From 223afe37fad4b89418633f8f91a522ae6cefab85 Mon Sep 17 00:00:00 2001
From: Ivan Smirnov <i.s.smirnov@gmail.com>
Date: Sat, 2 Jul 2016 15:33:04 +0100
Subject: [PATCH] Add documentation re: PYBIND11_DTYPE macro

---
 docs/advanced.rst | 24 ++++++++++++++++++++++++
 1 file changed, 24 insertions(+)

diff --git a/docs/advanced.rst b/docs/advanced.rst
index 1620db71..649454c9 100644
--- a/docs/advanced.rst
+++ b/docs/advanced.rst
@@ -1358,6 +1358,30 @@ template paramenter, and it ensures that non-conforming arguments are converted
 into an array satisfying the specified requirements instead of trying the next
 function overload.
 
+NumPy structured types
+======================
+
+In order for ``py::array_t`` to work with structured (record) types, we first need
+to register the memory layout of the type. This could be done via ``PYBIND11_DTYPE``
+macro which expects the type followed by field names:
+
+.. code-block:: cpp
+
+    struct A {
+        int x;
+        double y;
+    };
+
+    struct B {
+        int z;
+        A a;
+    };
+
+    PYBIND11_DTYPE(A, x, y);
+    PYBIND11_DTYPE(B, z, a);
+
+    /* now both A and B can be used as template arguments to py::array_t */
+
 Vectorizing functions
 =====================