forward declare instead of including boost/thread.hpp (BVLC#1009)

This means that Caffe::Get has to be moved to common.cpp, and loses its
"inline" (but there are no real performance implications).

include/caffe/common.hpp

@@ -2,7 +2,6 @@
 #define CAFFE_COMMON_HPP_
 
 #include <boost/shared_ptr.hpp>
-#include <boost/thread.hpp>
 #include <gflags/gflags.h>
 #include <glog/logging.h>
 
@@ -69,6 +68,9 @@ private:\
 // See PR #1236
 namespace cv { class Mat; }
 
+// Avoid issues with including boost/thread in NVCC source (#1009).
+namespace boost { template <typename T> class thread_specific_ptr; }
+
 namespace caffe {
 
 // We will use the boost shared_ptr instead of the new C++11 one mainly
@@ -99,12 +101,7 @@ void GlobalInit(int* pargc, char*** pargv);
 class Caffe {
  public:
   ~Caffe();
-  inline static Caffe& Get() {
-    if (!singleton_.get()) {
-      singleton_.reset(new Caffe());
-    }
-    return *singleton_;
-  }
+  static Caffe& Get();
   enum Brew { CPU, GPU };
 
   // This random number generator facade hides boost and CUDA rng

src/caffe/common.cpp

@@ -1,3 +1,4 @@
+#include <boost/thread.hpp>
 #include <glog/logging.h>
 #include <cstdio>
 #include <ctime>
@@ -39,6 +40,13 @@ void GlobalInit(int* pargc, char*** pargv) {
   ::google::InstallFailureSignalHandler();
 }
 
+Caffe& Caffe::Get() {
+  if (!singleton_.get()) {
+    singleton_.reset(new Caffe());
+  }
+  return *singleton_;
+}
+
 #ifdef CPU_ONLY  // CPU-only Caffe.
 
 Caffe::Caffe()