Fix the type to use the new c_init_code interface and work on something other than OSX.

theano/sandbox/gpuarray/basic_ops.py

@@ -151,17 +151,13 @@ class GpuFromHost(Op):
         return """
         PyArrayObject *%(name)s_tmp;
         int %(name)serr;
-        if ((PyObject *)GpuArray_default_context == Py_None) {
-            PyErr_SetString(PyExc_ValueError, "No default context, gpuarray not initialized?");
-            %(fail)s
-        }
         %(name)s_tmp = PyArray_GETCONTIGUOUS(%(inp)s);
         if (%(name)s_tmp == NULL) {
             // PyArray_GETCONTIGUOUS sets an error message if it fails
             %(fail)s
         }
         Py_XDECREF(%(out)s);
-        %(out)s = new_GpuArray((PyObject *)&GpuArrayType, GpuArray_default_context);
+        %(out)s = new_GpuArray((PyObject *)&GpuArrayType, GpuArray_default_context());
         if (%(out)s == NULL) {
             Py_DECREF(%(name)s_tmp);
             // new_GpuArray calls __new__ which will set an error message
@@ -169,8 +165,8 @@ class GpuFromHost(Op):
             %(fail)s
         }
         %(name)serr = GpuArray_empty(&%(out)s->ga,
-                                     GpuArray_default_context->ops,
-                                     GpuArray_default_context->ctx,
+                                     GpuArray_default_context()->ops,
+                                     GpuArray_default_context()->ctx,
                                      %(typecode)s,
                                      PyArray_NDIM(%(inp)s),
                                      (size_t *)PyArray_DIMS(%(inp)s),
@@ -279,7 +275,7 @@ class GpuFromCuda(Op):
         ssize_t *%(name)sstr;
 
         cuCtxGetCurrent(&%(name)scur);
-        if (%(name)scur != cuda_get_ctx(GpuArray_default_context->ctx)) {
+        if (%(name)scur != cuda_get_ctx(GpuArray_default_context()->ctx)) {
             PyErr_SetString(PyErr_ValueError, "Ambient cuda context is not the same as output context.");
             %(fail)s
         }
@@ -308,7 +304,7 @@ class GpuFromCuda(Op):
             %(fail)s
         }
 
-        %(name)sdata = cuda_make_buf(GpuArray_default_context->ctx,
+        %(name)sdata = cuda_make_buf(GpuArray_default_context()->ctx,
                                      (CUdeviceptr)%(in)s->devdata,
                                      (size_t)%(in)s->data_allocated);
         if (%(name)sdata == NULL) {
@@ -319,7 +315,7 @@ class GpuFromCuda(Op):
             %(fail)s
         }
         %(name)serr = GpuArray_fromdata(&%(out)s->ga,
-                                        GpuArray_default_context->ops,
+                                        GpuArray_default_context()->ops,
                                         %(name)sdata, 0, GA_FLOAT, %(in)s->nd,
                                         %(name)sdims, %(name)sstr, 1);
         free(%(name)sdims);

theano/sandbox/gpuarray/type.py

@@ -174,15 +174,23 @@ class GpuArrayType(Type):
         }
         """ % {'name': name}
 
+    def c_init_code(self):
+        # We don't actually need the numpy API except in
+        # HostFromGpu and GpuFromHost and those case will be covered
+        # by the TensorType parameter
+        return ['import_pygpu__gpuarray();']
+
     def c_headers(self):
-        return ['pygpu/gpuarray.h', 'compyte/array.h', 'compyte/kernel.h',
-                'compyte/error.h']
+        # We need arrayobject for the PyArrayDescr struct def
+        # (even if we just use a pointer to it in a function def)
+        return ['<compyte/array.h>', '<compyte/kernel.h>', '<compyte/error.h>',
+                '<numpy/arrayobject.h>', '<gpuarray_api.h>']
 
     def c_header_dirs(self):
-        return [pygpu.get_include()]
+        return [pygpu.get_include(), numpy.get_include()]
 
     def c_code_cache_version(self):
-        return (0,)
+        return (1,)
 
 
 class _operators(tensor.basic._tensor_py_operators):
@@ -265,7 +273,7 @@ theano.compile.register_view_op_c_code(GpuArrayType, """
 
 theano.compile.register_deep_copy_op_c_code(GpuArrayType, """
     Py_XDECREF(%(oname)s);
-    %(oname)s = new_GpuArray((PyObject *)&GpuArrayType, GpuArray_default_context);
+    %(oname)s = new_GpuArray((PyObject *)&GpuArrayType, GpuArray_default_context());
     if (!%(oname)s) { %(fail)s }
     int err;
     err = GpuArray_copy(&%(oname)s->ga, &%(iname)s->ga, GA_ANY_ORDER);