Diagnostic Attribute Error

SQLFlow compiles each statement in the program into an execution plan and executes them. As the TRAIN statement above, SQLFlow uses clause to train a specific model called DNNClassifier, using clause to configure the training arguments.

Sometimes users may make some configuration mistake on WITH clause, then the job would fail during execution and return some uncertain error message.

1. Early testing, we can do early testing before running the job; users can wait less time and cluster save resources.
2. More accurate diagnostic message.

We want to document the compiler-readable description of model parameters in the docstring of Python function or class that define a model.

A docstring contains multiple lines:

• A line starting with is the check rule in Python code.
• A line starting with argument name and document followed by a colon :.

If a user set some invalid parameters as the following SQL statement:

We expected the SQLFlow GUI show the error message as:

This documentation using native Python code to express the check rule, another PR designed a new Python library to make the code shorter and simpler, will make more discussion in the future.