[psosnows]

Dear Foamers,

I implemented a derived boundary condition, and added a new method. Now I am trying to use this new method "from the main code".

In general, the derived patch field is:

class myDerivedPatchField
:
    public fixedGradientFvPatchScalarField
{
   TypeName("myDerivedPatch");
//(...) some standard overloading of updateCoeffs(), constructors, etc
   public:
   bool myMethod() 
   {
      return true;
   }
}

Now supposing I created a volScalarField F, which has as one of its boundaries myDerivedPatchField, I want to call that method in the main code of the solver:

//(...)
int main(int argc, char *argv[])
{
   //(...) declarations, loading, creating volScalarField F
   forAll(F.boundaryField(),b)
   {
      if(isA<myDerivedPatchField>(F.boundaryField()[b]))
      {
         Info << refCast<myDerivedPatchField>(F.boundaryField()[b]).myMethod() <<endl;
      }
   }
}

Unfortunately isA<myDerivedPatchField>(F.boundaryField()[b]) returns always false. From what I get, F.boundaryField()[b] at this level of the code is a fvPatchField, and does not match myDerivedPatchField. At the same time, without the isA<>() check, refCast breaks as it encounters some other boundary type (like zeroGradient).

What should I do to cast an added method in the main code?

Thank you in advance!
Pawel

[hua1015]

Dear Psosnows，
I also met this problem but failed to correct it. Have you got the right way?I will be thankful for your suggestions.

[wyldckat]

Greetings to all!

Can either one of you attach the 3 things necessary for replicating this issue? Namely:

1. The derived boundary condition.
2. The modified solver.
3. A test case to use them.

Because I don't know the answer, but if I have access to the 3, I can easily solve your problem.


Best regards,
Bruno

[ngj]

Good evening,

I do not have access to a prober computer, so I do not have the means of checking the hypothesis, but...:

You should carefully check the "isA<>()" command, because if it take a patchField and not a fvPatchField it makes all the difference. Your boundary condition is derived from a patchField deep down, so you might encounter this problem, because the many derived levels are merely ignored.

Good luck,

Niels

[ngj]

Also, it might be that isA asks for the physical type of the boundary condition, i.e. you might want to check on the typeName instead,. Something like this:

if (F.boundaryField()[p]::typeName == Foam::myDerivedFvPatchField::typeName)

Note that I have not tried to compile this, so it might fail. Especially, I am uncertain of the need of "()" after typeName.

Kind regards,

Niels