maximum recursion depth exceeded

[zxpatric]

what to achieve:

Expect to use pyro to share 2M data acquired in Jython (client) to Python. Tried file but it was too slow. Open to suggestion of other ways.

Client using pyrolite-4.15.jar:

self._pyroP.call("saveLeft", bytearray('123'))

Server Pyro 4.53 :

@Pyro4.expose
def saveLeft(self, data):
    dataLen = len(data)
    self._leftData = bytearray(dataLen)
    self._leftData[0:dataLen] = data

Error Message:

Exception in thread Thread-1:Traceback (most recent call last): File , line 222, in _Thread__bootstrap self.run() File , line 149, in run self._pyroP.call("saveLeft", bytearray('123')) RuntimeError: maximum recursion depth exceeded (Java StackOverflowError)

[irmen]

Can you paste the full java stacktrace. Also: what happens when you upgrade to the latest Pyrolite release.

[zxpatric]

I tried the latest build pyrolite-4.18.jar and upgraded Pyro4 to Pyro4-4.54. Still same thing.

Jython doesn't give the full java stacktrace or I did find howto. I wrote a pure java application but still struggle to find out java equivalent of jython call "self._pyroP.call("saveLeft", bytearray('123'))"

I tried byte[] data but rejected by serpent. I tried string but it doesn't reproduce the recursion error. Please suggest.

        String _uri = "PYRO:obj_221d338e088844cb934ce386c8056465@localhost:57892";
        System.out.println(_uri.getClass().getName());
        PyroProxy _pyroP = new PyroProxy(new PyroURI(_uri));
        System.out.println(_pyroP.call("greeting"));

// byte[] CDRIVES = hexStringToByteArray("e04fd020ea3a6910a2d808002b30309d"); // byte[] data = "Any String you want".getBytes(); // System.out.println(data.getClass().getName()); String data = "Any String you want"; _pyroP.call("saveLeft", data);

        System.out.println(_pyroP.call ("getLeft"));

        _pyroP.close();

[irmen]

Here's some Java example code that uses the native Java API to Pyrolite https://github.com/irmen/Pyrolite/blob/master/java/src/test/java/net/razorvine/examples/EchoExample.java

A bytearray is obviously just new byte[] {0x12, 0x34, 0x45} in Java....? Does that crash Pyrolite if you pass that as an argument to call?

[irmen]

I can't reproduce your problem.

Here are some tests I did with Jython. (First I started the built-in Pyro test echoserver, python -m Pyro4.test.echoserver)

$ jython
Jython 2.7.1b3 (default:df42d5d6be04, Feb 3 2016, 03:22:46)
[OpenJDK 64-Bit Server VM (Azul Systems, Inc.)] on java1.8.0_102
Type "help", "copyright", "credits" or "license" for more information.
>>> import serpent
>>> serpent.dumps(bytearray('123'))
"# serpent utf-8 python2.6\n{'data':'MTIz','encoding':'base64'}"
>>> import Pyro4
>>> p=Pyro4.Proxy("PYRONAME:test.echoserver")
>>> p.echo("test")
u'test'
>>> p.echo(bytearray('1234'))
{u'data': u'MTIzNA==', u'encoding': u'base64'}
>>> _["data"].decode("base-64")
'1234'

It is by design that serpent encodes binary data in a base-64 encoded string structure. That is easily converted back to the actual bytes as you can see.

Now for the native Pyrolite Java API, there's no problem what I can see with giving it bytes:

package net.razorvine.examples;

import java.io.IOException;
import java.util.Map;

import javax.xml.bind.DatatypeConverter;

import net.razorvine.pyro.NameServerProxy;
import net.razorvine.pyro.PyroProxy;
import net.razorvine.pyro.PyroURI;

public class EchoBytes
{
    public static void main(String[] args) throws IOException
    {
        // First start Pyro's test echoserver in a terminal:  python -m Pyro4.test.echoserver -N

        // get proxy via name server:
//      NameServerProxy ns = NameServerProxy.locateNS(null);
//      PyroProxy p = new PyroProxy(ns.lookup("test.echoserver"));
//      ns.close();
        // or get it directly:
        PyroProxy p = new PyroProxy(new PyroURI("PYRO:test.echoserver@localhost:50205"));

        byte[] data = new byte[]{0x12, 0x34, 0x45};

