10 SWIG and Tcl

Caution: This chapter is under repair!

This chapter discusses SWIG's support of Tcl. SWIG currently requires Tcl 8.0 or a later release. Earlier releases of SWIG supported Tcl 7.x, but this is no longer supported.

Preliminaries

To build a Tcl module, run SWIG using the -tcl option :

$ swig -tcl example.i
If building a C++ extension, add the -c++ option:

$ swig -c++ -tcl example.i

This creates a file example_wrap.c or example_wrap.cxx that contains all of the code needed to build a Tcl extension module. To finish building the module, you need to compile this file and link it with the rest of your program.

Getting the right header files

In order to compile the wrapper code, the compiler needs the tcl.h header file. This file is usually contained in the directory

/usr/local/include
Be aware that some Tcl versions install this header file with a version number attached to it. If this is the case, you should probably make a symbolic link so that tcl.h points to the correct header file.

Compiling a dynamic module

The preferred approach to building an extension module is to compile it into a shared object file or DLL. To do this, you will need to compile your program using comands like this (shown for Linux):

$ swig -tcl example.i
$ gcc -c example.c
$ gcc -c example_wrap.c -I/usr/local/include
$ gcc -shared example.o example_wrap.o -o example.so
The exact commands for doing this vary from platform to platform. SWIG tries to guess the right options when it is installed. Therefore, you may want to start with one of the examples in the SWIG/Examples/tcl directory. If that doesn't work, you will need to read the man-pages for your compiler and linker to get the right set of options. You might also check the SWIG Wiki for additional information.

When linking the module, the name of the output file has to match the name of the module. If the name of your SWIG module is "example", the name of the corresponding object file should be "example.so". The name of the module is specified using the %module directive or the -module command line option.

Static linking

An alternative approach to dynamic linking is to rebuild the Tcl interpreter with your extension module added to it. In the past, this approach was sometimes necesssary due to limitations in dynamic loading support on certain machines. However, the situation has improved greatly over the last few years and you should not consider this approach unless there is really no other option.

The usual procedure for adding a new module to Tcl involves writing a special function Tcl_AppInit() and using it to initialize the interpreter and your module. With SWIG, the tclsh.i and wish.i library files can be used to rebuild the tclsh and wish interpreters respectively. For example:

%module example

extern int fact(int);
extern int mod(int, int);
extern double My_variable;

%include tclsh.i       // Include code for rebuilding tclsh

The tclsh.i library file includes supporting code that contains everything needed to rebuild tclsh. To rebuild the interpreter, you simply do something like this:

$ swig -tcl example.i
$ gcc example.c example_wrap.c \
        -Xlinker -export-dynamic \
        -DHAVE_CONFIG_H -I/usr/local/include/ \
	-L/usr/local/lib -ltcl -lm -ldl \
	-o mytclsh

You will need to supply the same libraries that were used to build Tcl the first time. This may include system libraries such as -lsocket, -lnsl, and -lpthread. If this actually works, the new version of Tcl should be identical to the default version except that your extension module will be a built-in part of the interpreter.

Comment: In practice, you should probably try to avoid static linking if possible. Some programmers may be inclined to use static linking in the interest of getting better performance. However, the performance gained by static linking tends to be rather minimal in most situations (and quite frankly not worth the extra hassle in the opinion of this author).

Using your module

To use your module, simply use the Tcl load command. If all goes well, you will be able to this:

$ tclsh
% load ./example.so
% fact 4
24
%
A common error received by first-time users is the following:
% load ./example.so
couldn't find procedure Example_Init
% 
This error is almost always caused when the name of the shared object file doesn't match the name of the module supplied using the SWIG %module directive. Double-check the interface to make sure the module name and the shared object file match. Another possible cause of this error is forgetting to link the SWIG-generated wrapper code with the rest of your application when creating the extension module.

Another common error is something similar to the following:

% load ./example.so
couldn't load file "./example.so": ./example.so: undefined symbol: fact
% 
This error usually indicates that you forgot to include some object files or libraries in the linking of the shared library file. Make sure you compile both the SWIG wrapper file and your original program into a shared library file. Make sure you pass all of the required libraries to the linker.

Sometimes unresolved symbols occur because a wrapper has been created for a function that doesn't actually exist in a library. This usually occurs when a header file includes a declaration for a function that was never actually implemented or it was removed from a library without updating the header file. To fix this, you can either edit the SWIG input file to remove the offending declaration or you can use the %ignore directive to ignore the declaration.

Finally, suppose that your extension module is linked with another library like this:

$ gcc -shared example.o example_wrap.o -L/home/beazley/projects/lib -lfoo \
      -o example.so
If the foo library is compiled as a shared library, you might get the following problem when you try to use your module:
% load ./example.so
couldn't load file "./example.so": libfoo.so: cannot open shared object file:
No such file or directory
%        
This error is generated because the dynamic linker can't locate the libfoo.so library. When shared libraries are loaded, the system normally only checks a few standard locations such as /usr/lib and /usr/local/lib. To fix this problem, there are several things you can do. First, you can recompile your extension module with extra path information. For example, on Linux you can do this:
$ gcc -shared example.o example_wrap.o -L/home/beazley/projects/lib -lfoo \
      -Xlinker -rpath /home/beazley/projects/lib \
      -o example.so
Alternatively, you can set the LD_LIBRARY_PATH environment variable to include the directory with your shared libraries. If setting LD_LIBRARY_PATH, be aware that setting this variable can introduce a noticeable performance impact on all other applications that you run. To set it only for Tcl, you might want to do this instead:
$ env LD_LIBRARY_PATH=/home/beazley/projects/lib tclsh
Finally, you can use a command such as ldconfig to add additional search paths to the default system configuration (this requires root access and you will need to read the man pages).

Compilation of C++ extensions

Compilation of C++ extensions has traditionally been a tricky problem. Since the Tcl interpreter is written in C, you need to take steps to make sure C++ is properly initialized and that modules are compiled correctly.

On most machines, C++ extension modules should be linked using the C++ compiler. For example:

% swig -c++ -tcl example.i
% g++ -c example.cxx
% g++ -c example_wrap.cxx -I/usr/local/include
% g++ -shared example.o example_wrap.o -o example.so
In addition to this, you may need to include additional library files to make it work. For example, if you are using the Sun C++ compiler on Solaris, you often need to add an extra library -lCrun like this:

% swig -c++ -tcl example.i
% CC -c example.cxx
% CC -c example_wrap.cxx -I/usr/local/include
% CC -G example.o example_wrap.o -L/opt/SUNWspro/lib -o example.so -lCrun
Of course, the extra libraries to use are completely non-portable---you will probably need to do some experimentation.

Sometimes people have suggested that it is necessary to relink the Tcl interpreter using the C++ compiler to make C++ extension modules work. In the experience of this author, this has never actually appeared to be necessary. Relinking the interpreter with C++ really only includes the special run-time libraries described above---as long as you link your extension modules with these libraries, it should not be necessary to rebuild Tcl.

