[vtk-developers] Tcl wrapping and output array arguments

[David Gobbi]

On Mon, Mar 30, 2015 at 8:31 AM, Jorge Perez <josp.jorge at gmail.com> wrote:

> I was also studying the vtkWrapTcl.c code, just the part related to the
> tasks described in 2).
>
> For instance, regarding to vtkVariant value,  the function int
> checkFunctionSignature(ClassInfo *data) reject to wrap a method call if the
> return value is vtkVariant or expect an argument of type vtkVariant in one
> of this checks:
>
> if ((returnType & VTK_PARSE_INDIRECT) != VTK_PARSE_POINTER) // the return
> type is value not a pointer
>
> {
>
> args_ok = 0;
>
> }
>
> ....
>
> if ((argType & VTK_PARSE_INDIRECT) != VTK_PARSE_POINTER) // the argument
> is a value not a pointer
>
> {
>
> args_ok = 0;
>
> }
>

Yes, exactly.  You want to add "if" statements that allow "special" types
like vtkVariant and vtkBoundingBox to get through.

  if ((argType & VTK_PARSE_INDIRECT) == VTK_PARSE_REF || /* pass by
reference e.g. "&param" */
      (argType & VTK_PARSE_INDIRECT) == 0) /* pass-by-value */
      {
      if (strcmp(currentFunction->ReturnClass, "vtkVariant") == 0 ||
          strcmp(currentFunction->ReturnClass, "vtkBoundingox") == 0)
        {
        /* is a special object passed by reference or passed by value */
        }
      else
        {
        /* is unrecognized */
        args_ok = 0;
        }

I think that we can wrap (at least manually) some classes (vtkVariant,
> vtkBoundingBox, ...) not derived from vtkObject as new Tcl_ObjType and
> provide in vtkTclUtil some helper functions to convert from C++ instances
> to Tcl_Obj and the other way from Tcl_Obj to C++ instance.
>
>
> I'm experimenting with this idea: implement new Tcl_ObjType to wrap
> vtkVariant and vtkBoundingBox.
>

Sounds like it should work, but it has been so long since I've worked with
Tcl that I don't remember anything about creating new tcl object types!


By the way the generated C code could be based in Tcl_Obj *objv[] interface
> instead of char *argv[]
>

If you think it would be a good change, put it on gitlab.kitware.com and
I'll take a look.

Cheers,
 - David
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/vtk-developers/attachments/20150330/a9655ba2/attachment-0001.html>