when column type is decimal, should add precision and scale

[cdmikechen]

When using HoodieHiveClient to create a hive table, I found that a decimal column just only show DECIMAL in schema and don't have precision and scale. So that hoodie will report error, if hoodie is analysing this schema.
Try to fix this problem.

[vinothchandar]

@cdmikechen thanks for the fix. does this happen when hive syncing a DECIMAL column?

[cdmikechen]

@vinothchandar yes. I have a hive table with decimal column.
When I used run_sync_tool.sh to sync a hoodie data to hive table, hoodie used hive jdbc to read hive schema and checked difference from hive and hoodie. I found a decimal column in hive table didn't have precision and scale, but hoodie table had, so that hoodie thought they were different column type and reported error.

[cdmikechen]

I found this problem when I revised a change about not using com.databricks:spark-avro_2.11 and letting hoodie support timestamp and decimal column type below spark2.4.
I have completed spark test about supporting timestamp and decimal column type but have some problem in testing hive. When I finish testing I will take a PR.

[vinothchandar]

sg. In parallel, let me try to fully understand these gaps. overall lg otherwise

[vinothchandar]

@cdmikechen have not been able to spend time on this much.. any updates from your testing?

[vinothchandar]

This timestamp handling has dogged us for a while. :( if you understand it fully, can you please put up a HIP with your suggestions.. we can then divvy up the actual impl..

[cdmikechen]

@vinothchandar add a pr: #770
You can see this pr, the problem of timestamp and decimal data types should be solved in spark. But timestamp can not be solved in Hive, because Hive don't support parquet logical_type.

[bvaradar]

@cdmikechen - @n3nash or myself will review this in couple of days and will get back to you

[vinothchandar]

@cdmikechen finally understood :) .. Thanks. merging!