Hello, experts.
I've started my custom driver implementation from
ProjectableFilterableTable. And then found that certain aggregation SQLs
cause the "projected" (the 3rd argument of int[]) to be passed as null.
As far as I understand, this means that all columns are needed. The
problem is that I have dozens of columns, and each SqlTypeName.DOUBLE
column is dynamically calculated. I.e. my table can be seen as a complex
view, and each numerical column may have a different formula/procedure
bound to it.
So, here is a list of nub questions:
1. In a ProjectableFilterableTable, can I do anything about keeping
projected list received by ProjectableFilterableTable.scan()?
If not:
2. What is the best suited base class/interface I need to consider
instead of ProjectableFilterableTable?
Here
(http://mail-archives.apache.org/mod_mbox/calcite-dev/201701.mbox/raw/%[email protected]%3e/1)
I see an advice to use QueryableTable.
3. In examples, I see CsvTranslatableTable implementing QueryableTable
and TranslatableTable. Why may I need the TranslatableTable?
4. From the other point of view, DruidTable implements only
TranslatableTable. Why doesn't it implement QueryableTable?
Thank you!
- Alexey.