        Object result = p.call("echo", data);
        System.out.println("return value:");
        System.out.println("[class: "+result.getClass()+"]");
        System.out.println(result);

        // bytes/bytearry is serialized in Serpent as a structure with encoding (base64) and the data (ascii-encoded bytes)
        @SuppressWarnings("unchecked")
        Map<String, String> h = (Map<String,String>)result;
        if(h.get("encoding").equals("base64")) {
            byte[] resultdata = DatatypeConverter.parseBase64Binary(h.get("data"));
            System.out.println("result data:");
            for(byte b: resultdata)
            {
                System.out.println("byte 0x"+Integer.toHexString(b));
            }

        } else {
            System.err.println("unexpected encoding: "+h.get("encoding"));
        }

        p.close();
    }
}

When running this, it prints:

return value:
[class: class java.util.HashMap]
{encoding=base64, data=EjRF}
result data:
byte 0x12
byte 0x34
byte 0x45

So I'm closing this issue. If you can't solve your problem you can reopen it but you have to provide a minimal set of working program code that I can run to reproduce the problem.

[zxpatric]

I still can reproduce the error in my project. Have uploaded a zip of PyCharm project at https://drive.google.com/open?id=0B185R70hbZHkZVkybi1sWnhCQzQ

Let me know if it runs for you. Thanks.

[zxpatric]

Sorry to get back to you late. Since I can't reopen the issue, a comment has been given at the end of this closed issue.

On Sat, Feb 11, 2017 at 7:41 AM, Irmen de Jong notifications@github.com wrote:

Closed #149 https://github.com/irmen/Pyro4/issues/149.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/irmen/Pyro4/issues/149#event-958326281, or mute the thread https://github.com/notifications/unsubscribe-auth/ACnBWWn--eY35-RsmXCvUDeUM4lZC03Fks5rbayQgaJpZM4L9w8S .

[irmen]

I'm willing to look at it further but only if you trim down the code a lot. I don't have the time to try to get this full 30 mb project to work and debug all the code of it.... Please create a smaller example that shows the issue, and tell us exactly what to run to reproduce the problem?

[zxpatric]

the code is not much. The zip includes a jython-standalone-2.7.0.jar of 36M that let you run the Jython 2.7.0...

On Mon, Feb 13, 2017 at 1:22 PM, Irmen de Jong notifications@github.com wrote:

I'm willing to look at it further but only if you trim down the code a lot. I don't have the time to try to get this full 30 mb project to work and debug all the code of it.... Please create a smaller example that shows the issue, and tell us exactly what to run to reproduce the problem?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/irmen/Pyro4/issues/149#issuecomment-279476945, or mute the thread https://github.com/notifications/unsubscribe-auth/ACnBWeNegCKMj_Vs2wLl5StAsDEMh1jHks5rcJ9pgaJpZM4L9w8S .

[zxpatric]

Download the project and run main.py from console (or best from pycharm), you will see the problem.

On Mon, Feb 13, 2017 at 1:24 PM, Patrick Zhou zxpatric@gmail.com wrote:

the code is not much. The zip includes a jython-standalone-2.7.0.jar of 36M that let you run the Jython 2.7.0...

On Mon, Feb 13, 2017 at 1:22 PM, Irmen de Jong notifications@github.com wrote:

I'm willing to look at it further but only if you trim down the code a lot. I don't have the time to try to get this full 30 mb project to work and debug all the code of it.... Please create a smaller example that shows the issue, and tell us exactly what to run to reproduce the problem?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/irmen/Pyro4/issues/149#issuecomment-279476945, or mute the thread https://github.com/notifications/unsubscribe-auth/ACnBWeNegCKMj_Vs2wLl5StAsDEMh1jHks5rcJ9pgaJpZM4L9w8S .

[irmen]

It is not going to run as provided on my system. Please get rid of unneeded code and dependencies (it crashes on numpy and pyqt) that are of no consequence to reproducing the issue, and I'm willing to take another look.