If you aren't entirely sure about the linking of a C++ extension, you might look at an existing C++ program. On many Unix machines, the ldd command will list library dependencies. This should give you some clues about what you might have to include when you link your extension module. For example:

$ ldd swig
        libstdc++-libc6.1-1.so.2 => /usr/lib/libstdc++-libc6.1-1.so.2 (0x40019000)
        libm.so.6 => /lib/libm.so.6 (0x4005b000)
        libc.so.6 => /lib/libc.so.6 (0x40077000)
        /lib/ld-linux.so.2 => /lib/ld-linux.so.2 (0x40000000)
$

As a final complication, a major weakness of C++ is that it does not define any sort of standard for binary linking of libraries. This means that C++ code compiled by different compilers will not link together properly as libraries nor is the memory layout of classes and data structures implemented in any kind of portable manner. In a monolithic C++ program, this problem may be unnoticed. However, in Tcl, it is possible for different extension modules to be compiled with different C++ compilers. As long as these modules are self-contained, this probably won't matter. However, if these modules start sharing data, you will need to take steps to avoid segmentation faults and other erratic program behavior. If working with lots of software components, you might want to investigate using a more formal standard such as COM.

Compiling for 64-bit platforms

On platforms that support 64-bit applications (Solaris, Irix, etc.), special care is required when building extension modules. On these machines, 64-bit applications are compiled and linked using a different set of compiler/linker options. In addition, it is not generally possible to mix 32-bit and 64-bit code together in the same application.

To utilize 64-bits, the Tcl executable will need to be recompiled as a 64-bit application. In addition, all libraries, wrapper code, and every other part of your application will need to be compiled for 64-bits. If you plan to use other third-party extension modules, they will also have to be recompiled as 64-bit extensions.

If you are wrapping commercial software for which you have no source code, you will be forced to use the same linking standard as used by that software. This may prevent the use of 64-bit extensions. It may also introduce problems on platforms that support more than one linking standard (e.g., -o32 and -n32 on Irix).

Setting a package prefix

To avoid namespace problems, you can instruct SWIG to append a package prefix to all of your functions and variables. This is done using the -prefix option as follows :

swig -tcl -prefix Foo example.i

If you have a function "bar" in the SWIG file, the prefix option will append the prefix to the name when creating a command and call it "Foo_bar".

Using namespaces

Alternatively, you can have SWIG install your module into a Tcl namespace by specifying the -namespace option :

swig -tcl -namespace example.i

By default, the name of the namespace will be the same as the module name, but you can override it using the -prefix option.

When the -namespace option is used, objects in the module are always accessed with the namespace name such as Foo::bar.

Building Tcl/Tk Extensions under Windows 95/NT

Building a SWIG extension to Tcl/Tk under Windows 95/NT is roughly similar to the process used with Unix. Normally, you will want to produce a DLL that can be loaded into tclsh or wish. This section covers the process of using SWIG with Microsoft Visual C++. although the procedure may be similar with other compilers.

Running SWIG from Developer Studio

If you are developing your application within Microsoft developer studio, SWIG can be invoked as a custom build option. The process roughly follows these steps :

Now, assuming all went well, SWIG will be automatically invoked when you build your project. Any changes made to the interface file will result in SWIG being automatically invoked to produce a new version of the wrapper file. To run your new Tcl extension, simply run tclsh or wish and use the load command. For example :


MSDOS > tclsh80
% load example.dll
% fact 4
24
%

Using NMAKE

Alternatively, SWIG extensions can be built by writing a Makefile for NMAKE. To do this, make sure the environment variables for MSVC++ are available and the MSVC++ tools are in your path. Now, just write a short Makefile like this :

# Makefile for building various SWIG generated extensions

SRCS          = example.c
IFILE         = example
INTERFACE     = $(IFILE).i
WRAPFILE      = $(IFILE)_wrap.c

# Location of the Visual C++ tools (32 bit assumed)

TOOLS         = c:\msdev
TARGET        = example.dll
CC            = $(TOOLS)\bin\cl.exe
LINK          = $(TOOLS)\bin\link.exe
INCLUDE32     = -I$(TOOLS)\include
MACHINE       = IX86

# C Library needed to build a DLL

DLLIBC        = msvcrt.lib oldnames.lib  

# Windows libraries that are apparently needed
WINLIB        = kernel32.lib advapi32.lib user32.lib gdi32.lib comdlg32.lib 
winspool.lib

# Libraries common to all DLLs
LIBS          = $(DLLIBC) $(WINLIB) 

# Linker options
LOPT      = -debug:full -debugtype:cv /NODEFAULTLIB /RELEASE /NOLOGO /
MACHINE:$(MACHINE) -entry:_DllMainCRTStartup@12 -dll

# C compiler flags

CFLAGS    = /Z7 /Od /c /nologo
TCL_INCLUDES  = -Id:\tcl8.0a2\generic -Id:\tcl8.0a2\win
TCLLIB        = d:\tcl8.0a2\win\tcl80.lib

tcl::
	..\..\swig -tcl -o $(WRAPFILE) $(INTERFACE)
	$(CC) $(CFLAGS) $(TCL_INCLUDES) $(SRCS) $(WRAPFILE)
	set LIB=$(TOOLS)\lib
	$(LINK) $(LOPT) -out:example.dll $(LIBS) $(TCLLIB) example.obj example_wrap.obj

To build the extension, run NMAKE (you may need to run vcvars32 first). This is a pretty minimal Makefile, but hopefully its enough to get you started. With a little practice, you'll be making lots of Tcl extensions.

Primitive Tcl Interface

This section describes the basic Tcl interface. The object-based interface is described shortly.

Functions

C functions are turned into new Tcl commands with the same usage as the C function. Default/optional arguments are also allowed. An interface file like this :

%module example
int foo(int a);
double bar (double, double b = 3.0);
...

Will be used in Tcl like this :

set a [foo 2]
set b [bar 3.5 -1.5]
set b [bar 3.5]              # Note : default argument is used 

There isn't much more to say...this is pretty straightforward.

Global variables

For global variables, SWIG uses Tcl's variable tracing mechanism to provide direct access. For example:

// example.i
%module example
...
double My_variable;
...

# Tcl script
puts $My_variable            # Output value of C global variable
set My_variable 5.5          # Change the value

Compatibility Note: Variable tracing is currently supported for all C/C++ datatypes. In older versions of SWIG, only variables of type int, double, and char * could be linked. All other types were accessed using special function calls.

Constants

Constants are installed as new Tcl variables. For example:
%module example
#define FOO 42
is accessed as follows:
% puts $FOO
42
%
No attempt is made to enforce the read-only nature of a constant. Therefore, a user could reassign the value if they wanted. You will just have to be careful.

A peculiarity of installing constants as variables is that it is necessary to use the Tcl global statement to access constants in procedure bodies. For example:

proc blah {} {
   global FOO
   bar $FOO
}
If a program relies on a lot of constants, this can be extremely annoying. To fix the problem, consider using the following typemap rule:
%apply int CONSTANT { int x };
#define FOO 42
...
void bar(int x);
When applied to an input argument, the CONSTANT rule allows a constant to be passed to a function using its actual value or a symbolic identifier name. For example:
proc blah {} {
   bar FOO
}
When an identifier name is given, it is used to perform an implicit hash-table lookup of the value during argument conversion. This allows the global statement to be ommitted.

