Strange observation: MethodHandle.invokeWithArguments() would not work, whereas Method.invoke() would with the very same arguments

[Rony G. Flatscher]

Hmm, I ran into another such problem, where using core reflection invoke(...) succeeds whereas using
MethodHandle.invokeWithArguments(...) does not.

The problem is also linked to the resulting object from Arrays.asList(T... a), this time in the
context of a stream example causing a list that contains a single String array, instead of a list
that contains the String array elements as its individual list elements.

Here is the ooRexx code (the tilde '~' is the explicit message operator, Rexx is caseless and
dynamically typed), which was originally transcribed 1:1 from a Java example and which works with
core reflection:

    -- define a Rexx string containing letter 'k' in some of its words wordstring="Just a bunch of
    words to test for killer items containing a k" -- turn Rexx-string into a Java-string: this
    allows us to use Java's String methods refWordString=.bsf~new("java.lang.String", wordstring) --
    convert the Java string into a Java List (a Collection):
    alist=bsf.loadClass("java.util.Arrays")~asList(refWordString~split(" ")) -- create a RexxProxy
    of our Worker class which implements the two functional interface -- methods that we use
    rexxWorker=BsfCreateRexxProxy(.worker~new, , "java.util.function.Predicate",
    "java.util.function.Consumer") -- now run a filter stream operation on the list -- (the filter
    just selects words containing the letter 'k' or 'K') sa=alist~stream~filter(rexxWorker)~toArray
    -- "filter" employs the Predicate interface -- print the results for verification: loop y over
    sa say y end say "-----------------------" - now run a foreach operation on a stream -- the
    consumer here just prints inputs with some surrounding brackets alist~stream~foreach(rexxWorker)
    -- "forEach" employs the Consumer interface ::requires BSF.CLS -- get access to the Java bridge
    ::class Worker -- define Rexx class that implements the functional methods -- implements the
    interface java.util.function.Predicate ::method test -- will return .true for strings containing
    'k' or 'K', .false else use arg s -- fetch argument (from Java) return s~caselessPos('k')>0 --
    implements the interface java.util.function.Consumer ::method accept -- will show each string
    use arg s -- fetch argument (from Java) say ">>"s"<<" -- show argument

The solution here was the same as already described earlier: in order to use
MethodHandle.invokeWithArguments() do not use the result of Arrays.asList(array) and apply stream()
on it , but rather create the stream directly with Arrays.stream(array) each time a stream is needed.

It is as if the object returned by Arrays.asList(...) is being handled differently if using the core
reflection invoke(...) compared to using MethodHandle.invokeWithArguments(...) for unknown reasons
(the arguments are processed the same in both cases).