[zxpatric]

Please get https://drive.google.com/open?id=0B185R70hbZHkV2NVN053cVJzc1U

you shall be able to run and see the following. Let me know if anything.

self._pyroP.call("saveLeft", self._data)
self._pyroP.call("saveRight", self._data)

RuntimeError: maximum recursion depth exceeded (Java StackOverflowError) IllegalArgumentException: java.lang.IllegalArgumentException: couldn't introspect javabean: java.lang.reflect.InvocationTargetException

On Mon, Feb 13, 2017 at 1:31 PM, Irmen de Jong notifications@github.com wrote:

It is not going to run as provided on my system. Please get rid of unneeded code and dependencies (it crashes on numpy and pyqt) that are of no consequence to reproducing the issue, and I'm willing to take another look.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/irmen/Pyro4/issues/149#issuecomment-279479337, or mute the thread https://github.com/notifications/unsubscribe-auth/ACnBWRG7CPwGwHfd_TfGHtzooUdkESDBks5rcKFxgaJpZM4L9w8S .

[irmen]

Thanks, I'll have a look.

In the meantime, I noticed that jython has its own share of classes representing various Python types. See http://javadox.com/org.python/jython-standalone/2.7-rc1/org/python/core/PyObject.html and its derived classes I suspect 2 things:

• the problem is specific to Serpent, not Pyro or Pyrolite. If I have proof this issue should reallly be moved to the serpent project ( https://github.com/irmen/Serpent )
• the problem is because Serpent doesn't know about these Jython classes, and when calling the serializer from Jython code, it somehow gets stuck in a loop.

[zxpatric]

Thanks. I made some efforts to use the Pyro to communicate memory blob between Python 3.X, Python 2.7 and/or Jython (python 2.7 based and communicate through Pyrolite package). So far, there are still some obstacles though I think case #152 is the closest I can get. Let me know if anything I can help on.

On Mon, Feb 13, 2017 at 4:42 PM, Irmen de Jong notifications@github.com wrote:

Thanks, I'll have a look.

In the meantime, I noticed that jython has its own share of classes representing various Python types. See http://javadox.com/org.python/ jython-standalone/2.7-rc1/org/python/core/PyObject.html and its derived classes I suspect 2 things:

• the problem is specific to Serpent, not Pyro or Pyrolite. If I have proof this issue should reallly be moved to the serpent project ( https://github.com/irmen/Serpent https://github.com/irmen/Serpent )
• the problem is because Serpent doesn't know about these Jython classes, and when calling the serializer from Jython code, it somehow gets stuck in a loop.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/irmen/Pyro4/issues/149#issuecomment-279531959, or mute the thread https://github.com/notifications/unsubscribe-auth/ACnBWRhC2zXLa0ZwTs0MUAyIKlBmBfCkks5rcM5hgaJpZM4L9w8S .

[irmen]

question: is there any need to use Jython at all? Can't you use Python itself for client as well?

[zxpatric]

Unfortunately device has only Java API complete so far...

On Feb 13, 2017 5:42 PM, "Irmen de Jong" notifications@github.com wrote:

question: is there any need to use Jython at all? Can't you use Python itself for client as well?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/irmen/Pyro4/issues/149#issuecomment-279547605, or mute the thread https://github.com/notifications/unsubscribe-auth/ACnBWdrI0uP8YzdGvHr-iKwxfd_pKDl7ks5rcNxLgaJpZM4L9w8S .

[irmen]

I see, but have you tried to use plain Java plus the native Pyrolite java api, instead of jython? That is a lot faster. See my comment above on some tips to use it, have you tried that? https://github.com/irmen/Pyro4/issues/149#issuecomment-279141493

[irmen]

Please try a new Serpent jar build. (no need to upgrade pyrolite). Hopefully it fixes the issue with the bytearray. Please let me know

[zxpatric]

OK. But I think you are right that if there are possibly other problems with Jython, why not just use pure java instead of Jython? They can both be called (Popen) from Python 3.X...

Switching the codes and will try both tomorrow.

On Mon, Feb 13, 2017 at 9:09 PM, Irmen de Jong notifications@github.com wrote:

Please try a new Serpent jar build. (no need to upgrade pyrolite). Hopefully it fixes the issue with the bytearray. Please let me know

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/irmen/Pyro4/issues/149#issuecomment-279586264, or mute the thread https://github.com/notifications/unsubscribe-auth/ACnBWdjNoEzESyEOh1OWeKu2BguxcBn0ks5rcQzSgaJpZM4L9w8S .

[zxpatric]

Yes. The "maximum recursion depth exceeded" has been fixed after upgrading the serpent package. I have also moved my codes away from using Jython.