Github user tomatophantastico commented on the issue:
https://github.com/apache/metamodel/pull/163
To me, there are two use cases / requirements here:
1. Make queries transferable
2. Ease of exploring a schema and its data via SQL
w.r.t 1)
Yes, i think the data contexts are the place where this should be
implemented. As most (all?) non-jdbc datacontexts are extending
QueryPostProcessingContext, this could perhaps be done without too many changes.
i generally think it is not a good idea to deal with a table
"Backup_23423_hash_by_Jörg_do_not_modify.csv" in the query, especially if it
is an export of the lets say person table. Having an alias would clear up the
semantics of the query.
"select * from person" will work on the source system, "select * from
tables[0]" not.
w.r.t. 2)
Just iterating over tables without having the table name, does not really
make sense to me. How should the result of such a query look like?
Most DBMs implement this via custom commands (list tables/documents...) or
using tables (SELECT table_schema,table_name FROM information_schema.tables).
Maybe we can cook up sth. more convenient here. But i would prefer to
change the semantics of SQL.
---
