Addons: macOS build from source fails

[drewszurko]

System information

• Have I written custom code (as opposed to using a stock example script provided in TensorFlow):
• OS Platform and Distribution (e.g., Linux Ubuntu 16.04): macOS Mojave 10.14.4
• TensorFlow installed from (source or binary): binary
• TensorFlow version (use command below): tf-nightly-2.0-preview==2.0.0.dev20190419
• TensorFlow Addons installed from (source, PyPi): source
• TensorFlow Addons version: 0.2.1
• Python version and type (eg. Anaconda Python, Stock Python as in Mac, or homebrew installed Python etc): pyenv Python 3.6.8
• Bazel version (if compiling from source): 0.24.1
• GCC/Compiler version (if compiling from source): Apple LLVM version 10.0.1 (clang-1001.0.46.3)
• Is GPU used? (yes/no): no
• GPU model (if used): N/A

Describe the bug
During source install bazel build build_pip_pkg returned the following error:

libtensorflow_framework.so.2: No such file or directory

I replaced all instances of libtensorflow_framework.so.2 with libtensorflow_framework.2.dylib in BUILD.tpl and tf_configure.bzl files.

Attempted fresh source install. bazel build build_pip_pkg returned new error:

bazel-out/darwin-fastbuild/genfiles/external/local_config_tf/include/tensorflow/core/framework/op_def.pb.h:10:10: fatal error: 'google/protobuf/port_def.inc' file not found
#include <google/protobuf/port_def.inc>

[seanpmorgan]

Hi @drewszurko thanks for reporting this.. we are very close to having working macOS builds and I apologize for the inconvenience.

For the first issue I have a branch cut that will lookup the tensorflow shared library name:
https://github.com/seanpmorgan/addons/commit/0fe21efd8a5b52b70fd2e997f0d85b8d8f5b2074

Just waiting on a fix in tensorflow core for your second issue:
https://github.com/tensorflow/tensorflow/issues/27697#event-2289154760

[seanpmorgan]

Both issues are now resolved. Addons can be built from source using OSX without the need to modify the source code or download coreutils.

We will hopefully have a packaged 0.3 macos whl in the next couple days.

cc @drewszurko @ageron