There’s also the “_MAP” column returned by the MongoDB and Elasticsearch adapters. It contains a (key, value) pair for the top-level fields of each row. It DOES appear in SELECT *.
We had the idea of rewriting SELECT empno FROM Emp to SELECT _MAP[‘empno’] FROM Emp if Emp had a particular row type but I don’t think we ever did it. Julian > On Apr 17, 2017, at 7:27 PM, James Taylor <[email protected]> wrote: > > Hey Jacques, > The EXTENDS (col1 type, col2 type) keyword can be used for this (and the > EXTENDS keyword is optional). Something like this in Phoenix: > > SELECT foo FROM bar (my_col VARCHAR) WHERE my_col = 'bas' > > Phoenix uses this for dynamic column support. > > Thanks, > James > > On Mon, Apr 17, 2017 at 7:17 PM, Jacques Nadeau <[email protected]> wrote: > >> Hey All, >> >> I remember there being a syntax in Calcite that allowed one to define an >> _EXTRA column or columns that could be referenced in Calcite but wouldn't >> be included in a select *. If I recall, the Splunk adapter used this at one >> point. Does this actually exist or have I lost my mind. I can't seem to >> find a reference to this anywhere. >> >> thanks! >>
