jython-users — Discussion of and questions about using Jython

Re: [Jython-users] jythonc error?

[John M. <joh...@ya...>]

Whew!  It works now.

I added jython.jar to the CLASSPATH.  That makes sense based on the
error msg below.  I'm suprised though since I don't remember needing to
do this before.  For now I'm thrilled just to have the jar.

Not only do I have a jar file but it also doesn't require jython.jar to
run.  That's great!  Everything is in one jar file.


--- John Mudd <joh...@ya...> wrote:
> I've come up with a list of steps before trying jythonc.
> 	1) purge the cache directory
> 	2) remove all class files
> 	3) remove the jpywork directories & files
> 	4) touch all of the .py files.
> Following these steps I have a different error.  I seem to be closer.
> 
> Any suggestions?
> 
> Compiling .java to .class...
> Compiling with args:['/opt/java1.3/bin/javac', ...........very
> long...]
> 1  ./jpywork/com/fiserv/ist/titan/repository/disk/whrandom.java:3:
> package org.python.core does not exist
> import org.python.core.*;
> ^
> ./jpywork/com/fiserv/ist/titan/repository/disk/whrandom.java:237:
> cannot resolve symbol
> symbol  : class PyObject  
> location: class com.fiserv.ist.titan.repository.disk.whrandom
>     public static void moduleDictInit(PyObject dict) {
>                                       ^
> ./jpywork/com/fiserv/ist/titan/repository/disk/whrandom.java:9:
> cannot
> resolve symbol
> symbol  : class PyFunctionTable  
> location: class
> com.fiserv.ist.titan.repository.disk.whrandom._PyInner
>     public static class _PyInner extends PyFunctionTable implements
> PyRunnable {                                         ^
> 
> 
> __________________________________________________
> Do You Yahoo!?
> Get email at your own domain with Yahoo! Mail. 
> http://personal.mail.yahoo.com/
> 
> _______________________________________________
> Jython-users mailing list
> Jyt...@li...
> http://lists.sourceforge.net/lists/listinfo/jython-users


__________________________________________________
Do You Yahoo!?
Get email at your own domain with Yahoo! Mail. 
http://personal.mail.yahoo.com/

[Jython-users] jythonc error?

[John M. <joh...@ya...>]

I've come up with a list of steps before trying jythonc.
	1) purge the cache directory
	2) remove all class files
	3) remove the jpywork directories & files
	4) touch all of the .py files.
Following these steps I have a different error.  I seem to be closer. 
Any suggestions?

Compiling .java to .class...
Compiling with args:['/opt/java1.3/bin/javac', ...........very long...]
1  ./jpywork/com/fiserv/ist/titan/repository/disk/whrandom.java:3:
package org.python.core does not exist
import org.python.core.*;
^
./jpywork/com/fiserv/ist/titan/repository/disk/whrandom.java:237:
cannot resolve symbol
symbol  : class PyObject  
location: class com.fiserv.ist.titan.repository.disk.whrandom
    public static void moduleDictInit(PyObject dict) {
                                      ^
./jpywork/com/fiserv/ist/titan/repository/disk/whrandom.java:9: cannot
resolve symbol
symbol  : class PyFunctionTable  
location: class com.fiserv.ist.titan.repository.disk.whrandom._PyInner
    public static class _PyInner extends PyFunctionTable implements
PyRunnable {                                         ^


__________________________________________________
Do You Yahoo!?
Get email at your own domain with Yahoo! Mail. 
http://personal.mail.yahoo.com/

Re: [Jython-users] More on problem with subclassing Java class

[<bc...@wo...>]

[Kent Johnson}

>At 3:42 AM -0700 4/4/01, Finn Bock wrote:
>>Keep in mind that python class can't overload a method name. Instead it
>>can define different numbers of arguments to the same method.
>
>So, if a jython class overrides a method from a Java superclass,
>- The jython class can only have one implementation of the method (no 
>overloaded methods in Python)
>- Any call to that method name, regardless of actual arguments, will 
>be sent to the jython method
>
>Is that right? 

Yes.

