Python package is missing ModuleSpec in tensorflow.__spec__ in tf 1.14.0

[JonasAmrich]

System information

• Have I written custom code (as opposed to using a stock example script provided in TensorFlow): no
• OS Platform and Distribution (e.g., Linux Ubuntu 16.04): Linux-4.9.125-linuxkit-x86_64-with-Ubuntu-18.04-bionic
• Mobile device (e.g. iPhone 8, Pixel 2, Samsung Galaxy) if the issue happens on mobile device: no
• TensorFlow installed from (source or binary): preinstalled in docker image
• TensorFlow version (use command below): v1.14.0-rc1-22-gaf24dc91b5 1.14.0
• Python version: 3.6.8
• Bazel version (if compiling from source):
• GCC/Compiler version (if compiling from source):
• CUDA/cuDNN version:
• GPU model and memory:

Describe the current behavior

In TF 1.14.0 the module spec in tensorflow.__spec__ is None:

>>> import tensorflow
>>> print(tensorflow.__spec__)
None

Describe the expected behavior

This is different from tf 1.13.1 where it works as expected:

>>> import tensorflow
>>> print(tensorflow.__spec__)
ModuleSpec(name='tensorflow', loader=<_frozen_importlib_external.SourceFileLoader object at 0x7f038cfc3cf8>, origin='/usr/local/lib/python3.5/dist-packages/tensorflow/__init__.py', submodule_search_locations=['/usr/local/lib/python3.5/dist-packages/tensorflow'])

Missing spec causes some problems, e.g. pkgutil now fails when trying to find tensorflow. Note that the first call to find_loader is successful, it only fails after tensorflow is imported:

Python 3.6.8 (default, Jan 14 2019, 11:02:34) 
[GCC 8.0.1 20180414 (experimental) [trunk revision 259383]] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import pkgutil
>>> pkgutil.find_loader('tensorflow')
<_frozen_importlib_external.SourceFileLoader object at 0x7f62372c7160>
>>> import tensorflow
>>> pkgutil.find_loader('tensorflow')
Traceback (most recent call last):
  File "/usr/lib/python3.6/pkgutil.py", line 490, in find_loader
    spec = importlib.util.find_spec(fullname)
  File "/usr/lib/python3.6/importlib/util.py", line 102, in find_spec
    raise ValueError('{}.__spec__ is None'.format(name))
ValueError: tensorflow.__spec__ is None

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/lib/python3.6/pkgutil.py", line 496, in find_loader
    raise ImportError(msg.format(fullname, type(ex), ex)) from ex
ImportError: Error while finding loader for 'tensorflow' (<class 'ValueError'>: tensorflow.__spec__ is None)

Code to reproduce the issue
See above

Other info / logs

I've tested this using official tf docker image (tensorflow/tensorflow:1.14.0-py3) and also using python docker image (python:3.6) with tensorflow installed with pip.

[hhsecond]

importlib throws the same error. Can't do custom imports in 1.14.

Traceback (most recent call last):
  File "path/test/test_model.py", line 62, in testTFGraph
    save_tensorflow(sess, path, output=['output'])
  File "path/model.py", line 55, in save_tensorflow
    if not is_installed('tensorflow'):
  File "path/_util.py", line 18, in is_installed
    if importlib.util.find_spec(p) is None:
  File "path/lib/python3.7/importlib/util.py", line 114, in find_spec
    raise ValueError('{}.__spec__ is None'.format(name))
ValueError: tensorflow.__spec__ is None

[mihaimaruseac]

Another data point: this also happens in nightly

[annarev]

Sorry for the breakage! This is caused by adding a module wrapper that prints deprecation messages. I will send a change to fix it.

[ymodak]

Closing this issue since the associated PR has been merged. Thanks!

[tensorflow-bot]

Are you satisfied with the resolution of your issue?
Yes
No

[SumNeuron]

Is there a work around in the mean time?

[mihaimaruseac]

We'll get the 1.14.1 patch release this week.

[mihaimaruseac]

Update: Instead of a 1.14.1 patch release, we will get a 1.15 version in a few weeks. Sorry for the extra delay this causes.