some nashorn testing

[Andreas Rieber]

On 08.01.13 19:12, Jim Laskey (Oracle) wrote:
> On 2013-01-08, at 1:22 PM, Andreas Rieber <rieberandreas at gmail.com> wrote:
>
>> Hi,
>>
>> i tested Nashorn with a project i am currently working on (scripting
>> java). The first thing i spotted is that Rhino comes with print() and
>> println() as default functions in:
>>
>> jdk8/jdk/src/share/classes/com/sun/script/javascript/RhinoScriptEngine.java
>>
>> while Nashorn has only a print():
>>
>> jdk8/nashorn/src/jdk/nashorn/api/scripting/NashornScriptEngine.java
>> jdk8/nashorn/src/jdk/nashorn/api/scripting/resources/engine.js
>>
>> Will this be changed or is that a migration issue?
> We opted for print because we were primarily using V8 performance tests and v8 only supports print.  Standalone rhino does not support println either.  We did have a switch for turning on println, but it fell out of favour to be consistent with other VMs.
>
> You can duplicate the behaviour with;
>
> function print(s) { java.lang.System.out.print("" + s); }
> function println(s) { java.lang.System.out.println("" + s); }

Well, i would use the print/println versions from 
jdk8/jdk/src/share/classes/com/sun/script/javascript/RhinoScriptEngine.java 
because they use context.getWriter(). One might set/change with 
context.setWriter().


>> importClass works only with: load("nashorn:mozilla_compat.js");
>> What is with importPackage (also used in jrunscript)?
>>
> importClass is being phased out since it duplicates the builtin behaviour of Packages and JavaClass.
>
> importPackage(java.awt);
> importClass(java.awt.Frame);
>
> is just
>
> var awt = java.awt;
> var Frame = java.awt.Frame;
>
>
>> Next problem i had was related to java array creation. Here the sample
>> from JSE 7 Documentation "Java Scripting Programmer's Guide":
>>
>> // create Java String array of 5 elements
>> var a = java.lang.reflect.Array.newInstance(java.lang.String, 5);
>>
>> // Accessing elements and length access is by usual Java syntax
>> a[0] = "scripting is great!";
>> print(a.length);
>>
>> It works with Rhino but fails with Nashorn:
>>
>> Exception in thread "main" java.lang.RuntimeException: java.lang.NoSuchMethodException: None of the fixed arity signatures [(java.lang.Class, int[]), (java.lang.Class, int)] or the variable arity signatures [(java.lang.Class, int...)] of the method java.lang.reflect.Array.newInstance match the argument types [org.dynalang.dynalink.beans.StaticClass, java.lang.Integer]
>>
>> The default number type for Rhino in vararg functions is double while
>> Nashorn has also integer (nice), might be a migration issue.
>
> We are pushing the Java syntax here to eliminate confusion between the java class and the JavaClass javascript object.  This eliminates the confusion when accessing members of the (String) class when a javascript object and members of the class itself when a java object.
>
> jjs> java.lang.String
> [JavaClass java.lang.String]
> jjs> java.lang.String.class
> class java.lang.String
>
> so
>
> var a = java.lang.reflect.Array.newInstance(java.lang.String.class, 5);

I worked around the issues, no problem. I use IcedTea7 which comes with 
Rhino (no standalone Rhino). So i just wrote it down for others which 
might want to migrate later.

>> Is jrunscript and the scriptpad sample application
>> (jdk8/jdk/src/share/sample/scripting/scriptpad) now also developed
>> under Nashorn project? I would like to contribute a patch for
>> jrunscript (printf never worked and i added also a sprintf), the
>> scriptpad needed only a few fixes to work with Rhino. I signed the OCA
>> recently.
> Contributions welcome.

Coming soon, need to change the patch so it works with Rhino and 
Nashorn. The number handling is pretty interesting to handle now.

>
> I just see Attila answered differently, but both forms work.
>
>
>> best regards
>> Andreas
>>