>Would it be possible for the runtime to look at the 
>number of arguments, at least, 

I suppose jython could for the very common case of methods. Looking at
argument count (or better looking at argument types as well as count)
just strikes me as unpythonic. Keep in mind that the method could be any
kind of callable and checking the argument count/types of all callables
is not possible.

>and look up the inheritance hierarchy to find a method match?

This is what the overriden "showData" methods look like. It attempts to
find a python method called "showData" and then call it if found.

        public void showData(test.Data arg0) {
            PyObject inst = Py.jfindattr(this, "showData");
            if (inst != null) inst._jcall(new Object[] {arg0});
            else super.showData(arg0);
        }
        public void showData(java.lang.String arg0, int arg1) {
            PyObject inst = Py.jfindattr(this, "showData");
            if (inst != null) inst._jcall(new Object[] 
                      {arg0, Py.newInteger(arg1)});
            else super.showData(arg0, arg1);
        }

Fitting in an arg count/type match into this simple dictionary lookup is
difficult.

regards,
finn

Fwd: RE: [Jython-users] overloading methods

[Glen S. <gl...@en...>]

----------  Forwarded Message  ----------
Subject: RE: [Jython-users] overloading methods
Date: Wed, 4 Apr 2001 11:44:26 -0700
From: Glen Starchman <gl...@en...>


On Wed, 04 Apr 2001, Brian Zimmer wrote:
> 
> The correct way to do this is:
> 
> def mything(a, b, c=None):
>  """ do something, like check if 'c' is not None """
>  if c:
>   print a, b, c
>  else:
>   print a, b

But that begins to get complicated if you need to "overload" with several
different types of parameters, doesn't it? 

I am not a big fan of overloading, but sometimes it is a necessary evil. 

I found an interesting approach with Lua for overloading (bear in mind, I don't
know Lua very well and this is just my interpretation of how it works):

function fn_iii(a,b,c)  #takes three integers
function fn_is(a,b)	 #takes an integer and a string 
function fn_ii(a,b)	#takes two integers

to call the function, just call fn() with the correct number of params and the
runtime will analyze the params and call the appropriate function. Obviously
the problem comes into play when a param is not a built-in type (eg...some
random object). Not an exactly elegant solution, but it appears to work.

-- 
c=(0x47, 0x6c, 0x65, 0x6e, 0x40, 0x65, 
 0x6e, 0x61, 0x62, 0x6c, 0x65, 0x64, 
 0x76, 0x65, 0x6e, 0x74, 0x75, 0x72, 
 0x65, 0x73, 0x2e, 0x63, 0x6f, 0x6d)
for z in map(eval("lambda _y:_y "),
range(len(c))):print repr((map(
lambda q:chr(c[z]),c))).join(
chr(c[z])).lower(),
-------------------------------------------------------

-- 
c=(0x47, 0x6c, 0x65, 0x6e, 0x40, 0x65, 
 0x6e, 0x61, 0x62, 0x6c, 0x65, 0x64, 
 0x76, 0x65, 0x6e, 0x74, 0x75, 0x72, 
 0x65, 0x73, 0x2e, 0x63, 0x6f, 0x6d)
for z in map(eval("lambda _y:-_y "),
range(len(c))):print repr((map(
lambda q:chr(c[z]),c))).join(
chr(c[z])).lower(),

Re: [Jython-users] overloading methods

[Tim H. <tim...@ie...>]

RE: [Jython-users] overloading methods
  ----- Original Message -----=20
  From: Brian Zimmer=20
  To: Tom Whittaker ; jyt...@li...=20
  Sent: Wednesday, April 04, 2001 10:35 AM
  Subject: RE: [Jython-users] overloading methods


  The correct way to do this is:=20

  def mything(a, b, c=3DNone):=20
   """ do something, like check if 'c' is not None """=20
   if c:=20
    print a, b, c=20
   else:=20
    print a, b=20

You may want to be careful testing for c being true rather than testing =
None specifically since:

    mything("spam", "eggs", []) # or () or "" or 0 or some instances.

will take the second branch which may not be want you want. It's =
probably better to use:

def mything(a, b, c=3DNone):=20
 """ do something, like check if 'c' is not None """=20
 if c is not None:=20
  print a, b, c=20
 else:=20
  print a, b=20



