Re: How do I found out which .i rule/pattern generates a specific (illegal) C++ code fragment?

Kustaa

It is very hard to know what is going on if you just post the output. It's
like asking someone to diagnose a problem in some C++ code and just show
them the assembler output. I suggest you read about the C and SWIG
preprocessors to see what is really happening and cut it all down to a
minimal bit of code to really understand what is going on. Look for missing
definitions and declarations of the relevant types that SWIG actually
parses. Post a few lines of standalone cutdown code here for further help.

William


On Sat, 25 Jul 2020 at 08:35, Kustaa Nyholm <[email protected]>
wrote:

> >
>
> Further details, maybe my *question* was actually wrong!
>
> There seems to be actually 12 occurrences of the same code/pattern in the
> generated file but only the
> one I referred to in the original question causes compile error.
>
>
>
>
> For example this is ok:
>
> // From SWIG generated wrapper
> SWIGEXPORT jlong JNICALL
> Java_org_jcae_opencascade_jni_OccJavaJNI_BRepBuilderAPI_1MakeShape_1shape(JNIEnv
> *jenv, jclass jcls, jlong jarg1, jobject jarg1_) {
>    ...
>   TopoDS_Shape *result = 0 ;
>    ...
>   result = (TopoDS_Shape *) &((BRepBuilderAPI_MakeShape const
> *)arg1)->Shape();
>
>
> // This is the OpenCascade C++ declaration (AFAIU):
> class BRepBuilderAPI_MakeShape {
>         ....
>         virtual const TopoDS_Shape &    Shape ()
>
>
>
>
>
> But this results in 'taking the address of a temporary object of type'
> error:
>
> // From SWIG generated wrapper
>
> SWIGEXPORT jlong JNICALL
> Java_org_jcae_opencascade_jni_OccJavaJNI_BRepBuilderAPI_1ModifyShape_1modifiedShape(JNIEnv
> *jenv, jclass jcls, jlong jarg1, jobject jarg1_, jlong jarg2, jobject
> jarg2_) {
>   ...
>   TopoDS_Shape *result = 0 ;
>   ...
>   result = (TopoDS_Shape *) &((BRepBuilderAPI_ModifyShape const
> *)arg1)->ModifiedShape((TopoDS_Shape const &)*arg2);
>
> // This is the OpenCascade C++ declaration (AFAIU):
> class BRepBuilderAPI_ModifyShape {
>         ...
>         virtual TopoDS_Shape    ModifiedShape (const TopoDS_Shape &S)
> const
>
>
>
> The difference that I see on the C++/OCC side is the 'const' in the return
> type.
>
>
> If that is the problem how can I fix this in the '.i' files without
> touching OCC sources.
> (I suspect the missing 'const' is really a bug or omission but it is what
> it is).
>
>
> wbr Kusti
>
>
> >
> >
> >
> > _______________________________________________
> > Swig-user mailing list
> > [email protected]
> > https://lists.sourceforge.net/lists/listinfo/swig-user
>
>
>
> _______________________________________________
> Swig-user mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/swig-user
>

_______________________________________________
Swig-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/swig-user