
> > > >
> > > >        public MyBean(Object arg) {
> > > >            calledM = "new1";
> > > >        }
> > > >
> > > >        public MyBean(Integer arg) {
> > > >            calledM = "new2";
> > > >        }
> > > >
> > > >    }
> > > >
> > > >    public static void main(String argv[]) throws Exception {
> > > >        Statement stmt = new Statement(MyBean.class, "new",
> > > >                new Object[] { new Integer(17) });
> > > >
> > > >        stmt.execute();
> > > >        if (!MyBean.calledM.equals("new2")) {
> > > >            System.out.println("FAIL");
> > > >        } else {
> > > >            System.out.println("PASS");
> > > >        }
> > > >    }
> > > >
> > > > }
> > > >
> > > > The result is "FAIL" on RI. However, it behaves correctly for regular
> > > > methods (that aren't constructors). I think I should post "Non-bug
> > > > differences from RI" JIRA. I am also going to correct our
> > > > implementation of Statement since it is now "compatible" with RI. Any
> > > > objections?
> > > >
> > > > Thanks,
> > > >
> > > > --
> >
> > --
> > Alexei Zakharov,
> > Intel Middleware Product Division

--
Alexei Zakharov,
Intel Middleware Product Division

---------------------------------------------------------------------
Terms of use : http://incubator.apache.org/harmony/mailing.html
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to