SWIG

ERMX Jumps Into Huge Advanced Nitride Super Conductors!

EntreMetrix Inc. (ERMX)
$0.16

ERMX has just announced it involvement in a joint venture to manufacture
Advanced Nitride Devices. These have Military application as well as
Energy Super Conductors. This takes ERMX to a whole new level. Google
"Nitride Devices" and get on ERMX first thing Monday!

xcu file contains information that the Package Manager uses to add the
templates to the appropriate location. "We've been on Linux for years,
but I'm still amazed. By clicking the Sign up button, you indicate that
you agree to our Terms and Conditions and have read and understand our
Privacy Policy. From the very beginning, users could add new features to
the office suite by installing so-called UNO packages. It was actually
BSD merged with System V at one point, so there were a lot of people who
felt that we should be using BSD as a license. JOST is in relatively
early development, and it still shows some rough edges.

GM: What did that involve?

These scripts are my invaluable helpmates, and in lieu of LASH
compatibility they are wonderful time savers.

Because the Lorem ipsum generator contains only one command, it makes
sense to install it under the Tools menu.

So, although we'd written the code, we didn't actually have the right to
disclose it, because to do so would have disclosed the trade secrets.

org also introduces the new .

You easily can turn your templates, autotext entries, gallery art and
macros into extensions that can be installed with a couple of clicks.

It accepted that the manuscripts were looted, but argued that the
university had acquired them in "good faith". The university has had
concerns that simply to hand them back would create a precedent that
would cause problems for other institutions, such as the British Museum
and British Library. " Chan says DreamWorks Animation has a good
relationship with Red Hat, working closely to ensure that HP
workstations work with Red Hat Linux for DreamWorks.

Edinburgh University, despite a direct plea from the government in Addis
Ababa, has refused to hand over four manuscripts with likely links to
the troops' invasion.

We saw Mozilla being released, and then we saw the Open Source
Initiative picking up the Debian Social Contract and turning it into the
Open Source Definition.
In my opinion JOST has grown from its original design as a host for
natively-compiled Linux VST plugins, and it is becoming a powerful
modular environment for complex audio synthesis and processing. When I
joined Sun, Sun was right at the forefront of some really quite radical
commercial open-source experiments-one of them, starting OpenOffice.
Their hostility wasn't moderated by a recognition that Java came from
what would now be recognised as an open-source company. Someone else
might have the right to it.

But this does not necessarily mean that is the end of the matter. It
also agreed that opportunities for collaboration, such as providing
loans for exhibitions and participating in projects that enhance access
to and awareness of Ethiopian culture, be explored.

Hello everybody,

I have a problem with the Python interface for my C library because of 
the global interpreter lock (GIL) causing incorrect behavior in the 
multi threaded clients.

After looking around on the Internet I found that if I wrap my C 
functions by Py_BEGIN_ALLOW_THREADS and Py_END_ALLOW_THREADS macros I'll 
get needed behavior. It was proposed to use the %except directive to 
wrap automaticaly my functions with these macros. It looks like follows:

%except(python){
      Py_BEGIN_ALLOW_THREADS
      $function
      Py_END_ALLOW_THREADS
}
which should add to all wrapper functions the needed macros before and 
after the function call. This will release and take the GIL before the C 
function call and regain it just after it.

Currently I found that in swig 1.3 the %except directive is deprecated. 
In any case when I run swig with %except the macros are not in the 
wrapper file. I am looking to the swig 1.3 documentation and I cannot 
find anything which can replace the %except directive. The use of 
%exception do not result in needed wrapping.

Can somebody provide me a hint how I can wrap my C functions with the 
above macros in swig 1.3 ?