Pedantically yours,

-tim

RE: [Jython-users] overloading methods

[Brian Z. <bz...@ig...>]

The correct way to do this is:

def mything(a, b, c=3DNone):
 """ do something, like check if 'c' is not None """
 if c:
  print a, b, c
 else:
  print a, b

then when you call mything with either 2 or 3 arguments the runtime
figures out if 'c' should be passed, if not, it uses the default value,
in this case None.

mything(1, 2)
 - and -
mything(1, 2, 3)

will both work.

brian


-----Original Message-----
From: jyt...@li...
[mailto:jyt...@li...]On Behalf Of Tom
Whittaker
Sent: Wednesday, April 04, 2001 12:22 PM
To: jyt...@li...
Subject: [Jython-users] overloading methods


Please pardon my basic "newbie" question...I've not been able to find
the
answer to this in the tutorial, in O'Reilly, or thru web searching.

I am making a .py file with a bunch of methods.  I'd like to overload
(at
least) one of them, like:

def mything(a,b,c)
  print a,b,c
def mything(a,b)
  print a,b

What I've found is that only the last one in the .py file seems to be
called
every time.  If I try to reference the above with:

file.mything(1,2,3)

I get an error message like: "not enough arguments; expected 3 got 2".

I've overcome this by using a single method like:

def mything(*a)

and looking at the len(a).  Since Jython does such a great job when
trying to
figure out which overloaded method to use in a Java class, I thought I
must be
missing something. Is there a better/preferred way to accomplish this?

Thanks ahead for your advice & guidance,

tom
--=20
Tom Whittaker (to...@ss...)
University of Wisconsin-Madison
Space Science and Engineering Center
Phone/VoiceMail: 608/262-2759
Fax: 608/262-5974

_______________________________________________
Jython-users mailing list
Jyt...@li...
http://lists.sourceforge.net/lists/listinfo/jython-users

[Jython-users] overloading methods

[Tom W. <to...@ss...>]

Please pardon my basic "newbie" question...I've not been able to find the
answer to this in the tutorial, in O'Reilly, or thru web searching.

I am making a .py file with a bunch of methods.  I'd like to overload (at
least) one of them, like:

def mything(a,b,c)
  print a,b,c
def mything(a,b)
  print a,b

What I've found is that only the last one in the .py file seems to be called
every time.  If I try to reference the above with:

file.mything(1,2,3)

I get an error message like: "not enough arguments; expected 3 got 2".

I've overcome this by using a single method like:

def mything(*a)

and looking at the len(a).  Since Jython does such a great job when trying to
figure out which overloaded method to use in a Java class, I thought I must be
missing something. Is there a better/preferred way to accomplish this?

Thanks ahead for your advice & guidance,

tom
-- 
Tom Whittaker (to...@ss...)
University of Wisconsin-Madison
Space Science and Engineering Center
Phone/VoiceMail: 608/262-2759
Fax: 608/262-5974

[Jython-users] jythonc error, too many open files?

[John M. <joh...@ya...>]

I decided to build a jar file and use the jython.jar with it.  Now I
get errors just trying to build the jar as follows.  


