It would be nice to have a return_type argument that allowed results other than a pandas dataframe. Currently, I think that pyarrow.Table, ibis.Table, and pd.DataFrame (defaulting to pd.DataFrame would be a good set of options.
This would look like:
ensdb.genes(return_type=ibis.expr.types.Table)
(I generally like the idea of passing a type instead of a string literal here, but I see how the ibis case is unwieldy)
Advantages
Allows us to work with a few use cases:
pd.DataFrame: useful for people who don't even need to know they are using SQL. Very much the default dataframe type
ibisTable: lazy representation for those who explicitly want to work with the SQL
arrow In between, or if you want to pass it to some other library! I believe this also has potential to be lazy via chunks.
Disadvantages
Mainly complexity. Also keeping data types consistent.
It would probably be pretty easy to start with only supporting pandas and ibis, where the ibis case is just not running .execute()
Description of feature
It would be nice to have a
return_typeargument that allowed results other than a pandas dataframe. Currently, I think thatpyarrow.Table,ibis.Table, andpd.DataFrame(defaulting topd.DataFramewould be a good set of options.This would look like:
(I generally like the idea of passing a type instead of a string literal here, but I see how the ibis case is unwieldy)
Advantages
Allows us to work with a few use cases:
ibisTable: lazy representation for those who explicitly want to work with the SQLarrowIn between, or if you want to pass it to some other library! I believe this also has potential to be lazy via chunks.Disadvantages
Mainly complexity. Also keeping data types consistent.
It would probably be pretty easy to start with only supporting pandas and ibis, where the ibis case is just not running
.execute()