Thank you

Robert

Hi,

I was trying to experiment with swig and C#. Here's what I did using a
cygwin shell on Windows.

file: example.c

int fact(int n) {
  if(n==0)
    return 1;
  else
    return n*fact(n-1);
}

$ swig -csharp -module example example.c
$ gcc -c example.c example_wrap.c
$ ld -G example.o example_wrap.o -o example.dll

I then created a visual studio C# project. Which included the following files:

example.cs
examplePINVOKE.cs

I added the example.dll as an item to the project and ensured that the
DLL gets copied to the output directory whenever I build the project.
The project also included a .cs file with the main function

file: Program.cs

using System;
using System.Collections.Generic;
using System.Text;

namespace ConsoleApplication2
{
    class Program
    {
        static void Main(string[] args)
        {
            Console.WriteLine(example.fact(5));
        }
    }
}

Unfortunately when I build and run the project nothing happens. The
program doesn't seem to proceed forward after the call to
"example.fact". Can anyone help me with this?

Also, I couldn't find any documentation which ran through an example
of creating a simple wrapper for C#.

Thanks
Rohit

Navnidhi Sharma wrote:
> How do we map C calls that take void * in C#?
> 
>  
> 
>  
> 
> For example:
> 
>  
> 
> In c:
> 
> void*  foo(void *p);
> 
>  
> 
The default wrapping produces the type wrapper class SWIGTYPE_p_void.
How do you want it wrapped? You could use typemaps to change it to a
HandleRef if you wanted to.

William

Shields, Daniel P.(Technology) wrote:
> I have successfully written a typemap to convert a java.util.Date to
> boost ptime.
> This works fine for functions of the form:
> 
> fun(const boost::posix_time::ptime&)
> 
> I have since added functions with an additional argument:
> 
> fun(const SomeClass&, const boost::posix_time::ptime&)
> 
> The generated java code will not compile, complaining about a missing
> symbol in the moduleJNI.java:
> 
> symbol  : constructor
> SWIGTYPE_p_boost__posix_time__ptime(java.util.Date,boolean)
> 
> Do I need to write an additional typemap for the function with two
> arguments? Is there
> a shorthand way of doing this?
> 
You can use the same typemap no matter how many arguments there are in
your function or where. The type can be anywhere in the argument list
and the type can occur as many times as you wish.

I suspect you have not cleaned your build properly since experimenting
with the typemaps and the problem class is an old one that needs
deleting. If not, post a complete standalone example to demonstrate.

William

Wayne Marquette wrote:
> How do I use swig to call a function which has a pointer to a pointer
>  as an arg.  Most solutions I have seen to do this involve writing
> inline C code?   Can this be done with either typemaps library or the
> pointer library?
> 

Solutions are often target language dependent. The Java.html chapter has
a section on pointer to pointers.

William

Hi all,

I'm using swigwin-1.3.31.zip, Python 2.5.1, and Visual Studio .Net .  I'm working with the Python variables example given with swigwin.  I'm running on Windows XP.  I'm not using cygwin.

I have 3 main issues...1) when I compile I have warnings, 2) When I run the Python variable example I noticed that the float value returned by python was not the same as returned from C, and 3) When I run the entire variable example the example code fails when dealing with pointers.  Help with any of these questions would be great.

#1 Compile Warnings:
1>------ Build started: Project: example, Configuration: Release Win32 ------
1>Performing Custom Build Step
1>Invoking SWIG...
1>c:MainStorageVSWorkspaceexampleexample>C:swigwinswig.exe -python "c:swigwinExamplespythonvariablesexample.i" 
1>Compiling...
1>example_wrap.c
1>........swigwinExamplespythonvariablesexample_wrap.c(346) : warning C4244: 'return' : conversion from '__w64 int' to 'int', possible loss of data
1>........swigwinExamplespythonvariablesexample_wrap.c(1108) : warning C4244: 'initializing' : conversion from 'Py_ssize_t' to 'int', possible loss of data
1>example.c
1>Linking...
1>   Creating library C:swigwinExamplespythonvariables_example.lib and object C:swigwinExamplespythonvariables_example.exp
1>Generating code
1>Finished generating code
1>Embedding manifest...
1>Build log was saved at "file://C:swigwinExamplespythonvariablesBuildLog.htm"
1>example - 0 error(s), 2 warning(s)
========== Build: 1 succeeded, 0 failed, 0 up-to-date, 0 skipped ==========