$ make
jythonc --all --package com.ry.disk \
        --jar ../../../../../../../classes/ry.jar hield.py nnMgr.py
skServer.py skUsageThread.py leMgr.py rtSession.py stMisc.py ckMgr.py
sgQueue.py skSet.py bjFile.py bjIndex.py adSession.py spy.py
processing hield
processing getopt
processing string
$ make
jythonc --all --package com.ry.disk \
        --jar ../../../../../../../classes/ry.jar hield.py nnMgr.py
skServer.py
skUsageThread.py leMgr.py rtSession.py stMisc.py ckMgr.py sgQueue.py
skSet.py bjFile.py bjIndex.py adSession.py spy.py
processing hield
processing getopt
processing string
processing stMisc
processing traceback
processing linecache
processing stat
processing javaos
processing javapath
processing nnMgr
processing skServer
processing bjDiskSet
processing glob
Traceback (innermost last):
  File "/home/mudd/jython-2.1a1/Tools/jythonc/jythonc.py", line 5, in ?
  File "/home/mudd/jython-2.1a1/Tools/jythonc/main.py", line 300, in
main
  File "/home/mudd/jython-2.1a1/Tools/jythonc/main.py", line 219, in
doCompile
  File "/home/mudd/jython-2.1a1/Tools/jythonc/compile.py", line 215, in
compilefile
  File "/home/mudd/jython-2.1a1/Tools/jythonc/compile.py", line 231, in
compile
  File "/home/mudd/jython-2.1a1/Tools/jythonc/compile.py", line 296, in
addDependencies
  File "/home/mudd/jython-2.1a1/Tools/jythonc/compile.py", line 215, in
compilefile
  File "/home/mudd/jython-2.1a1/Tools/jythonc/compile.py", line 231, in
compile
  File "/home/mudd/jython-2.1a1/Tools/jythonc/compile.py", line 296, in
addDependencies
  File "/home/mudd/jython-2.1a1/Tools/jythonc/compile.py", line 215, in
compilefile
  File "/home/mudd/jython-2.1a1/Tools/jythonc/compile.py", line 24, in
getdata
IOError: File not found - /new1/mudd/jython-2.1a1/Lib/glob.py (Too many
open files)
gmake2: *** [../../../../../../../classes/ry.jar] Error 255
$




FYI:  The file "glob.py" does seem to be in place.  I guess the "too
many open files" problem prevented accessing this file.

$ pwd
/new1/mudd/jython-2.1a1/Lib
$ ls -l glob.py
-rw-r--r--   1 mudd       general       1197 Apr  4 11:49 glob.py
$ 



The reason that the home directory used above is /home/mudd in some
cases and /new1/mudd in some is because my home directory is linked
from /new1 to /home.

$ cd /home
$ ls -l mudd
lrwxr-xr-x   1 mudd   general      10 Sep 21  2000 mudd -> /new1/mudd
$



__________________________________________________
Do You Yahoo!?
Get email at your own domain with Yahoo! Mail. 
http://personal.mail.yahoo.com/

RE: [Jython-users] How can I iterate over a Map?

[DITTGEN P. <pie...@cr...>]

> 
>    for aKey in myMap.getKeys():
> 

Correct is:
	for aKey in myMap.keys():

> or
>    for (aKey, aValue) in myMap.entries()
> 

Correct is:
	for (aKey, aValue) in myMap.items():


Regards
Pierre

[Jython-users] How can I iterate over a Map?

[Kent J. <kjo...@tr...>]

Is there a way for a jython program to iterate over the data in a 
Collection using a for loop? If I try

   for aKey in myMap.getKeys():

or
   for (aKey, aValue) in myMap.entries()

I get an AttributeError on __getitem__

Do I have to explicitly use Iterators to make this work?

P.S. I am using Java 1.1 collections if it matters

Kent Johnson                   Transparent Language, Inc.
kjo...@tr...       http://www.transparent.com

Re: [Jython-users] A big problem for me!

[D-Man <ds...@ri...>]

On Wed, Apr 04, 2001 at 11:01:51PM -0400, cindy wrote:
| > The following code worked for me :
| >
| > from java.awt import Component , AWTEvent
| >
| > class C( Component ) :
| >     def __init__( self ) :
| >             self.super__enableEvents( AWTEvent.WINDOW_EVENT_MASK )
| 
| Ok. So this means that ever time I wanted to use this method, "enableEvents()" I
| have to
| inherit Component. But lets say that I would like to extend some other
| component, like
| JFrame. Then if I want to use enableEvents() in the same class that I want to
| extend
| JFrame, I would have to inherit Componet and do a composit on JFrame. Is this
| correct?

Not quite.  If you look at the top of the documenation for JFrame you
will see

Class JFrame

java.lang.Object 
  |
  +--java.awt.Component 
        |
        +--java.awt.Container 
              |
              +--java.awt.Window 
                    |
                    +--java.awt.Frame 
                          |
                          +--javax.swing.JFrame 

JFrame already inherits from java.awt.Component.  The term "is-a" is
often used with inheritance.  A JFrame object is-a Component object.
If you inherit from JFrame, you are indirectly inheriting from
Component.