Pointers

Pointers to C/C++ objects are represented as character strings such as the following :

_100f8e2_p_Vector

A NULL pointer is represented by the string "NULL". NULL pointers can also be explicitly created as follows :

_0_p_Vector

As much as you might be inclined to modify a pointer value directly from Tcl, don't. The hexadecimal encoding is not necessarily the same as the logical memory address of the underlying object. Instead it is the raw byte encoding of the pointer value. The encoding will vary depending on the native byte-ordering of the platform (i.e., big-endian vs. little-endian). Similarly, don't try to manually cast a pointer to a new type by simply replacing the type-string. This is may not work like you expect and it is particularly dangerous when casting C++ objects. If you need to cast a pointer or change its value, consider writing some helper functions instead. For example:
%inline %{
/* C-style cast */
Bar *FooToBar(Foo *f) {
   return (Bar *) f;
}

/* C++-style cast */
Foo *BarToFoo(Bar *b) {
   return dynamic_cast<Foo*>(b);
}

Foo *IncrFoo(Foo *f, int i) {
    return f+i;
}
%}
If you need to type-cast a lot of objects, it may indicate a serious weakness in your design. Also, if working with C++, you should always try to use the new C++ style casts. For example, in the above code, the C-style cast may return a bogus result whereas as the C++-style cast will return NULL if the conversion can't be performed.

Structures

SWIG generates a basic low-level interface to C structures. For example :

struct Vector {
	double x,y,z;
};

gets mapped into the following collection of C functions :

double Vector_x_get(Vector *obj)
double Vector_x_set(Vector *obj, double x)
double Vector_y_get(Vector *obj)
double Vector_y_set(Vector *obj, double y)
double Vector_z_get(Vector *obj)
double Vector_z_set(Vector *obj, double z)

These functions are then used in the resulting Tcl interface. For example :

# v is a Vector that got created somehow
% Vector_x_get $v
3.5
% Vector_x_set $v 7.8            # Change x component

Similar access is provided for unions and the data members of C++ classes.

C++ Classes

C++ classes are handled by building a set of low level accessor functions. Consider the following class :

class List {
public:
  List();
  ~List();
  int  search(char *item);
  void insert(char *item);
  void remove(char *item);
  char *get(int n);
  int  length;
static void print(List *l);
};

When wrapped by SWIG, the following functions are created :

List    *new_List();
void     delete_List(List *l);
int      List_search(List *l, char *item);
void     List_insert(List *l, char *item);
void     List_remove(List *l, char *item);
char    *List_get(List *l, int n);
int      List_length_get(List *l);
int      List_length_set(List *l, int n);
void     List_print(List *l);

Within Tcl, we can use the functions as follows :

% set l [new_List]
% List_insert $l Ale
% List_insert $l Stout
% List_insert $l Lager
% List_print $l
Lager
Stout
Ale
% puts [List_length_get $l]
3
% puts $l
_1008560_p_List
% 

C++ objects are really just pointers (which are represented as strings). Member functions and data are accessed by simply passing a pointer into a collection of accessor functions that take the pointer as the first argument.

While somewhat primitive, the low-level SWIG interface provides direct and flexible access to almost any C++ object. As it turns out, it is possible to do some rather amazing things with this interface as will be shown in some of the later examples. SWIG also generates an object-based interface that can be used in addition to the basic interface just described here.

The object-based interface

In addition to the low-level accessors, SWIG also generates an object-based interface to C structures and C++ classes. This interface supplements the low-level SWIG interface already defined--in fact, both can be used simultaneously. To illustrate this interface, consider the previous List class :

class List {
public:
  List();
  ~List();
  int  search(char *item);
  void insert(char *item);
  void remove(char *item);
  char *get(int n);
  int  length;
static void print(List *l);
};

Using the object oriented interface requires no additional modifications or recompilation of the SWIG module (the functions are just used differently).

Creating new objects

The name of the class becomes a new command for creating an object. There are 5 methods for creating an object (MyObject is the name of the corresponding C++ class)

MyObject o                   # Creates a new object named `o' 

MyObject o -this $objptr     # Turn a pointer to an existing C++ object into a 
                             # Tcl object named `o'

MyObject -this $objptr       # Turn the pointer $objptr into a Tcl "object"

MyObject -args args          # Create a new object and pick a name for it. A handle
                             # will be returned and is the same as the pointer value.

MyObject                     # The same as MyObject -args, but for constructors that
                             # take no arguments.

Thus, for our List class, you can create new List objects as follows :

List l                       # Create a new list l

set listptr [new_List]       # Create a new List using low level interface
List l2 -this $listptr       # Turn it into a List object named `l2'

set l3 [List]                # Create a new list. The name of the list is in $l3

List -this $listptr          # Turn $listptr into a Tcl object of the same name

Assuming you're not completely confused at this point, the best way to think of this is that there are really two different ways of representing an object. One approach is to simply use the pointer value as the name of an object. For example :

_100e8f8_p_List

The second approach is to allow you to pick a name for an object such as "foo". The different types of constructors are really just a mechanism for using either approach.

Invoking member functions

Member functions are invoked using the name of the object followed by the method name and any arguments. For example :

% List l
% l insert "Bob"
% l insert "Mary"
% l search "Dave"
0
% ...

Or if you let SWIG generate the name of the object, it works like this:

% set l [List]
% $l insert "Bob"            # Note $l contains the name of the object
% $l insert "Mary"
% $l search "Dave"
0
%

Deleting objects

Since objects are created by adding new Tcl commands, they can be deleted by simply renaming them. For example :

% rename l ""                # Destroy list object `l'
It is also possible to explicitly delete the object using the delete method. For example:
% l -delete
If applicable, SWIG will automatically call the corresponding C/C++ destructor when the object is destroyed.

Accessing member data

Member data of an object can be accessed using the cget method. The approach is quite similar to that used in [incr Tcl] and other Tcl extensions. For example :

% l cget -length             # Get the length of the list
13

The cget method currently only allows retrieval of one member at a time. Extracting multiple members will require repeated calls.

The member -this contains the pointer to the object that is compatible with other SWIG functions. Thus, the following call would be legal

% List l                            # Create a new list object
% l insert Mike
% List_print [l cget -this]         # Print it out using low-level function

Changing member data

To change the value of member data, the configure method can be used. For example :

% l configure -length 10     # Change length to 10 (probably not a good idea, but
                             # possible).

In a structure such as the following :

struct Vector {
	double x, y, z;
};

you can change the value of all or some of the members as follows :

% v configure -x 3.5 -y 2 -z -1.0

The order of attributes does not matter.

Managing Object Ownership

By default, objects created from Tcl are owned by the Tcl interpreter and are automatically destroyed when the correponding Tcl variable goes out of scope. However, sometimes it is necessary to change the owneship of an object managed by the interpreter. For instance, if you stored an object in another data structure, you might want Tcl to disown the object. Similarly, it may make sense for Tcl to acquire ownership of an object returned by C.

