commit dbc0004f995556b288b43077458b61683ac93855
author Tullie Murrell <tullie@fb.com> Wed Aug 29 23:40:15 2018 -0700
committer Facebook Github Bot <facebook-github-bot@users.noreply.github.com> Wed Aug 29 23:55:28 2018 -0700
tree 2aeefd7cbc8e2430186fd5a315b9ed30de89f9e3
parent 23af7deea7efa1172a54133cba5d13358cdf3cc0

Remove use_count() == 1 in Tensor::Extend (#11046)

Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/11046

As suggested by jerryzh168, temporary fix for a new constraint that was added D9350686 is to remove this assert. Long term jerryzh168 is going to work out a better way of handling this.

Reviewed By: jerryzh168

Differential Revision: D9566323

fbshipit-source-id: e4630c7cbe0cc68a084974ea7048654811fae01f

caffe2/core/tensor.h

diff --git a/caffe2/core/tensor.h b/caffe2/core/tensor.h
index 21dc126..f1934f5 100644
--- a/caffe2/core/tensor.h
+++ b/caffe2/core/tensor.h
@@ -275,9 +275,6 @@
     CAFFE_ENFORCE_GE_WITH_CALLER(dims_.size(), 1);
     CAFFE_ENFORCE_GE_WITH_CALLER(
         num, 0, "`num` must be non-negative for Extend");
-    CAFFE_ENFORCE(
-        storage_.use_count() == 1,
-        "Can't call Extend on shared storage, please call Resize instead");
     auto newDims = dims_;
     newDims[0] += num;
     if (!storage_->data()) {