If you look farther down on the documentation page you will see a list
of  "Methods inherited from Frame" and "Methods inherited from Window"
and ... and "Methods inherited from Component".  Since the methods are
listed there, they exist in objects of the type "JFrame".

FYI, the documentation (for JDK1.2.2) is at

http://java.sun.com/products/jdk/1.2/docs/api/index.html

while the JFrame page is

http://java.sun.com/products/jdk/1.2/docs/api/javax/swing/JFrame.html


-D

RE: [Jython-users] A big problem for me!

[Brian Z. <bz...@ig...>]

( I'm perfectly comfortable being corrected when I'm wrong. )

Point well made, I'll be more careful next time ;)

D-Man is indeed correct and offered a better explanation of object vs.
class than I did.

Sorry for any confusion and thanks for the clarification.

brian


-----Original Message-----
From: jyt...@li...
[mailto:jyt...@li...]On Behalf Of D-Man
Sent: Wednesday, April 04, 2001 9:30 AM
To: Jyt...@li...
Subject: Re: [Jython-users] A big problem for me!



( I don't mean to be condescending at all, but I want to make sure this
  doesn't cause some confusion.  )

An object can't be abstract, only classes can.  Objects also can't be
subclassed, only classes can be subclassed.  Also, 'self' hasn't been
"called", it was only used to access an attribute. =20

(calling it would look like

self() or self( <some arguments here> ) and would require defining the
function  __call_

)



I think Brian knows this, but mixed a couple of terms around in
writing this.  Other than the wording issues, everything Brian said is
accurate and (hopefully) helpful to you.

-D


On Tue, Apr 03, 2001 at 03:05:05PM -0500, Brian Zimmer wrote:
|=20
|    Wayne,
|   =20
|    You can call "self" on an abstract object, you just can't
instantiate
|    an abstract object.  An abstract object has all the state and
|    behaviour of a normal, concrete object it just can't live on it's
|    own.  It needs to be subclassed and made abstract.  For a good
example
|    of this, check out the java.io.* packages, in particular, either
the
|    Reader/Writer or InputStream/OutputStream classes and their
|    subclasses.  Look at FileReader and you'll notice it is an
|    implementation of the abstract Reader class, all FileReader
provides
|    to Reader is how to get the initial data, in this case a File.
|   =20
|    Python really doesn't have the notion of an abstract class, in
general
|    an abstract class or interface is implemented by defining a bunch
of
|    methods which raise Exceptions.
|   =20
|    So in your case, you can't instantiate a Component, but you can get
|    subclasses of Component and then call methods on them.
|   =20
|    hope this helps,
|   =20
|    brian

_______________________________________________
Jython-users mailing list
Jyt...@li...
http://lists.sourceforge.net/lists/listinfo/jython-users

Re: [Jython-users] A big problem for me!

[cindy <inf...@pi...>]

> The following code worked for me :
>
> from java.awt import Component , AWTEvent
>
> class C( Component ) :
>     def __init__( self ) :
>             self.super__enableEvents( AWTEvent.WINDOW_EVENT_MASK )

Ok. So this means that ever time I wanted to use this method, "enableEvents()" I
have to
inherit Component. But lets say that I would like to extend some other
component, like
JFrame. Then if I want to use enableEvents() in the same class that I want to
extend
JFrame, I would have to inherit Componet and do a composit on JFrame. Is this
correct?
Thanks.
Wayne

Re: [Jython-users] A big problem for me!

[D-Man <ds...@ri...>]

( I don't mean to be condescending at all, but I want to make sure this
  doesn't cause some confusion.  )

An object can't be abstract, only classes can.  Objects also can't be
subclassed, only classes can be subclassed.  Also, 'self' hasn't been
"called", it was only used to access an attribute.  

(calling it would look like

self() or self( <some arguments here> ) and would require defining the
function  __call_

)



I think Brian knows this, but mixed a couple of terms around in
writing this.  Other than the wording issues, everything Brian said is
accurate and (hopefully) helpful to you.