To handle object ownership, two object methods are available:

% obj -disown            # Release ownership
% obj -acquire           # Acquire ownership
When Tcl owns an object, it is released when the Tcl variable is destroyed. Otherwise, the Tcl variable is destroyed without calling the corresponding C/C++ destructor.

Relationship with pointers

The object oriented interface is mostly compatible with all of the functions that accept pointer values as arguments. Here are a couple of things to keep in mind :

Here is a script that illustrates how these things work :

# Example 1 : Using a named object

List l                       # Create a new list
l insert Dave                # Call some methods
l insert Jane	
l insert Pat
List_print [l cget -this]    # Call a static method (which requires the pointer value)

# Example 2: Let SWIG pick a name

set l [List]                 # Create a new list
$l insert Dave               # Call some methods
$l insert Jane
$l insert Pat
List_print $l                # Call static method (name of object is same as pointer)

# Example 3: Already existing object
set l [new_List]             # Create a raw object using low-level interface
List_insert $l Dave          # Call some methods (using low-level functions)
List -this $l                # Turn it into a Tcl object instead
$l insert Jane
$l insert Part
List_print $l                # Call static method (uses pointer value as before).


Everything from this point on is out of date-- in progress

Examples

Accessing arrays

In some cases, C functions may involve arrays and other objects. In these instances, you may have to write helper functions to provide access. For example, suppose you have a C function like this :

// Add vector a+b -> c
void vector_add(double *a, double *b, double *c, int size);

SWIG is quite literal in its interpretation of double *--it is a pointer to a double. To provide access, a few helper functions can be written such as the following :

// SWIG helper functions for double arrays
%inline %{
double *new_double(int size) {
	return (double *) malloc(size*sizeof(double));
}
void delete_double(double *a) {
	free a;
}
double get_double(double *a, int index) {
	return a[index];
}
void set_double(double *a, int index, double val) {
	a[index] = val;
}
%}

Using our C functions might work like this :

# Tcl code to create some arrays and add them

set a [new_double 200]
set b [new_double 200]
set c [new_double 200]

# Fill a and b with some values
for {set i 0} {$i < 200} {incr i 1} {
	set_double $a $i 0.0
	set_double $b $i $i
}

# Add them and store result in c
vector_add $a $b $c 200

The functions get_double and set_double can be used to access individual elements of an array. To convert from Tcl lists to C arrays, one could write a few functions in Tcl such as the following :

# Tcl Procedure to turn a list into a C array
proc Tcl2Array {l} {
	set len [llength $l]
	set a [new_double $len]
	set i 0
	foreach item $l {
		set_double $a $i $item
		incr i 1
	}
	return $a
}

# Tcl Procedure to turn a C array into a Tcl List
proc Array2Tcl {a size} {
	set l {}
	for {set i 0} {$i < size} {incr i 1} {
		lappend $l [get_double $a $i]
	}
	return $l
}

While not optimal, one could use these to turn a Tcl list into a C representation. The C representation could be used repeatedly in a variety of C functions without having to repeatedly convert from strings (Of course, if the Tcl list changed one would want to update the C version). Likewise, it is relatively simple to go back from C into Tcl. This is not the only way to manage arrays--typemaps can be used as well. The SWIG library file `array.i' also contains a variety of pre-written helper functions for managing different kinds of arrays.

Exception handling

The %except directive can be used to create a user-definable exception handler in charge of converting exceptions in your C/C++ program into Tcl exceptions. The chapter on exception handling contains more details, but suppose you extended the array example into a C++ class like the following :

class RangeError {};   // Used for an exception

class DoubleArray {
  private:
    int n;
    double *ptr;
  public:
    // Create a new array of fixed size
    DoubleArray(int size) {
      ptr = new double[size];
      n = size;
    }
    // Destroy an array
    ~DoubleArray() {
       delete ptr;
    }
    // Return the length of the array
    int   length() {
      return n;
    }

    // Get an item from the array and perform bounds checking.
    double getitem(int i) {
      if ((i >= 0) && (i < n))
        return ptr[i];
      else
        throw RangeError();
    }

    // Set an item in the array and perform bounds checking.
    void setitem(int i, double val) {
      if ((i >= 0) && (i < n))
        ptr[i] = val;
      else {
        throw RangeError();
      }
    }
  };

The functions associated with this class can throw a C++ range exception for an out-of-bounds array access. We can catch this in our Tcl extension by specifying the following in an interface file :

%except(tcl) {
  try {
    $function                // Gets substituted by actual function call
  }
  catch (RangeError) {
    interp->result = "Array index out-of-bounds";
    return TCL_ERROR;
  }
}

or in Tcl 8.0

%except(tcl8) {
  try {
    $function                // Gets substituted by actual function call
  }
  catch (RangeError) {
    Tcl_SetStringObj(tcl_result,"Array index out-of-bounds");
    return TCL_ERROR;
  }
}

When the C++ class throws a RangeError exception, our wrapper functions will catch it, turn it into a Tcl exception, and allow a graceful death as opposed to just having some sort of mysterious program crash. We are not limited to C++ exception handling. Please see the chapter on exception handling for more details on other possibilities, including a method for language-independent exception handling..

Typemaps

This section describes how SWIG's treatment of various C/C++ datatypes can be remapped using the %typemap directive. While not required, this section assumes some familiarity with Tcl's C API. The reader is advised to consult a Tcl book. A glance at the chapter on SWIG typemaps will also be useful.

What is a typemap?

A typemap is mechanism by which SWIG's processing of a particular C datatype can be changed. A simple typemap might look like this :

%module example

%typemap(tcl,in) int {
	$target = (int) atoi($source);
	printf("Received an integer : %d\n",$target);
}
...
extern int fact(int n);

Typemaps require a language name, method name, datatype, and conversion code. For Tcl, "tcl" should be used as the language name. For Tcl 8.0, "tcl8" should be used if you are using the native object interface. The "in" method in this example refers to an input argument of a function. The datatype `int' tells SWIG that we are remapping integers. The supplied code is used to convert from a Tcl string to the corresponding C datatype. Within the supporting C code, the variable $source contains the source data (a string in this case) and $target contains the destination of a conversion (a C local variable).

When the example is compiled into a Tcl module, it will operate as follows :

% fact 6
Received an integer : 6
720
%

A full discussion of typemaps can be found in the main SWIG users reference. We will primarily be concerned with Tcl typemaps here.

Tcl typemaps

The following typemap methods are available to Tcl modules :

%typemap(tcl,in) Converts a string to input function arguments

%typemap(tcl,out) Converts return value of a C function to a string

%typemap(tcl,freearg) Cleans up a function argument (if necessary)

%typemap(tcl,argout) Output argument processing

%typemap(tcl,ret) Cleanup of function return values

%typemap(tcl,const) Creation of Tcl constants

%typemap(memberin) Setting of C++ member data

%typemap(memberout) Return of C++ member data

%typemap(tcl, check) Check value of function arguments.

