Separating creation of query from execution of it
Currently, we have helper functions that actually build the query and then immediately execute returning the retrieved data and the query itself.
I suggest that those functions only return the statement, which would be called at initialization time storing it in an instance attribute of the constraint and then in retrieve it executes it.
My rationale to do this is the following case. I want to create the constraint and I want to obtain the SQL query that said constraint will produce without having to actually perform the test (or more exactly the retrieval). This is particularly interesting for explorative use of this library and when the execution of the query could take a long time, e.g., more than 10 minutes.
This has no priority, just a suggested idea, which also establishes better the responsibility of the retrieve method.
- [x] AggregateNumericRangeEquality
- [ ] Uniques
- [ ] NRows
- [ ] ...
- [ ] Add
selectionasabstractmethodinConstraintbase class.
Sounds cool!