#2 Float Error...Notice fvar...I turned off the other code that was erroring...
Variables (values printed from Pyth
ivar      = 42
svar      = -31000
lvar      = 65537
uivar     = 123456
usvar     = 61000
ulvar     = 654321
scvar     = -13
ucvar     = 251
fvar      = 3.14159011841
dvar      = 2.1828
cvar      = S
strvar    = Hello World
cstrvar   = Goodbye
name      = Bill

Variables (values printed from C)
ivar      = 42
svar      = -31000
lvar      = 65537
uivar     = 123456
usvar     = 61000
ulvar     = 654321
scvar     = -13
ucvar     = 251
fvar      = 3.14159
dvar      = 2.1828
cvar      = S
strvar    = Hello World
cstrvar   = Goodbye
iptrvar   = 00972B50
name      = Bill
ptptr     = 00972B60 (37, 42)
pt        = (10, 20)
status    = 1

#3 Error...My code breaks at the following line located in the runme.py
print "iptrvar   =", example.cvar.iptrvar

Thanks so much,

Matt

Hi all.

The problem is when I use the file std_complex.i in a function with
complex<double>
*.

For example when I use pointer int *, float * ..., I use cpointer.i.
>>>x=example.intP()
>>>x.assign(9)

With this sintax, I call a function in python to function C++ with pointer
int or float.

But exist a class equal for complex<double> *. or I create a Typemap for
this problem?
Whith serius this typemap more less?

Thanks for your time and sorry for my English.

Hi,

I would have a finished project and want to make a script interface to 
it. It looks like the follows:

semnet.h:
template<class T>
class semNet{
...
};

SemNetTypes.h:
class snParentSemNet : public semNet<SemNet_trait>{
...
};

i want to make class A available in a script language. How do I write 
the binding.i file?
I tried to write a file:

binding.i
%module semnet
%{
#include <utilFESemNetTypes/SemNetTypes.h>
%}

namespace semnet{
        class snParentSemNet{
        public:typedef semNet<SemNet_trait> super;
        snParentSemNet(std::string name) : super();
        std::string getlable();
        relType_cits getRelTypeConstIterator();
        node_t* addNode( node_t* node );
        node_t* removeNode( node_t* node );
  };
}

Do I have to write the class again like I did in my binding.i?

Cheers

Anton

I have identified a bug in the handling of varargs using the %varargs
feature in python when wrapping varargs of non-primitive (but pointer)
type.


Using Swig 1.3.29, and the following file, I get:

%module(docstring=3D=22Demos Bad Varargs=22) testSwig;

%=7B
typedef struct ST =7B=7D *pST;=20
int demoBadVararg(int, ...);
int demoGoodVararg(int, ...);
%=7D

%varargs(1, int extra=3D0) demoGoodVararg;
int demoGoodVararg(int x, ...);

%varargs(1, pST st =3D NULL) demoBadVararg;
int demoBadVararg(int x, ...);


The warning that it prints out is:
testSwig.interface:14: Warning(509): Overloaded
demoBadVararg(int,pST,pST) is shadowed by demoBadVararg(int,pST) at
testSwig.interface:14.


I believe that the generated code is incorrect, as it does not put the
trailing 'NULL' argument when it calls the function, which will crash
certain code that is expecting it.

As an example, the generated code that I got was:



SWIGINTERN PyObject *_wrap_demoBadVararg(PyObject *self, PyObject *args)
=7B
  int argc;
  PyObject *argv=5B3=5D;
  int ii;
 =20
  if (=21PyTuple_Check(args)) SWIG_fail;
  argc =3D PyObject_Length(args);
  for (ii =3D 0; (ii < argc) && (ii < 2); ii++) =7B
    argv=5Bii=5D =3D PyTuple_GET_ITEM(args,ii);
  =7D
  if (argc =3D=3D 1) =7B
    int _v;
    =7B
      int res =3D SWIG_AsVal_int(argv=5B0=5D, NULL);
      _v =3D SWIG_CheckState(res);
    =7D
    if (_v) =7B
      return _wrap_demoBadVararg__SWIG_2(self, args);
    =7D
  =7D
=2E..

