Error on non-optimized join does not pretty print conditions correctly

[smola]

Expected Behavior

When I get the join optimization error (i.e. This Join could not be optimized. This might severely impact the performance...) I should be able to read the list of conditions that could not match.

Current Behavior

I get a default toString representation that is not that helpful.

18/04/09 09:35:30 ERROR GitOptimizer: * Reason:                                                                      *
18/04/09 09:35:30 ERROR GitOptimizer: * Obtained unsupported conditions: Set(org.apache.spark.sql.catalyst.expressio *
18/04/09 09:35:30 ERROR GitOptimizer: * ns.AttributeEquals@53a96fc) 

Possible Solution

Use a human-readable string representation.

Steps to Reproduce (for bugs)

1. Execute a query with a non-optimizable join because of unsupported conditions. Example query at https://github.com/src-d/engine/issues/364
2. Check logs.

Context

I'm exploring complex queries on engine. Better text representation of these errors might remove some guess work involved in identifying problems.

Your Environment

• Version used: engine-jupyter 0.5.5 on Docker
• Operating System and version: pipeline-staging cluster