-D


On Tue, Apr 03, 2001 at 03:05:05PM -0500, Brian Zimmer wrote:
| 
|    Wayne,
|    
|    You can call "self" on an abstract object, you just can't instantiate
|    an abstract object.  An abstract object has all the state and
|    behaviour of a normal, concrete object it just can't live on it's
|    own.  It needs to be subclassed and made abstract.  For a good example
|    of this, check out the java.io.* packages, in particular, either the
|    Reader/Writer or InputStream/OutputStream classes and their
|    subclasses.  Look at FileReader and you'll notice it is an
|    implementation of the abstract Reader class, all FileReader provides
|    to Reader is how to get the initial data, in this case a File.
|    
|    Python really doesn't have the notion of an abstract class, in general
|    an abstract class or interface is implemented by defining a bunch of
|    methods which raise Exceptions.
|    
|    So in your case, you can't instantiate a Component, but you can get
|    subclasses of Component and then call methods on them.
|    
|    hope this helps,
|    
|    brian

Re: [Jython-users] More on problem with subclassing Java class

[Kent J. <kjo...@tr...>]

At 3:42 AM -0700 4/4/01, Finn Bock wrote:
>Keep in mind that python class can't overload a method name. Instead it
>can define different numbers of arguments to the same method.

So, if a jython class overrides a method from a Java superclass,
- The jython class can only have one implementation of the method (no 
overloaded methods in Python)
- Any call to that method name, regardless of actual arguments, will 
be sent to the jython method

Is that right? Would it be possible for the runtime to look at the 
number of arguments, at least, and look up the inheritance hierarchy 
to find a method match?

Kent

Re: [Jython-users] A big problem for me!

[<bc...@wo...>]

[D-Man]

>So with that knowledge, a class that inherits from java.awt.Component
>should have a member 'enableEvents'.  The problem here is it is
>'protected'.  In Java that means that only subclasses and classes in
>the same package can access the function.  Other classes can't call
>it.  Apparently jython does some name mangling on protected members by
>prefixing "super__" to it.  (I certainly hope it is mentioned in the
>documentation, but like a <ahem> genius^H^H^H^H^H^Hidiot I haven't
>checked ;-))

It is mentioned in the section "Calling Methods in Your Superclass"

   http://www.jython.org/docs/subclassing.html

regards,
finn

Re: [Jython-users] More on problem with subclassing Java class

[<bc...@wo...>]

[Kent Johnson]

>Here is a complete test case. When using the first showData in 
>MyCallback, it works. With the second callback (as below) it gives
>
>Traceback (innermost last):
>   File "<console>", line 1, in ?
>   File "./Project/Test.py", line 12, in ?
>TypeError: showData() takes at least 3 arguments (2 given)

Keep in mind that python class can't overload a method name. Instead it
can define different numbers of arguments to the same method. Since the
python "showData" method will subclass both java showData methods, you
will have to figure out which invocation that occured. Something like:

	def showData(self, inName, inValue=None):
		if inValue == None:
			# The "inName" arg is actually a Data instance
			Test.Callback.showData(self, inName)
		else:
			print inName
			print inValue

As you can see, implementing an overloaded java method in python is not
a situation where jython shines.

regards,
finn

Re: [Jython-users] chr/unichr question

[<bc...@wo...>]

On Mon, 2 Apr 2001 16:40:40 -0500 (CDT), you wrote:

>Hello all,
>
>I have a question about the builtins chr().  The Python Library
>Reference for 2.1b2 says it "Return a string of one character whose ASCII
>code is the integer i, e.g., chr(97) returns the string 'a'. This is the
>inverse of ord(). The argument must be in the range [0..255], inclusive;
>ValueError will be raised if i is outside that range."
>
>When interpreting this for Jython, the obvious catches are that it
>specifies "range [0..255]", and raises ValueError for i outsite this
>range.
>
>This doesn't really seem to apply considering how Jython only has the
>2-byte character string.  Different behavior for Jython's builtin chr()
>makes sense.  However, is this difference in chr() official? 

I think it is. As long as CPython strings are 8-bits, I expect this
difference to stand.

>I didn't
>see a comment on http://jython.sourceforge.net/docs/differences.html about
>this,

I'll add an item about the chr() difference.