Which calls:
SWIGINTERN PyObject *_wrap_demoBadVararg__SWIG_2(PyObject
*SWIGUNUSEDPARM(self), PyObject *args) =7B
  PyObject *resultobj =3D 0;
  int arg1 ;
  int result;
  int val1 ;
  int ecode1 =3D 0 ;
  PyObject * obj0 =3D 0 ;
 =20
  if (=21PyArg_ParseTuple(args,(char *)=22O:demoBadVararg=22,&obj0))
SWIG_fail;
  ecode1 =3D SWIG_AsVal_int(obj0, &val1);
  if (=21SWIG_IsOK(ecode1)) =7B
    SWIG_exception_fail(SWIG_ArgError(ecode1), =22in method '=22
=22demoBadVararg=22 =22', argument =22 =221=22=22 of type '=22 =
=22int=22=22'=22);
  =7D=20
  arg1 =3D static_cast< int >(val1);
  result =3D (int)demoBadVararg(arg1);
  resultobj =3D SWIG_From_int(static_cast< int >(result));
  return resultobj;
fail:
  return NULL;
=7D

I believe that the underlying call should be demoBadVarArg(arg1, NULL).


Is this correct?







- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - =
- - - - - - -

This message is intended only for the personal and confidential use of the =
designated recipient(s) named above.  If you are not the intended =
recipient of this message you are hereby notified that any review, =
dissemination, distribution or copying of this message is strictly =
prohibited.  This communication is for information purposes only and =
should not be regarded as an offer to sell or as a solicitation of an =
offer to buy any financial product, an official confirmation of any =
transaction, or as an official statement of Lehman Brothers.  Email =
transmission cannot be guaranteed to be secure or error-free.  Therefore, =
we do not represent that this information is complete or accurate and it =
should not be relied upon as such.  All information is subject to change =
without notice.

--------
IRS Circular 230 Disclosure:
Please be advised that any discussion of U.S. tax matters contained within =
this communication (including any attachments) is not intended or written =
to be used and cannot be used for the purpose of (i) avoiding U.S. tax =
related penalties or (ii) promoting, marketing or recommending to another =
party any transaction or matter addressed herein.

... snip ...
> ------------------------------
>=20
> Message: 2
> Date: Tue, 26 Jun 2007 14:34:08 +0100
> From: "Shields, Daniel P.(Technology)" <Daniel.P.Shields@...>
> Subject: Re: [Swig-user] C++ -> Java Wrapping templates within
> 	templates..
> To: <swig-user@...>
> Message-ID:
>
<5C590B9888C81E41BB0EA58D146560770159A57E@....
> gs.com>
>=20
> Content-Type: text/plain;	charset=3D"us-ascii"
>=20
>  >-----Original Message-----
> >From: swig-user-bounces@...
> >[mailto:swig-user-bounces@...] On Behalf Of
> >Cooper_JohnD@...
> >Sent: Tuesday, June 26, 2007 2:12 PM
> >To: wsf@...
> >Cc: swig-user@...
> >Subject: Re: [Swig-user] C++ -> Java Wrapping templates within
> >templates..
> >Importance: High
>=20
> ...snip...
>=20
> >
> >> William
> >
> >Thanks William,
> >
> >I have gotten past these issues, using a solution similar to what you
> >suggested.
> >
> >I have one minor issue now that would be nice to solve via the .i if
> >possible.
> >
> >I now generate 2 classes, 1 Vector_ClassId (C++ type) and 2
> >std_Vector_ClassId (Swig parable type) both classes are good but only
> >the std_Vector_ClassId has the vector methods in it. If I go into the
> >java code and simply say that Vector_ClassId extends
std_Vector_ClassId
> >all is good.
> >
> >Is there a way in the .i file that I can specify, for java
generation,
> >that std_Vector_ClassId is the base class for Vector_ClassId
> >and add the
> >super() call into the constructor.
> >
> >Thanks
> >
> >John
> >
>=20
> You should be able to add the java code using:
>=20
> %typemap (javacode) YourCppClassName
> %{
> 	/* put your java code here */
> %}
>=20
> Whether this will work for a default constructor I'm not sure.
> I guess you could always rename this.
>=20
> Daniel.
>=20
>=20

Hi Daniel,

I've tried something like that;