Typemap variables

The following variables may be used within the C code used in a typemap:

$source Source value of a conversion

$target Target of conversion (where the result should be stored)

$type C datatype being remapped

$mangle Mangled version of data (used for pointer type-checking)

$value Value of a constant (const typemap only)

$arg Original function argument (usually a string)

Name based type conversion

Typemaps are based both on the datatype and an optional name attached to a datatype. For example :

%module foo

// This typemap will be applied to all char ** function arguments
%typemap(tcl,in) char ** { ... }

// This typemap is applied only to char ** arguments named `argv'
%typemap(tcl,in) char **argv { ... }

In this example, two typemaps are applied to the char ** datatype. However, the second typemap will only be applied to arguments named `argv'. A named typemap will always override an unnamed typemap.

Due to the name-based nature of typemaps, it is important to note that typemaps are independent of typedef declarations. For example :

%typemap(tcl, in) double {
	... get a double ...
}
void foo(double);            // Uses the above typemap
typedef double Real;
void bar(Real);              // Does not use the above typemap (double != Real)

To get around this problem, the %apply directive can be used as follows :

%typemap(tcl,in) double {
	... get a double ...
}
void foo(double);

typedef double Real;         // Uses typemap
%apply double { Real };      // Applies all "double" typemaps to Real.
void bar(Real);              // Now uses the same typemap.

Converting a Tcl list to a char **

A common problem in many C programs is the processing of command line arguments, which are usually passed in an array of NULL terminated strings. The following SWIG interface file allows a Tcl list to be used as a char ** object.

%module argv

// This tells SWIG to treat char ** as a special case
%typemap(tcl,in) char ** {
        int tempc;
        if (Tcl_SplitList(interp,$source,&tempc,&$target) == TCL_ERROR) 
		return TCL_ERROR;
}

// This gives SWIG some cleanup code that will get called after the function call
%typemap(tcl,freearg) char ** {
        free((char *) $source);
}

// Return a char ** as a Tcl list
%typemap(tcl,out) char ** {
        int i = 0;
        while ($source[i]) {
             Tcl_AppendElement(interp,$source[i]);
             i++;
        }
}

// Now a few test functions
%inline %{
int print_args(char **argv) {
    int i = 0;
    while (argv[i]) {
         printf("argv[%d] = %s\n", i,argv[i]);
         i++;
    }
    return i;
}


// Returns a char ** list
char **get_args() {
    static char *values[] = { "Dave", "Mike", "Susan", "John", "Michelle", 0};
    return &values[0];
}

// A global variable
char *args[] = { "123", "54", "-2", "0", "NULL", 0 };

%}
%include tclsh.i

When compiled, we can use our functions as follows :

% print_args {John Guido Larry}
argv[0] = John
argv[1] = Guido
argv[2] = Larry
3
% puts [get_args]
Dave Mike Susan John Michelle
% puts [args_get]
123 54 -2 0 NULL
%

Perhaps the only tricky part of this example is the implicit memory allocation that is performed by the Tcl_SplitList function. To prevent a memory leak, we can use the SWIG "freearg" typemap to clean up the argument value after the function call is made. In this case, we simply free up the memory that Tcl_SplitList allocated for us.

Remapping constants

By default, SWIG installs C constants as Tcl read-only variables. Unfortunately, this has the undesirable side effect that constants need to be declared as "global" when used in subroutines. For example :

proc clearscreen { } {
	global GL_COLOR_BUFFER_BIT
	glClear $GL_COLOR_BUFFER_BIT
}

If you have hundreds of functions however, this quickly gets annoying. Here's a fix using hash tables and SWIG typemaps :

// Declare some Tcl hash table variables
%{
static Tcl_HashTable  constTable;      /* Hash table          */
static int           *swigconst;       /* Temporary variable  */
static Tcl_HashEntry *entryPtr;        /* Hash entry          */
static int            dummy;           /* dummy value         */
%}

// Initialize the hash table (This goes in the initialization function)

%init %{
  Tcl_InitHashTable(&constTable,TCL_STRING_KEYS);
%}

// A Typemap for creating constant values
// $source = the value of the constant
// $target = the name of the constant

%typemap(tcl,const) int, unsigned int, long, unsigned long {
  entryPtr = Tcl_CreateHashEntry(&constTable,"$target",&dummy);
  swigconst = (int *) malloc(sizeof(int));
  *swigconst = $source;
  Tcl_SetHashValue(entryPtr, swigconst);
  /* Make it so constants can also be used as variables */
  Tcl_LinkVar(interp,"$target", (char *) swigconst, TCL_LINK_INT | TCL_LINK_READ_ONLY);
};

// Now change integer handling to look for names in addition to values
%typemap(tcl,in) int, unsigned int, long, unsigned long {
  Tcl_HashEntry *entryPtr;
  entryPtr = Tcl_FindHashEntry(&constTable,$source);
  if (entryPtr) {
    $target = ($type) (*((int *) Tcl_GetHashValue(entryPtr)));
  } else {
    $target = ($type) atoi($source);
  }
}

In our Tcl code, we can now access constants by name without using the "global" keyword as follows :

proc clearscreen { } {
	glClear GL_COLOR_BUFFER_BIT
}

Returning values in arguments

The "argout" typemap can be used to return a value originating from a function argument. For example :

// A typemap defining how to return an argument by appending it to the result
%typemap(tcl,argout) double *outvalue {
        char dtemp[TCL_DOUBLE_SPACE];
        Tcl_PrintDouble(interp,*($source),dtemp);
        Tcl_AppendElement(interp, dtemp);
}

// A typemap telling SWIG to ignore an argument for input
// However, we still need to pass a pointer to the C function
%typemap(tcl,ignore) double *outvalue {
	static double temp;         /* A temporary holding place */
	$target = &temp;
}

// Now a function returning two values
int mypow(double a, double b, double *outvalue) {
        if ((a < 0) || (b < 0)) return -1;
        *outvalue = pow(a,b);
        return 0;
};

When wrapped, SWIG matches the argout typemap to the "double *outvalue" argument. The "ignore" typemap tells SWIG to simply ignore this argument when generating wrapper code. As a result, a Tcl function using these typemaps will work like this :

% mypow 2 3     # Returns two values, a status value and the result
0 8
%

An alternative approach to this is to return values in a Tcl variable as follows :

%typemap(tcl,argout) double *outvalue {
	char temp[TCL_DOUBLE_SPACE];
	Tcl_PrintDouble(interp,*($source),dtemp);
	Tcl_SetVar(interp,$arg,temp,0);
}
%typemap(tcl,in) double *outvalue {
	static double temp;
	$target = &temp;
}

Our Tcl script can now do the following :

% set status [mypow 2 3 a]
% puts $status
0
% puts $a
8.0
%
Here, we have passed the name of a Tcl variable to our C wrapper function which then places the return value in that variable. This is now very close to the way in which a C function calling this function would work.

