Saver errors in 0.7.0

[aymericdamien]

In 0.7, there are different errors about Saver, first Warnings in serialization, such as moving averages, or dictionaries:

WARNING:tensorflow:Error encountered when serializing moving_average_variables.
Type is unsupported, or the types of the items don't match field type in CollectionDef.
unbound method to_proto() must be called with Variable instance as first argument (got Tensor instance instead)
WARNING:tensorflow:Error encountered when serializing summary_tags.
Type is unsupported, or the types of the items don't match field type in CollectionDef.
'dict' object has no attribute 'name'

and then, it throws an OS Error:

  File "/Users/aymeric/anaconda2/lib/python2.7/site-packages/tensorflow/python/training/saver.py", line 970, in save
    self.export_meta_graph(meta_graph_file_name)
  File "/Users/aymeric/anaconda2/lib/python2.7/site-packages/tensorflow/python/training/saver.py", line 990, in export_meta_graph
    as_text=as_text)
  File "/Users/aymeric/anaconda2/lib/python2.7/site-packages/tensorflow/python/training/saver.py", line 1315, in export_meta_graph
    os.path.basename(filename), as_text=as_text)
  File "/Users/aymeric/anaconda2/lib/python2.7/site-packages/tensorflow/python/training/training_util.py", line 70, in write_graph
    gfile.MakeDirs(logdir)
  File "/Users/aymeric/anaconda2/lib/python2.7/site-packages/tensorflow/python/platform/default/_gfile.py", line 295, in MakeDirs
    os.makedirs(path, mode)
  File "/Users/aymeric/anaconda2/lib/python2.7/os.py", line 157, in makedirs
    mkdir(name, mode)
OSError: [Errno 2] No such file or directory: ''

But this is working well in 0.6.0, so Saver might have some issues in 0.7.0

[mrry]

Thanks for reporting this bug! You can safely ignore the warnings, which are related to the new MetaGraphDef support. If you want to be able to serialize your collections and avoid this warning, you can call the tf.register_proto_function() function for your collection.

The OSError is a bigger problem, and is caused by an incompatibility between our internal and external tf.gfile.MakeDirs() functions. As a temporary workaround, prepend "./" to the save_path argument when calling Saver.save(). I have a change in the pipeline that will fix this error.

[vrv]

If someone can validate that the above commit fixes the problem with Saver, that would be appreciated!

[vrv]

(I was able to run cifar10_train.py with no problems after the above commit).

[aymericdamien]

It is working fine now, thanks!