Making transformer plan log more obvious

[loudongfeng]

Is your feature request related to a problem or challenge? Please describe what you are trying to do.
Making the transformer plan log more obvious. Plan for sql

select count(*) from my_char where name = 'Nemon'

Now is

CHNativeColumnarToRow
+- *(2) HashAggregateTransformer(keys=[], functions=[count(1)], output=[count(1)#7])
   +- ShuffleQueryStage 0
      +- ColumnarExchangeAdaptor SinglePartition, ENSURE_REQUIREMENTS, false, [plan_id=49], [id=#49], [OUTPUT] List(count:LongType), [OUTPUT] List(count:LongType)
         +- *(1) HashAggregateTransformer(keys=[], functions=[partial_count(1)], output=[count#10L])
            +- *(1) ProjectExecTransformer
               +- *(1) FilterExecTransformer (isnotnull(name#2) AND (name#2 = Nemon))
                  +- FileScan orc tpcds_parquet.my_char[name#2] Batched: true, DataFilters: [isnotnull(name#2), (name#2 = Nemon)], Format: ORC, ...

A more obvious one will look like this

CHNativeColumnarToRow
+- ^(2) HashAggregateTransformer(keys=[], functions=[count(1)], output=[count(1)#7])
   +- ShuffleQueryStage 0
      +- ColumnarExchangeAdaptor SinglePartition, ENSURE_REQUIREMENTS, false, [plan_id=49], [id=#49], [OUTPUT] List(count:LongType), [OUTPUT] List(count:LongType)
         +- ^(1) HashAggregateTransformer(keys=[], functions=[partial_count(1)], output=[count#10L])
            +- ^(1) ProjectExecTransformer
               +- ^(1) FilterExecTransformer (isnotnull(name#2) AND (name#2 = Nemon))
                  +- NativeFileScan orc tpcds_parquet.my_char[name#2] Batched: true, DataFilters: [isnotnull(name#2), (name#2 = Nemon)],...

[PHILO-HE]

Are you proposing to use NativeFileScan to name the plan node to distinguish with Spark's? It can be achieved by overriding nodeName method. Looks reasonable. I am not quite sure why ^ is replacing * in your proposal. Any reasons?

[loudongfeng]

Using ^ , so we can distinguish plan between whole stage codegen and whole stage transformer.
In spark plan , * usually means the plan node is codegened, same ID means they are in the same whole stage codegen.
We can keep the ID part, but use a ^ or other symbols instead.