Mapping C structures into Tcl Lists

Suppose you have a C structure like this :

typedef struct {
  char  login[16];            /* Login ID  */
  int   uid;                  /* User ID   */
  int   gid;                  /* Group ID  */
  char  name[32];             /* User name */
  char  home[256];            /* Home directory */
} User;

By default, SWIG will simply treat all occurrences of "User" as a pointer. Thus, functions like this :

extern void add_user(User u);
extern User *lookup_user(char *name);

will work, but they will be weird. In fact, they may not work at all unless you write helper functions to create users and extract data. A typemap can be used to fix this problem however. For example :

// This works for both "User" and "User *"
%typemap(tcl,in) User * {
	int tempc;
	char **tempa;
	static User temp;
	if (Tcl_SplitList(interp,$source,&tempc,&tempa) == TCL_ERROR) return TCL_ERROR;
	if (tempc != 5) {
		free((char *) tempa);
		interp->result = "Not a valid User record";
		return TCL_ERROR;
	}
	/* Split out the different fields */
	strncpy(temp.login,tempa[0],16);
	temp.uid = atoi(tempa[1]);
	temp.gid = atoi(tempa[2]);
	strncpy(temp.name,tempa[3],32);
	strncpy(temp.home,tempa[4],256);
	$target = &temp;
	free((char *) tempa);
}

// Describe how we want to return a user record
%typemap(tcl,out) User * {
	char temp[20];
	if ($source) {
	Tcl_AppendElement(interp,$source->login);
	sprintf(temp,"%d",$source->uid);
	Tcl_AppendElement(interp,temp);
	sprintf(temp,"%d",$source->gid);
	Tcl_AppendElement(interp,temp);
	Tcl_AppendElement(interp,$source->name);
	Tcl_AppendElement(interp,$source->home);
	}
}

These function marshall Tcl lists to and from our User data structure. This allows a more natural implementation that we can use as follows :

% add_user {beazley 500 500 "Dave Beazley" "/home/beazley"}
% lookup_user beazley
beazley 500 500 {Dave Beazley} /home/beazley

This is a much cleaner interface (although at the cost of some performance). The only caution I offer is that the pointer view of the world is pervasive throughout SWIG. Remapping complex datatypes like this will usually work, but every now and then you might find that it breaks. For example, if we needed to manipulate arrays of Users (also mapped as a "User *"), the typemaps defined here would break down and something else would be needed. Changing the representation in this manner may also break the object-oriented interface.

Useful functions

The following tables provide some functions that may be useful in writing Tcl typemaps. Both Tcl 7.x and Tcl 8.x are covered. For Tcl 7.x, everything is a string so the interface is relatively simple. For Tcl 8, everything is now a Tcl object so a more precise set of functions is required. Given the alpha-release status of Tcl 8, the functions described here may change in future releases.

Integers

Tcl_Obj   *Tcl_NewIntObj(int Value);
void       Tcl_SetIntObj(Tcl_Obj *obj, int Value);
int        Tcl_GetIntFromObj(Tcl_Interp *, Tcl_Obj *obj, int *ip);
Floating Point
Tcl_Obj  *Tcl_NewDoubleObj(double Value);
void      Tcl_SetDoubleObj(Tcl_Obj *obj, double value);
int       Tcl_GetDoubleFromObj(Tcl_Interp *, Tcl_Obj *o, double *dp);
Strings
Tcl_Obj  *Tcl_NewStringObj(char *str, int len);
void      Tcl_SetStringObj(Tcl_Obj *obj, char *str, int len);
char     *Tcl_GetStringFromObj(Tcl_Obj *obj, int *len);
void      Tcl_AppendToObj(Tcl_Obj *obj, char *str, int len);
Lists
Tcl_Obj  *Tcl_NewListObj(int objc, Tcl_Obj *objv);
int       Tcl_ListObjAppendList(Tcl_Interp *, Tcl_Obj *listPtr, Tcl_Obj *elemListPtr);
int       Tcl_ListObjAppendElement(Tcl_Interp *, Tcl_Obj *listPtr, Tcl_Obj *element);
int       Tcl_ListObjGetElements(Tcl_Interp *, Tcl_Obj *listPtr, int *objcPtr, Tcl_Obj ***objvPtr);
int       Tcl_ListObjLength(Tcl_Interp *, Tcl_Obj *listPtr, int *intPtr);
int       Tcl_ListObjIndex(Tcl_Interp *, Tcl_Obj *listPtr, int index, Tcl_Obj_Obj **objptr);
int       Tcl_ListObjReplace(Tcl_Interp *, Tcl_Obj *listPtr, int first, int count, int objc, Tcl_Obj *objv);
Objects
Tcl_Obj *Tcl_DuplicateObj(Tcl_Obj *obj);
void     Tcl_IncrRefCount(Tcl_Obj *obj);
void     Tcl_DecrRefCount(Tcl_Obj *obj);
int      Tcl_IsShared(Tcl_Obj *obj);

Standard typemaps

The following typemaps show how to convert a few common kinds of objects between Tcl and C (and to give a better idea of how typemaps work)

Integer conversion

%typemap(in) int, short, long {
   int temp;
   if (Tcl_GetIntFromObj(interp, $input, &temp) == TCL_ERROR)
      return TCL_ERROR;
   $1 = ($1_ltype) temp;
}

%typemap(out) int, short, long {
   Tcl_SetIntObj($result,(int) $1);
}
Floating point conversion
%typemap(in) float, double {
   double temp;
   if (Tcl_GetDoubleFromObj(interp, $input, &temp) == TCL_ERROR)
       return TCL_ERROR;
   $1 = ($1_ltype) temp;
}

%typemap(out) float, double {
   Tcl_SetDoubleObj($result, $1);
}
String Conversion
%typemap(in) char * {
   int len;
   $1 = Tcl_GetStringFromObj(interp, &len);
   }
}

%typemap(out) char * {
   Tcl_SetStringObj($result,$1);
}

Pointer handling

REWRITE THIS!

SWIG pointers are mapped into Python strings containing the hexadecimal value and type. The following functions can be used to create and read pointer values.

These functions can be used in typemaps as well. For example, the following typemap makes an argument of "char *buffer" accept a pointer instead of a NULL-terminated ASCII string.

%typemap(tcl,in) char *buffer {
	if (SWIG_GetPtr($source, (void **) &$target, "$mangle")) {
		Tcl_SetResult(interp,"Type error. Not a pointer", TCL_STATIC);
		return TCL_ERROR;
	}
}

Note that the $mangle variable generates the type string associated with the datatype used in the typemap.

By now you hopefully have the idea that typemaps are a powerful mechanism for building more specialized applications. While writing typemaps can be technical, many have already been written for you. See the SWIG library reference for more information.

Configuration management with SWIG

After you start to work with Tcl for awhile, you suddenly realize that there are an unimaginable number of extensions, tools, and other packages. To make matters worse, there are about 20 billion different versions of Tcl, not all of which are compatible with each extension (this is to make life interesting of course).