%template(Vector_Attribute) NS1::Vector<Attribute>;
%template(std_Vector_Attribute) std::vector<Attribute>;

// note in C++ NS1::Vector<Attribute> derives from
std::vector<Attribute>
// however I get the following error during the gen on the first
template=20
// line:=20
// Warning(401): Nothing known about base class=20
// 'std::vector< NS1::NS2::SmartPointer< NS1::NS3::IAttribute=20
// >,std::allocator< NS1:: NS2::SmartPointer< NS1:: NS3::IAttribute > >
>'.=20
// Ignored.
//
// Warning(401): Maybe you forgot to instantiate=20
// 'std::vector< NS1:: NS2::SmartPointer< NS1:: NS3::IAttribute=20
// >,std::allocator< NS1:: NS2::SmartPointer< NS1:: NS3::IAttribute > >
>'=20
// using %template.


%typemap(javabase) Vector_Attribute "std_Vector_Attribute"

%typemap(javacode) Vector_Attribute %{
public $javaclassname(long cPtr, boolean cMemoryOwn) {
	super(cPtr, cMemoryOwn);
	swigCMemOwn =3D cMemoryOwn;
	swigCPtr =3D cPtr;
}
%}

But this does nothing. Any ideas?

Thanks

John

Shields, Daniel P.(Technology) wrote:
>  >-----Original Message-----
>> From: swig-user-bounces@... 
>> [mailto:swig-user-bounces@...] On Behalf Of 
>> Cooper_JohnD@...
>> Sent: Tuesday, June 26, 2007 2:12 PM
>> To: wsf@...
>> Cc: swig-user@...
>> Subject: Re: [Swig-user] C++ -> Java Wrapping templates within 
>> templates..
>> Importance: High
> 
> ...snip...
> 
>>> William
>> Thanks William,
>>
>> I have gotten past these issues, using a solution similar to what you
>> suggested.
>>
>> I have one minor issue now that would be nice to solve via the .i if
>> possible.
>>
>> I now generate 2 classes, 1 Vector_ClassId (C++ type) and 2
>> std_Vector_ClassId (Swig parable type) both classes are good but only
>> the std_Vector_ClassId has the vector methods in it. If I go into the
>> java code and simply say that Vector_ClassId extends std_Vector_ClassId
>> all is good.
>>
>> Is there a way in the .i file that I can specify, for java generation,
>> that std_Vector_ClassId is the base class for Vector_ClassId 
>> and add the
>> super() call into the constructor.
>>
>> Thanks
>>
>> John
>>
> 
> You should be able to add the java code using:
> 
> %typemap (javacode) YourCppClassName
> %{
> 	/* put your java code here */
> %}
> 
> Whether this will work for a default constructor I'm not sure.
> I guess you could always rename this.
> 

You can add in a Java base class, by using the javabase typemap (or
javainterfaces typemap for implementing interfaces). You can modify the
constructor code with the javaconstruct typemap. See the default
javaconstruct typemap in java.swg and search the
Examples/test-suite/java_*.i files for examples of typemaps.

