Why use shadoworflow gfile? (for input / output files)

Question

Tensorflow code uses file input / output methods that are different from python's built-in methods. According to the source code , it is useful as

"File I / O attachments without blocking the stream"

I am not sure in which cases it is useful and when it should not be used.

Any idea?

thank

Answer 1

This comment:

I / O file wrappers without blocking threads

... is a particularly useless description for the TensorFlow module tf.gfile!

The main roles of the module tf.gfile:

• API, Python file,
• TensorFlow ++ FileSystem API.

API ++ FileSystem , , Google ( gs://) HDFS ( hdfs://). TensorFlow tf.gfile, , TensorBoard ( ). , , API Python - .