>and I wanted to be sure before committing it to paper for a Jython
>publication (and risking a ValueError being raised in Jython2.1final or
>2.2).

It will not throw ValueErrors for 2.1 final.

The string/unicode split in CPython is not pleasing and will perhaps
change in later versions of CPython. If ever that happens, Jython may
change too, but only in order to reduce the differences between the two
implementations.

regards,
finn

Re: [Jython-users] Generated java classes dependency

[<bc...@wo...>]

[Pierre Dittgen]

>I'm quite new to jython, i find it very powerful as a script language.
>But at this time, I have a problem to make an independant jar from a jython
>script.
>I use Jython 2.0, with SUN JDK1.3.0 on Win2000 Pro.
>
>My script is called dir.py and contains:
>== begin ======
>import os
>
>for file in os.listdir('.'):
>	print file
>== end ========
>
>It works well as I use the command line:
>jython dir.py
>(it shows me the list of the files of the current directory)
>
>I want to build a jar file from this script.
>I used this command :
>jythonc -a -j dir.jar dir.py
>It created a jar file (called dir.jar) which contains normally all the
>needed jython libraries
>needed to make it run.
>But as I want to run it, using the command
>java -jar dir.jar
>I got an exception :
>== begin ==============
>
>java.lang.ExceptionInInitializerError: java.lang.NullPointerException
>        at org.python.modules.os.<clinit>(os.java:11)

It is a bug in the os.java module when running without an unset
sys.prefix. This normally happens for jythonc compiled applications.

As a workaround, you can set sys.prefix before importing the "os"
module.

  import sys; sys.prefix = "."
  import os
  ...

or you can upgrade to Jython-2.1a1 where the bug is fixed.

>I solved the problem giving the root dir of jython installation as a system
>property:
>java -Dpython.home=c:\dev\jython-2.0 -jar dir.jar
>
>But it annoys me...
>Is the generated jar (dir.jar) still dependant of the jython installation ?

No.

regards,
finn

RE: [Jython-users] Generated java classes dependency

[DITTGEN P. <pie...@cr...>]

> 
> Now this means that your app will have to ship some python 
> files in the JAR.
 

It annoyed me quite much and I looked further in the code (in
org.python.modules.os).
Before the line that causes me trouble, i.e. this:

    public static String __file__ = 
          Py.getSystemState().prefix.toString() + "/Lib/javaos.py";

there is a comment saying:

    // An ugly hack, but it keeps the site.py from CPython2.0 happy

So, apparently, it's just a hack, that's not used at runtime...
BUT, this hack requires the 'python.home' system property to be set..
So, i just tried to declare it and it works!

my example runs now with:
java -Dpython.home= -jar dir.jar
(dir.jar was created using --all option to include all jython libs useful
for my app)

It won't be very difficult to set this empty property at deployment time,
in any case easier than installing a jython distrib on the host machine.

Thanks for the help
Pierre

Re: [Jython-users] Generated java classes dependency

[Ben H. <Ben...@fi...>]

----- Original Message -----
From: "DITTGEN Pierre" <pie...@cr...>

> The code at line 11 in os.java is:
>
>     public static String __file__ =
>           Py.getSystemState().prefix.toString() + "/Lib/javaos.py";
>
> Is getSystemState() method returning a null value?
> Is prefix attribute null?

Definitely one of the above. Clearly the above line sets a file path to
javaos.py, presumably relative to a base JPython directory set
by -Dpython.home=c:\dev\jython-2.0. When the property isnt set Jython fails.

>
> But why, is there a reference to javaos.py?
> Do I need the jython libs as jython scripts (i.e. not as class files)
> to make my prog run?

My understanding is that Jython consists of both a java core interpreter and
multiple modules written in python. This is a good thing: it enables Jython
to provide that same Python infrastructure that CPython presents, and it
tests/proves that the core interpreter is python complaint.

Now this means that your app will have to ship some python files in the JAR.
They are too big so this shouldnt be a problem. You just need to devise a
way to set 'python.home' in your deployment. What you want is an absolute
URL to the JAR you are packed in, on the host machine, because the location
of .py files is fixed and relative to that.

