Gonzalo, I think the implementation is fairly straightforward. You’d write a new implementation of RelWriter.
But, as Alessandro says, the specification is the hard part. What *exactly* would you want instead of $0? (Those fields have names, and most of the time they are the same as the user wrote in SQL, but sometimes they are not. For example in the case where you’re joining EMP to DEPT and both sides have a field called DEPTNO.) Next step is for you to write a Jira case with a specification of what you want. Julian > On Apr 30, 2024, at 8:06 AM, Alessandro Solimando > <alessandro.solima...@gmail.com> wrote: > > Hi Gonzalo, > what would you expect when $i points to a complex expression from the input > relation(s)? > > It's easier to brainstorm around a concrete running example, would you have > one to share? (existing vs sought behavior) > > Best regards, > Alessandro > > On Tue, 30 Apr 2024 at 12:39, Gonzalo Ortiz Jaureguizar < > golthir...@gmail.com> wrote: > >> Hi community, >> >> I'm trying to improve explain plans in Apache Pinot and one of the main >> complaints of our customers is how difficult it is to read the explain >> plans having to transform the synthetic input names into the corresponding >> logical field. >> >> Is there a way to improve that? Ideally these inputs should conserve the >> names used in SQL. For example that is what Postgres does. >> >> Thanks for your help, >> >> Gonzalo >>
