Use move to avoid copying (pytorch#17188)

Michael Liu · facebook-github-bot · commit f2f40302941f · 2019-02-19T07:14:27.000-08:00
Summary: Pull Request resolved: pytorch#17188 Using flag "-Wreturn-std-move", compiler can identify the cases where a copy operation is performed when a move operation would have been available. Wrapped return statement with std::move to fix. For some reason, these files are not automatically modded. With D14115372 we should be able to turn on the compile flag Reviewed By: soumith Differential Revision: D14115786 fbshipit-source-id: e763b92eecbe4468027fc141d029618d1e9f280b
diff --git a/torch/csrc/autograd/VariableTypeManual.cpp b/torch/csrc/autograd/VariableTypeManual.cpp
@@ -341,7 +341,7 @@ Tensor VariableType::detach(const Tensor & self) const {
   if (jit::tracer::isTracing()) {
     jit::tracer::addOutput(node, result);
   }
-  return result;
+  return std::move(result);
 }
 
 Tensor & VariableType::detach_(Tensor & self) const {

Original file line number	Diff line number	Diff line change
`@@ -341,7 +341,7 @@ Tensor VariableType::detach(const Tensor & self) const {`
`341`	`341`	`if (jit::tracer::isTracing()) {`
`342`	`342`	`jit::tracer::addOutput(node, result);`
`343`	`343`	`}`
`344`		`- return result;`
	`344`	`+ return std::move(result);`
`345`	`345`	`}`
`346`	`346`
`347`	`347`	`Tensor & VariableType::detach_(Tensor & self) const {`