diff --git a/tests/test_shared_memory.py b/tests/test_shared_memory.py
index cf077c1..4f2e6fc 100644
--- a/tests/test_shared_memory.py
+++ b/tests/test_shared_memory.py
@@ -4,9 +4,7 @@
 
 def test_shared_memory():
     for test_context in xo.context.get_test_contexts():
-
         print(repr(test_context))
-
         if isinstance(test_context, xo.ContextCupy):
             test_context.default_block_size = 2
             test_context.default_shared_mem_size_bytes = (
@@ -44,10 +42,10 @@ class TestElement(xo.HybridClass):
                   // reduction with an array of 4 doubles using 2 blocks each 2 threads
                   // use reduction with interleaved addressing: https://developer.download.nvidia.com/assets/cuda/files/reduction.pdf
                   // all threads within a block have access to shared memory
-                            
+
                   unsigned int tid = threadIdx.x;  // thread ID within the block: 0,1
                   unsigned int gid = blockIdx.x*blockDim.x + threadIdx.x;  // global thread ID: 0,1,2,3
-                    
+
                   // init shared memory with chunk of input array
                   extern __shared__ double sdata[2];
                   sdata[tid] = input_arr[gid];
@@ -68,7 +66,6 @@ class TestElement(xo.HybridClass):
             def __init__(
                 self, _context=None, _buffer=None, _offset=None, _xobject=None
             ):
-
                 if _xobject is not None:
                     self.xoinitialize(
                         _xobject=_xobject,
diff --git a/xobjects/_version.py b/xobjects/_version.py
index c49a95c..75cf783 100644
--- a/xobjects/_version.py
+++ b/xobjects/_version.py
@@ -1 +1 @@
-__version__ = "0.2.8"
+__version__ = "0.2.9"
diff --git a/xobjects/hybrid_class.py b/xobjects/hybrid_class.py
index 2f0a743..2b504c6 100644
--- a/xobjects/hybrid_class.py
+++ b/xobjects/hybrid_class.py
@@ -301,7 +301,13 @@ def to_dict(self, copy_to_cpu=True):
 
         if hasattr(obj, "_store_in_to_dict"):
             for nn in obj._store_in_to_dict:
-                out[nn] = getattr(obj, nn)
+                ww = getattr(obj, nn)
+                if hasattr(ww, "to_dict"):
+                    out[nn] = ww.to_dict()
+                elif hasattr(ww, "_to_dict"):
+                    out[nn] = ww._to_dict()
+                else:
+                    out[nn] = ww
 
         return out