CRITICAL? TFDV 0.21.2 now relies on record_based_tfxio.py in tfx_bsl that has not yet been released

[jasonbrancazio]

This is related to tensorflow/tfx-bsl#3

The new release of tfdv depends on a file in tfx_bsl that has not been released yet.

STEPS TO REPRODUCE LOCALLY (assumes you have virtualenv and virtualenvwrapper installed):

mkvirtualenv jb_testing_tfdv_2 --python=python3.7
pip install tfx==0.21.0 tensorflow==2.1 tensorboard==2.1 tensorflow-data-validation==0.21.2
python
import tensorflow_data_validation

observe the following error:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/Users/jbrancazio/Virtualenvs/jb_testing_tfdv_2/lib/python3.7/site-packages/tensorflow_data_validation/__init__.py", line 33, in <module>
    from tensorflow_data_validation.coders.csv_decoder import DecodeCSV
  File "/Users/jbrancazio/Virtualenvs/jb_testing_tfdv_2/lib/python3.7/site-packages/tensorflow_data_validation/coders/csv_decoder.py", line 26, in <module>
    from tfx_bsl.tfxio import record_based_tfxio
ImportError: cannot import name 'record_based_tfxio' from 'tfx_bsl.tfxio' (/Users/jbrancazio/Virtualenvs/jb_testing_tfdv_2/lib/python3.7/site-packages/tfx_bsl/tfxio/__init__.py)

STEPS TO REPRODUCE (colab):

!pip install tfx==0.21.0 tensorflow==2.1 tensorflow-data-validation==0.21.2
# restart runtime
import tensorflow_data_validation

observe the following error:

ImportError                               Traceback (most recent call last)
<ipython-input-1-65f1eba19dbf> in <module>()
----> 1 import tensorflow_data_validation

1 frames
/usr/local/lib/python3.6/dist-packages/tensorflow_data_validation/__init__.py in <module>()
     31 
     32 # Import coders.
---> 33 from tensorflow_data_validation.coders.csv_decoder import DecodeCSV
     34 from tensorflow_data_validation.coders.tf_example_decoder import DecodeTFExample
     35 

/usr/local/lib/python3.6/dist-packages/tensorflow_data_validation/coders/csv_decoder.py in <module>()
     24 from tensorflow_data_validation import types
     25 from tfx_bsl.coders import csv_decoder as csv_decoder
---> 26 from tfx_bsl.tfxio import record_based_tfxio
     27 from typing import List, Iterable, Optional, Text
     28 

ImportError: cannot import name 'record_based_tfxio'

[paulgc]

@jasonbrancazio There will be a new tfx_bsl release today which should fix this issue.

[rmothukuru]

@jasonbrancazio,
Could run the command successfully in the Google Colab, as per the comment of @paulgc.
Please confirm if the issue is resolved and if we can close this issue. Thanks!

[jasonbrancazio]

resolved, though theoretically the requirement of tfx-bsl<0.22,>=0.21 could still expose a user to the bug in rare cases. A user needs tfx-bsl 0.21.2 or greater.

[rmothukuru]

@jasonbrancazio,
Can you please let us know if we can close this issue. Thanks!

[jasonbrancazio]

@rmothukuru Please see my comment above. You can consider editing https://github.com/tensorflow/data-validation/blob/master/setup.py#L101 to require tfx-bsl 0.21.2 or greater. Other than that, I consider the issue resolved.

[arghyaganguly]

Closing this as fix is available.Thanks.