While SWIG is certainly not a magical solution to the configuration management problem, it can help out alot in a number of key areas :

Writing a main program and Tcl_AppInit()

The traditional method of creating a new Tcl extension required a programmer to write a special function called Tcl_AppInit() that would initialize your extension and start the Tcl interpreter. A typical Tcl_AppInit() function looks like the following :

/* main.c */
#include <tcl.h>

main(int argc, char *argv[]) {
	Tcl_Main(argc,argv);
	exit(0);
}

int Tcl_AppInit(Tcl_Interp *interp) {
	if (Tcl_Init(interp) == TCL_ERROR) {
		return TCL_ERROR;
	}
	
	/* Initialize your extension */
	if (Your_Init(interp) == TCL_ERROR) {
		return TCL_ERROR;
	}

	tcl_RcFileName = "~/.myapp.tcl";
	return TCL_OK;
}

While relatively simple to write, there are tons of problems with doing this. First, each extension that you use typically has their own Tcl_AppInit() function. This forces you to write a special one to initialize everything by hand. Secondly, the process of writing a main program and initializing the interpreter varies between different versions of Tcl and different platforms. For example, in Tcl 7.4, the variable "tcl_RcFileName" is a C variable while in Tcl7.5 and newer versions its a Tcl variable instead. Similarly, the Tcl_AppInit function written for a Unix machine might not compile correctly on a Mac or Windows machine.

In SWIG, it is almost never necessary to write a Tcl_AppInit() function because this is now done by SWIG library files such as tclsh.i or wish.i. To give a better idea of what these files do, here's the code from the SWIG tclsh.i file which is roughly comparable to the above code

// tclsh.i : SWIG library file for rebuilding tclsh
%{

/* A TCL_AppInit() function that lets you build a new copy
 * of tclsh.
 *
 * The macro SWIG_init contains the name of the initialization
 * function in the wrapper file.
 */

#ifndef SWIG_RcFileName
char *SWIG_RcFileName = "~/.myapprc";
#endif

int Tcl_AppInit(Tcl_Interp *interp){

  if (Tcl_Init(interp) == TCL_ERROR)
    return TCL_ERROR;

  /* Now initialize our functions */
  if (SWIG_init(interp) == TCL_ERROR)
    return TCL_ERROR;

#if TCL_MAJOR_VERSION > 7 || TCL_MAJOR_VERSION == 7 && TCL_MINOR_VERSION >= 5
   Tcl_SetVar(interp,"tcl_rcFileName",SWIG_RcFileName,TCL_GLOBAL_ONLY);
#else
   tcl_RcFileName = SWIG_RcFileName;
#endif
  return TCL_OK;
}

#if TCL_MAJOR_VERSION > 7 || TCL_MAJOR_VERSION == 7 && TCL_MINOR_VERSION >= 4
int main(int argc, char **argv) {
  Tcl_Main(argc, argv, Tcl_AppInit);
  return(0);

}
#else
extern int main();
#endif

%}

This file is essentially the same as a normal Tcl_AppInit() function except that it supports a variety of Tcl versions. When included into an interface file, the symbol SWIG_init contains the actual name of the initialization function (This symbol is defined by SWIG when it creates the wrapper code). Similarly, a startup file can be defined by simply defining the symbol SWIG_RcFileName. Thus, a typical interface file might look like this :

%module graph
%{
#include "graph.h"
#define SWIG_RcFileName "graph.tcl"
%}

%include tclsh.i

... declarations ...


By including the tclsh.i, you automatically get a Tcl_AppInit() function. A variety of library files are also available. wish.i can be used to build a new wish executable, expect.i contains the main program for Expect, and ish.i, itclsh.i, iwish.i, and itkwish.i contain initializations for various incarnations of [incr Tcl].

Creating a new package initialization library

If a particular Tcl extension requires special initialization, you can create a special SWIG library file to initialize it. For example, a library file to extend Expect looks like the following :

// expect.i : SWIG Library file for Expect
%{

/* main.c - main() and some logging routines for expect

Written by: Don Libes, NIST, 2/6/90

Design and implementation of this program was paid for by U.S. tax
dollars.  Therefore it is public domain.  However, the author and NIST
would appreciate credit if this program or parts of it are used.
*/

#include "expect_cf.h"
#include <stdio.h>
#include INCLUDE_TCL
#include "expect_tcl.h"

void
main(argc, argv)
int argc;
char *argv[];
{
        int rc = 0;
        Tcl_Interp *interp = Tcl_CreateInterp();
        int SWIG_init(Tcl_Interp *);

        if (Tcl_Init(interp) == TCL_ERROR) {
                fprintf(stderr,"Tcl_Init failed: %s\n",interp->result);
                exit(1);
        }
        if (Exp_Init(interp) == TCL_ERROR) {
                fprintf(stderr,"Exp_Init failed: %s\n",interp->result);
                exit(1);
        }

        /* SWIG initialization. --- 2/11/96  */
        if (SWIG_init(interp) == TCL_ERROR) {
                fprintf(stderr,"SWIG initialization failed: %s\n", interp->result);
                exit(1);
        }

        exp_parse_argv(interp,argc,argv);
        /* become interactive if requested or "nothing to do" */
        if (exp_interactive)
                (void) exp_interpreter(interp);
        else if (exp_cmdfile)
                rc = exp_interpret_cmdfile(interp,exp_cmdfile);
        else if (exp_cmdfilename)
                rc = exp_interpret_cmdfilename(interp,exp_cmdfilename);

        /* assert(exp_cmdlinecmds != 0) */
        exp_exit(interp,rc);
        /*NOTREACHED*/
}
%}

In the event that you need to write a new library file such as this, the process usually isn't too difficult. Start by grabbing the original Tcl_AppInit() function for the package. Enclose it in a %{,%} block. Now add a line that makes a call to SWIG_init(). This will automatically resolve to the real initialization function when compiled.

Combining Tcl/Tk Extensions

A slightly different problem concerns the mixing of various extensions. Most extensions don't require any special initialization other than calling their initialization function. To do this, we also use SWIG library mechanism. For example :

// blt.i : SWIG library file for initializing the BLT extension
%{
#ifdef __cplusplus 
extern "C" {
#endif
extern int Blt_Init(Tcl_Interp *);
#ifdef __cplusplus
}
#endif
%}
%init %{
	if (Blt_Init(interp) == TCL_ERROR) {
		return TCL_ERROR;
	}
%}


// tix.i : SWIG library file for initializing the Tix extension
%{
#ifdef __cplusplus 
extern "C" {
#endif
extern int Tix_Init(Tcl_Interp *);
#ifdef __cplusplus
}
#endif
%}
%init %{
	if (Tix_Init(interp) == TCL_ERROR) {
		return TCL_ERROR;
	}
%}

Both files declare the proper initialization function (to be C++ friendly, this should be done using extern "C"). A call to the initialization function is then placed inside a %init %{ ... %} block.

To use our library files and build a new version of wish, we might now do the following :

// mywish.i : wish with a bunch of stuff added to it
%include wish.i
%include blt.i
%include tix.i

... additional declarations ...

Of course, the really cool part about all of this is that the file `mywish.i' can itself, serve as a library file. Thus, when building various versions of Tcl, we can place everything we want to use a special file and use it in all of our other interface files :

// interface.i
%module mymodule

%include mywish.i              // Build our version of Tcl with extensions

... C declarations ...

or we can grab it on the command line :

unix > swig -tcl -lmywish.i interface.i

Limitations to this approach

This interface generation approach is limited by the compatibility of each extension you use. If any one extension is incompatible with the version of Tcl you are using, you may be out of luck. It is also critical to pay careful attention to libraries and include files. An extension library compiled against an older version of Tcl may fail when linked with a newer version.