William

Cooper_JohnD@... wrote:
> Message: 3
> Date: Wed, 6 Jun 2007 22:13:00 +0000 (UTC)
> From: William S Fulton <wsf@...>
> Subject: Re: [Swig-user] Swig wrapper names Too Long
> To: swig-user@...
> Message-ID: <loom.20070607T000716-796@...>
> Content-Type: text/plain; charset=us-ascii
> 
>  <Cooper_JohnD <at> emc.com> writes:
> 
>>
>>
>>
>>
>> Hello, 
>>
>>  
>>
>> I am using SWIG to wrapper our C++ libraries for Java and
>> all I generally going well. The one thing that this causing a small
> issue is
>> the length of the names of some of SWIGWrappers generated, for
> instance I get
>> one proxy:
>>
>>
> SWIGTYPE_p_CSP__SetTCSP__RT__SmartPointerTCSP__I__IQualifiedName_t_std__
> lessTCSP
> __RT__SmartPointerTCSP__I__IQualifiedName_t_t_std__allocatorTCSP__RT__Sm
> artPoint
> erTCSP__I__IQualifiedName_t_t_t.java
>>  
>>
>> As you can imagine
>> this is a bit unwieldy for use in coding and causes pathing issues on
> some OSs.
>>  
>>
>> My question is,
>> is that some feature of interface language command that I could use to
> either
>> generate shorter names, or have something that works like
> %rename(SWIG*)
>>  
>>
>> I know I could
>> just refactor in eclipse, but that would be a pain to re-factor all of
> them
>> after each gen.
>>
> 
> The class you have given above is not a proxy class. It is a type
> wrapper class
> and its existence usually indicate a poor interface file. You should aim
> at
> removing them by ensuring you have the appropriate type information
> given to
> swig and eg appropriate usage of %template(). You can rename proxy
> classes with
> %rename but not type wrapper classes.
> 
> William
> 
> Thanks William point taken however my rule for the offending wrapper
> mentioned earlier is:
> 
> 
> SETHELPER(Set_Identity,Identity);
> 
> %define SETHELPER(type,intrf)
> %template(type) std::set<intrf>;
> %typemap(in) const type & {
> 	$1 = *(std::set<intrf> **)&($input);
> };
> %apply const type & { type * };
> %enddef
> 
> Which is trying to wrap the following c++ call:
> 
> class QualifiedName {
> public:
> 	QualifiedName(String Name);
> 	~QualifiedName();
> 	const String & getName()   const;
> };	// End class IQualifiedName
> 
> void listAttributes(const Token                 &token,
> 	              const tokenAttributeType_e  type,
>                     const std:vector<Attribute> &filter,
>                     std:set<QualifiedName>      &names) const;
> 
> 
> What am I doing wrong here?
> 

Not sure. Your example seems rather incomplete, how do you deal with
std::set and std::vector and what is String and what is tokenAttribute
etc.... Can you give a standalone example?

William

Nitro wrote:
> I think in your case you can just remove the anonymous union declaration,  
> while leaving it's "body" intact. After that it would look like:
> 
> class AMP_CXX_Export AmpOrderSiteSpecificFields_v2: public AsnType
> {
> public:
>     enum ChoiceIdEnum
>     {
>        nullCid = 0,
>        icapCid = 1,
>        pcxCid = 2
>     };
> 
>     enum ChoiceIdEnum choiceId;
> 
>     /// union causing the problem in SWIG ( hopefully not anymore :) )
>     AmpNull* null;
>     AmpOrderIcapFields_v2* icap;
>     AmpOrderPCXFields* pcx;
> 
> #if META
>     static const AsnChoiceTypeDesc desc_;
> #endif // META
> 
>     AmpOrderSiteSpecificFields_v2();
>     AmpOrderSiteSpecificFields_v2 (const AmpOrderSiteSpecificFields_v2 &);
>     ~AmpOrderSiteSpecificFields_v2();
> };
> 
> SWIG doesn't really care whether it's a union or not.
> 

Or %include the original header file and add getters/setters for the
union members back in like this:

%extend AmpOrderSiteSpecificFields_v2 {
  void setNull( AmpNull* ampnull ) {
    $self->null = ampnull;
  }
  void setIcap( AmpOrderIcapFields_v2* icap ) {
    $self->icap = icap;
  }
  void setPcx( AmpOrderPCXFields* pcx ) {
    $self->pcx = pcx;
  }
  AmpNull* getNull() {
    return $self->null;
  }
  AmpOrderIcapFields_v2* getIcap() {
    return $self->icap;
  }
  AmpOrderPCXFields* getPcx() {
    return $self->pcx;
  }
}

William

Metzner Robert wrote:
> 
> 
> Dear SWIG Users,
> 
> the last days i spent with wrapping Quantlib with a java Project in
> Eclipse.
> Now i saw that there are arount 60 - 70 SWIG Interface Files, but the
> Quantlib has
> arount 500 Source files.
> I specially miss the inteface to the BlackScholesCalculator class.
> 
> Did anyone create this or these interfaces ?
> Can anyone give me the interface i miss ?
> 
I think you are better off asking on the QuantLib mailing list.

William