I dont know if anyone has already solved this one finally for Jython. What I
would do is use the *useful* Classloader.getSystemResource(String resource)
method. This can take the relative name of a file, which it will search for
in all the roots of its classpath, and return you an absolute URL locating
the file.

Additionally, you could look at Bruce Eckels "Thinking in Patterns", which
is free online and has some useful advanced Jython examples in the
Interpreter chapter.

Regards
Ben

[Jython-users] jythonc, --all jar requires jython.jar to run.

[John M. <joh...@ya...>]

I switched to the latest jython but I still get an error when trying to
run a jythonc generated jar file unless I put the jython.jar in the
CLASSPATH.  I used the "--all" option when running jythonc.  Any
suggestions.

In this case yyy is a valid attribute for a super class that was
inherited by xxx.  This works when run as a jython script.  It only
works as a jythonc jar file if I put jython.jar in the CLASSPATH.

	AttributeError: instance of 'xxx' has no attribute 'yyy'




$ jython
Jython 2.1a1 on javaVM-1.3.0.01 (JIT: null)
Type "copyright", "credits" or "license" for more information.
>>> 


__________________________________________________
Do You Yahoo!?
Get email at your own domain with Yahoo! Mail. 
http://personal.mail.yahoo.com/

Re: [Jython-users] A big problem for me!

[cindy <inf...@pi...>]

Thanks for your time! I went the jython site and found the documentation.
the documentation has more meaning for me now, now that I had the
problem.
Wayne

> prefixing "super__" to it.  (I certainly hope it is mentioned in the
> documentation, but like a <ahem> genius^H^H^H^H^H^Hidiot I haven't
> checked ;-))
>

RE: [Jython-users] A big problem for me!

[Brian Z. <bz...@ig...>]

Wayne,

You can call "self" on an abstract object, you just can't instantiate an
abstract object.  An abstract object has all the state and behaviour of
a normal, concrete object it just can't live on it's own.  It needs to
be subclassed and made abstract.  For a good example of this, check out
the java.io.* packages, in particular, either the Reader/Writer or
InputStream/OutputStream classes and their subclasses.  Look at
FileReader and you'll notice it is an implementation of the abstract
Reader class, all FileReader provides to Reader is how to get the
initial data, in this case a File.

Python really doesn't have the notion of an abstract class, in general
an abstract class or interface is implemented by defining a bunch of
methods which raise Exceptions.

So in your case, you can't instantiate a Component, but you can get
subclasses of Component and then call methods on them.

hope this helps,

brian


-----Original Message-----
From: jyt...@li...
[mailto:jyt...@li...]On Behalf Of cindy
Sent: Wednesday, April 04, 2001 3:48 AM
To: D-Man
Cc: Jyt...@li...
Subject: Re: [Jython-users] A big problem for me!


D-Man wrote:

> class Foo :
>     def func( self ) :
>         print "func was called"
>
>     def runme( self ) :
>         func() # this won't work as you've seen in your code
>
>     def runme2( self ) :
>         self.func() # this will work as expected

This I undertstand, but I don't see how "self" can be used on a
class that is inherited and that is abstract, such as Component.
When I do
"self.enableEvents(java.awt.event.WindowEvent.WINDOW_EVENT_MASK)"
I get an AttribueError on enableEvents, which I would expect.
Thanks.
Wayne


_______________________________________________
Jython-users mailing list
Jyt...@li...
http://lists.sourceforge.net/lists/listinfo/jython-users

Re: [Jython-users] A big problem for me!

[cindy <inf...@pi...>]

D-Man wrote:

> class Foo :
>     def func( self ) :
>         print "func was called"
>
>     def runme( self ) :
>         func() # this won't work as you've seen in your code
>
>     def runme2( self ) :
>         self.func() # this will work as expected

This I undertstand, but I don't see how "self" can be used on a
class that is inherited and that is abstract, such as Component.
When I do "self.enableEvents(java.awt.event.WindowEvent.WINDOW_EVENT_MASK)"
I get an AttribueError on enableEvents, which I would expect.
Thanks.
Wayne