---rony


On 04.03.2018 19:05, Rony G. Flatscher wrote:
>
> OK, went back to do the changes (making sure to cast invokeWithArguments(...) to (Object[]) and
> discovered that I forgot a hard-coded override to use core reflection in all cases.
>
> Removing that hard-coded flag to use MethodHandles instead made the problem reappear, so I once
> more went after it in the past hours and think I can now explain this more exact: the case in
> question is a sequence of the following (transcribed to Java) statements that are executed in the
> BSF4ooRexx bridge on behalf of the Rexx program (which itself was transcribed from a Java example
> found on the Internet):
>
>   * "javafx.collections.FXCollections.observableArrayList()" returns an object that implements the
>     interface "javafx.collections.ObservableList", actually an instance of the non-exported
>     "com.sun.javafx.collections.ObservableListWrapper" class,
>   * a String array of month names gets turned into an ArrayList using
>     "java.util.Arrays.asList(monthNames)" returning listOfMonthNames (a "java.util.Arrays$ArrayList")
>   * "ObservableList.addAll(listOfMonthNames)" gets invoked,
>   * supplying the resulting ObservableListWrapper as an argument to
>     "javafx.scene.chart.CategoryAxis.setCategories(javafx.collections.ObservableList)" works when
>     invoked via core reflection, but causes an exception if invoked via a MethodHandle (the
>     exception being: "java.lang.ClassCastException: java.base/[Ljava.lang.String; cannot be cast
>     to java.base/java.lang.String")!
>
> The problem, it turns out is the use of "addAll(listOfMonthNames)" which causes the ObservableList
> to contain a single list member "java.util.Arrays$ArrayList at xxx" (toString() yielding
> "[Ljava.lang.String;@yyy"). The core reflection invocations work without an error, the
> MethodHandle.invokeWithArguments(...) version causes the above exception.
>
> Changing the code to use the String array monthNames (not its list version) as an argument (hence:
> "ObservableList.addAll(monthNames)" ) makes it run with both, the core reflection and the
> MethodHandle invocation versions.
>
> ---rony
>
>
>
>
>
> On 03.03.2018 19:02, Rony G. Flatscher wrote:
>> John, thank you very much for your kind reply and hints!
>>
>> On 02.03.2018 02:47, John Rose wrote:
>>> On Feb 12, 2018, at 11:59 AM, Rony G. Flatscher <Rony.Flatscher at wu.ac.at
>>> <mailto:Rony.Flatscher at wu.ac.at>> wrote:
>>>>
>>>> While testing a rather complex one (an adaption of the JavaFX address book example enhanced with a
>>>> BarChart, [1]), that exhibits a very strange behavior: when setting the values for the CategoryAxis
>>>> supplying an ObservableList of the month names in the current Locale, using a MethodHandle and
>>>> invoking it with invokeWithArguments() would yield (debug output):
>>>
>>> I just happened to see your message about your adventures with method handle:
>>>
>>> http://mail.openjdk.java.net/pipermail/jigsaw-dev/2018-February/013603.html
>>>
>>> This isn't really a jigsaw question, so I'm replying to mlvm-dev.
>> Yes, I just was not sure where to post such an e-mail (as on jigsaw MethodHandle's were suggested
>> I posted it there).
>>
>>> It looks like you are mixing or interconverting arrays of strings
>>> with lists of strings.  The print statements and CCE show that
>>> you are passing an array of strings into a place which expects
>>> a single string, and the print statements suggest you are
>>> in fact passing a list containing a string array into a place
>>> which expects a list of strings.  Either way there are too
>>> many brackets in your actual argument.
>>>
>>> The prime suspect when the number of brackets is off by one
>>> is varargs.  You code might be failing because of surprises
>>> in the overload resolution of invokeWA, which accepts
>>> a varargs Object array *and* a single List.
>>>
>>> Is your runtime invoke mechanism treating invokeWA as an ordinary
>>> method?  It is rather extraordinary, and may warrant a second look.
>>>
>>> String str;
>>> Object obj;
>>> Object[] aobj;
>>> String[] astr;
>>> List lst;
>>>
>>> plain single arity invocations:
>>>
>>> 1 mh.invokeWithArguments(str) => new Object[] { str }
>>> 2 mh.invokeWithArguments(obj) => new Object[] { obj }
>>>
>>> and yet:
>>>
>>> 3 mh.invokeWithArguments(aobj) => aobj (multiple args)
>>> 4 mh.invokeWithArguments(astr) => astr (multiple args again!)
>>> 5 mh.invokeWithArguments(lst) => lst.toArray() (multiple args again!)
>>>
>>> but again, a cast removes varargs:
>>>
>>> 6 mh.invokeWithArguments((Object) aobj) => new Object[] { aobj }
>>> 7 mh.invokeWithArguments((Object) astr) => new Object[] { astr }
>>> 8 mh.invokeWithArguments((Object) lst) => new Object[] { lst }
>>>
>>> Your bug looks like a confusion between two of these,
>>> perhaps 5 and 8.
>> The invocation occurs with an aobj kind of argument (an array of coerced arguments matching the
>> types of the parameterTypes array), defined as "Object [] coercedArgs=..."); nevertheless will
>> make sure to cast explicitly everywhere invokeWithArguments() gets used (there are
>> MethodHandle.invoke(...) as well, which I will replace with invokeWithArguments instead, just to
>> have everything use the same invocation method consistently in case arguments need to be
>> supplied; also foregoing bindTo().invoke(...), replacing it with appropriate
>> invokeWithArguments() for consistency).
>>
>> ---rony
>>
>> P.S.: After having rewritten the Java reflection (adding caching of Field, Method, Constructor
>> objects together with their corresponding MethodHandle objects) part I also rewrote some of the
>> core ooRexx/C++/JNI stuff to remove old-standing (18+ years) code with modern ooRexx-4.x-API
>> code, which allows to forgo many of the old restrictions, and improving speed in those corners as
>> well (still some cleanup to do). After getting the test units to pass, I turned to this
>> application, and that problem using MethodHandle invocations only does not surface anymore! 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.openjdk.java.net/pipermail/mlvm-dev/attachments/20180308/fd9c351f/attachment.html>