[SIP-1] Split the Table visualization into two visualizations

SIP-1 Split the Table visualization into two visualizations

Motivation

The current Table visualization is confusing as it shows 2 control sections, GROUP BY and NOT GROUPED BY that are mutually exclusive. It’s not clear that they are, and if both sections are used Superset will raise and error suggesting to clear some of the controls

Proposed Change

Split the current table visualization into 2 entry points: Table - Samples and Table - Aggregated. The GROUP BY and NOT GROUPED BY control sections will become a Query section in respective visualizations.

New or Changed Public Interfaces

The main change to users is the 2 entries points, essentially a new visualization in the visualizations menu.

Migration Plan and Compatibility

• a database migration re-assigning NOT GROUPED BY visualizations to the new type
• minor code changes introducing the new visualization

Rejected Alternatives

N/A

@betodealmeida what do you think of this template for SIPs? Took it from here: https://cwiki.apache.org/confluence/display/KAFKA/Kafka+Improvement+Proposals

test edit

[ktmud]

Didn’t know I accidentally fixed SIP No. 1!

[john-bodley]

@mistercrunch we agree that the current table UX is confusing, however we aren’t sure if this is really the long-term solution. The atomic vs. non-atomic issue is systemic to all visualization types, not just to the table. i.e., if you extrapolate this logic, we would have to have two time series charts, two bar charts, etc., to support both atomic tables and aggregated data sources.

It seems like a better solution would be to extend ad-hoc metrics to understand the difference between atomic vs non-atomic, i.e., a metric could be an aggregation on a non-atomic column or merely the value of an atomic column.