Change 109418220
	Update WORKSPACE to use gmock.BUILD from google/protobuf instead of a duplicate.
	Update google/protobuf's commit hash to include damieng@'s commit.
Change 109417314
	TensorFlow: add .gitignore to ignore some in-tree modified files.
Change 109400051
	Optionally build full TensorFlow for Android.
	1. --define ANDROID_TYPES=__ANDROID_TYPES_FULL__ to register ops
	   for all types, not just float. Today this increases codesize
	   by ~700K when compiled for ARM, though only for clients who
	   request full type support.
	2. Add more ops to android_extended_ops, sufficient to train on
	   the linear regression baseball codelab.
Change 109388118
	Fix the option changed in templatize. Oops.
Change 109382553
	Allows setting a function name in an op's attr in the py frontend.
Change 109380896
	Remove assert_same_graph in favor of op_scope. Change the latter to handle tensor-like objects such as SparseTensor, IndexedSlices, and Variable.

Base CL: 109418322

Change 109195845
	Fix TensorFlow for build against Bazel 0.1.2rc2

	Two things are currently broken with TensorFlow and Bazel 0.1.2:
	  - Bazel now use sandboxing by default on Linux and have fixed it for cc_* rules.
	    Undeclared headers are not mounted in the sandbox which make several cc_* rules
	    fails.
	  - Bazel now enforce strict header checking and some target were missing
	    headers even though the headers were mounted in the sandbox. This change
	    adds a "strict_headers" target that globs every headers of the core
	    library and add it to the `tf_cc_tests` targets.
Change 109162708
	Fix various website issues
	- Fix headline in os_setup.md
	- Fix #anchor links
Change 109162129
	Fix numbers in mnist tutorial, fixes #362
Change 109158967
	Fix typo in word2vec tutorial, fixes #347
Change 109151855
	Fix tile and its gradient for scalars on GPUs

	Eigen doesn't handle scalars on GPUs in all cases.  Fortunately, both
	tile and its gradient are the identity for scalars, so we can just copy
	the input to the output.

	Fixes https://github.com/tensorflow/tensorflow/issues/391.
Change 109140763
	Support int32 and int64 in tf.random_uniform

	This requires a new RandomUniformInt op on the C++ side since the op needs
	to know minval and maxval.

	Fixes https://github.com/tensorflow/tensorflow/issues/364.
Change 109140738
	Fix spacing in docs.
Change 109140030
	Fix content nav to not hide the bottom 100 or so px.
Change 109139967
	Add license files to TensorBoard files, fix mnist_with_summaries test
Change 109138333
	Fix typos in docstring
Change 109138098
	Fix some missing resources in the website.

	Fixes #366.
Change 109123771
	Make sparse_to_dense's default_value default to 0

	Nearly all uses of sparse_to_dense use 0 as the default.  The
	same goes for sparse_tensor_to_dense.

Base CL: 109198336

TensorFlow is an open source software library for numerical computation
using data flow graphs.

Base CL: 107276108