spark-instrumented-optimizer/python
Davies Liu 5d8f7b9e87 [SPARK-6677] [SQL] [PySpark] fix cached classes
It's possible to have two DataType object with same id (memory address) at different time, we should check the cached classes to verify that it's generated by given datatype.

This PR also change `__FIELDS__` and `__DATATYPE__` to lower case to match Python code style.

Author: Davies Liu <davies@databricks.com>

Closes #5445 from davies/fix_type_cache and squashes the following commits:

63b3238 [Davies Liu] typo
47bdede [Davies Liu] fix cached classes
2015-04-11 22:33:23 -07:00
..
docs [SPARK-6264] [MLLIB] Support FPGrowth algorithm in Python API 2015-04-09 15:10:10 -07:00
lib [SPARK-2305] [PySpark] Update Py4J to version 0.8.2.1 2014-07-29 19:02:06 -07:00
pyspark [SPARK-6677] [SQL] [PySpark] fix cached classes 2015-04-11 22:33:23 -07:00
test_support [SPARK-3634] [PySpark] User's module should take precedence over system modules 2014-09-24 12:10:09 -07:00
.gitignore [SPARK-3946] gitignore in /python includes wrong directory 2014-10-14 14:09:39 -07:00
run-tests [SPARK-6211][Streaming] Add Python Kafka API unit test 2015-04-09 23:14:24 -07:00