Dynamic loading

Newer versions of Tcl support dynamic loading. With dynamic loading, you compile each extension into a separate module that can be loaded at run time. This simplifies a number of compilation and extension building problems at the expense of creating new ones. Most notably, the dynamic loading process varies widely between machines and is not even supported in some cases. It also does not work well with C++ programs that use static constructors. Modules linked with older versions of Tcl may not work with newer versions as well (although SWIG only really uses the basic Tcl C interface). As a result, I usually find myself using both dynamic and static linking as appropriate.

Turning a SWIG module into a Tcl Package.

Tcl 7.4 introduced the idea of an extension package. By default, SWIG does not create "packages", but it is relatively easy to do. To make a C extension into a Tcl package, you need to provide a call to Tcl_PkgProvide() in the module initialization function. This can be done in an interface file as follows :

%init %{
        Tcl_PkgProvide(interp,"example","0.0");
%}

Where "example" is the name of the package and "0.0" is the version of the package.

Next, after building the SWIG generated module, you need to execute the "pkg_mkIndex" command inside tclsh. For example :

unix > tclsh
% pkg_mkIndex . example.so
% exit

This creates a file "pkgIndex.tcl" with information about the package. To use your

package, you now need to move it to its own subdirectory which has the same name as the package. For example :

./example/
	   pkgIndex.tcl           # The file created by pkg_mkIndex
	   example.so             # The SWIG generated module

Finally, assuming that you're not entirely confused at this point, make sure that the example subdirectory is visible from the directories contained in either the tcl_library or auto_path variables. At this point you're ready to use the package as follows :

unix > tclsh
% package require example
% fact 4
24
%

If you're working with an example in the current directory and this doesn't work, do this instead :

unix > tclsh
% lappend auto_path .
% package require example
% fact 4
24

As a final note, most SWIG examples do not yet use the package commands. For simple extensions it may be easier just to use the load command instead.

Building new kinds of Tcl interfaces (in Tcl)

One of the most interesting aspects of Tcl and SWIG is that you can create entirely new kinds of Tcl interfaces in Tcl using the low-level SWIG accessor functions. For example, suppose you had a library of helper functions to access arrays :

/* File : array.i */
%module array

%inline %{
double *new_double(int size) {
        return (double *) malloc(size*sizeof(double));
}
void delete_double(double *a) {
        free(a);
}
double get_double(double *a, int index) {
        return a[index];
}
void set_double(double *a, int index, double val) {
        a[index] = val;
}
int *new_int(int size) {
        return (int *) malloc(size*sizeof(int));
}
void delete_int(int *a) {
        free(a);
}
int get_int(int *a, int index) {
        return a[index];
}
int set_int(int *a, int index, int val) {
        a[index] = val;
}
%}

While these could be called directly, we could also write a Tcl script like this :

proc Array {type size} {
    set ptr [new_$type $size]
    set code {
        set method [lindex $args 0]
        set parms [concat $ptr [lrange $args 1 end]]
        switch $method {
            get {return [eval "get_$type $parms"]}
            set {return [eval "set_$type $parms"]}
            delete {eval "delete_$type $ptr; rename $ptr {}"}
        }
    }
    # Create a procedure
    uplevel "proc $ptr args {set ptr $ptr; set type $type;$code}"
    return $ptr
}

Our script allows easy array access as follows :

set a [Array double 100]                   ;# Create a double [100]
for {set i 0} {$i < 100} {incr i 1} {      ;# Clear the array
	$a set $i 0.0
}
$a set 3 3.1455                            ;# Set an individual element
set b [$a get 10]                          ;# Retrieve an element

set ia [Array int 50]                      ;# Create an int[50]
for {set i 0} {$i < 50} {incr i 1} {       ;# Clear it
	$ia set $i 0
}
$ia set 3 7                                ;# Set an individual element
set ib [$ia get 10]                        ;# Get an individual element

$a delete                                  ;# Destroy a
$ia delete                                 ;# Destroy ia

The cool thing about this approach is that it makes a common interface for two different types of arrays. In fact, if we were to add more C datatypes to our wrapper file, the Tcl code would work with those as well--without modification. If an unsupported datatype was requested, the Tcl code would simply return with an error so there is very little danger of blowing something up (although it is easily accomplished with an out of bounds array access).

Shadow classes

A similar approach can be applied to shadow classes. The following example is provided by Erik Bierwagen and Paul Saxe. To use it, run SWIG with the -noobject option (which disables the builtin object oriented interface). When running Tcl, simply source this file. Now, objects can be used in a more or less natural fashion.

# swig_c++.tcl
# Provides a simple object oriented interface using
# SWIG's low level interface.
#

proc new {objectType handle_r args} {
    # Creates a new SWIG object of the given type,
    # returning a handle in the variable "handle_r".
    #
    # Also creates a procedure for the object and a trace on
    # the handle variable that deletes the object when the
    # handle varibale is overwritten or unset
    upvar $handle_r handle
    #
    # Create the new object
    #
    eval set handle \[new_$objectType $args\]
    #
    # Set up the object procedure
    #
    proc $handle {cmd args} "eval ${objectType}_\$cmd $handle \$args"
    #
    # And the trace ...
    #
    uplevel trace variable $handle_r uw "{deleteObject $objectType $handle}"
    #
    # Return the handle so that 'new' can be used as an argument to a procedure
    #
    return $handle
}

proc deleteObject {objectType handle name element op} {
    #
    # Check that the object handle has a reasonable form
    #
    if {![regexp {_[0-9a-f]*_(.+)_p} $handle]} {
        error "deleteObject: not a valid object handle: $handle"
    }
    #
    # Remove the object procedure
    #
    catch {rename $handle {}}
    #
    # Delete the object
    #
    delete_$objectType $handle
}

proc delete {handle_r} {
    #
    # A synonym for unset that is more familiar to C++ programmers
    #
    uplevel unset $handle_r
}

To use this file, we simply source it and execute commands such as "new" and "delete" to manipulate objects. For example :

// list.i
%module List
%{
#include "list.h"
%}

// Very simple C++ example

class List {
public:
  List();  // Create a new list
  ~List(); // Destroy a list
  int  search(char *value);
  void insert(char *);  // Insert a new item into the list
  void remove(char *);  // Remove item from list
  char *get(int n);     // Get the nth item in the list
  int  length;          // The current length of the list
static void print(List *l);  // Print out the contents of the list
};

Now a Tcl script using the interface...

load ./list.so list       ; # Load the module
source swig_c++.tcl       ; # Source the object file

new List l
$l insert Dave
$l insert John
$l insert Guido
$l remove Dave
puts $l length_get

delete l

The cool thing about this example is that it works with any C++ object wrapped by SWIG and requires no special compilation. Proof that a short, but clever Tcl script can be combined with SWIG to do many interesting things.


SWIG 1.3 - Last Modified : January 31, 2002