Subj : Re: Problem with ambiguous method - under Solaris, works in Windows
To   : Merten Schumann
From : Igor Bukanov
Date : Fri Aug 13 2004 09:31 pm

At the end it turned out to be Rhino bug in code resolving overloaded 
Java methods. See http://bugzilla.mozilla.org/show_bug.cgi?id=255549

Regards, Igor

Igor Bukanov wrote:
> Merten Schumann wrote:
> 
>> Hello,
>>
>> ran into a strange problem: a script works when executed with Sun JDK
>> 1.4.2 under Windows but not under Solaris. Besides from the JDK, I used
>> the same environment, I mean, my application's .jar, same Rhino js.jar
>> ...
>>
>> Under Solaris, I do get when I call a (public static) method
>> getResultSet("bla", null, null)
>>
>> org.mozilla.javascript.EvaluatorException:
>> The choice of Java constructor getResultSet matching JavaScript argument
>> types (string,null,null) is ambiguous; candidate constructors are:
>>  interface de.rochade.srap.ResultSet
>> getResultSet(java.lang.String[],java.lang.String[],java.lang.String),
>>  interface de.rochade.srap.ResultSet
>> getResultSet(java.lang.String,java.lang.String,java.lang.String)
>>
>> According to http://www.mozilla.org/rhino/ScriptingJava.html, I think I
>> could understand why Rhino cannot decide due to it's dynamic behaviour
>> at Runtime some method overloading things. So, let's say with given
>>    getResultSet("foo", null, null)
>> it cannot decide what to take from
>>    getResultSet(String, String, String)
>>    getResult(String[], String[], String)
>> (although this would be not nice for me, anyway, I would have a
>> work-around for this)
>>
>> But, the strange thing: if Rhino cannot decide which method to take, why
>> is it different under Windows and Solaris??? I think this is an JVM
>> issue then ...
> 
> 
> Hm, what 
> java.lang.System.out.println(de.rochade.srap.ResultSet.getResultSet.toString()) 
> would print when executed on Solaris if you add such line to your script 
> before you call the method?
> 
>>
>> When I call getResultSet("foo", "foo", null), it works. That's strange
>> too. I mean, if Rhino cannot distinguish String from String[], shouldn't
>> it throw "ambiguous" all the time here? Or, does it think "ok, we have
>> two Strings here, this can only be the method (String, String, String)
>> since in (String[], String[], String) is only one String" :-) 
> 
> 
> It comes from LiveConnect 3.0 specs which Rhino implements, see
> 
> http://www.mozilla.org/js/liveconnect/lc3_method_overloading.html
> 
> Regards, Igor

.