Add Slice support for arbitrary queries (including parameterization)

[Jparks2532]

Slice support for arbitrary queries (including parameterization) would enable all types of use cases where a materialized table or view doesn't quite work. For instance, the following query template could serve as the source for a Slice:

query:

SELECT
    dim,
    SUM(cnt) AS first_moment
    SUM(cnt*cnt) AS second_moment
FROM (
    SELECT id_user, <dimension1> AS dim, COUNT(*) cnt
    FROM table1
    WHERE ds between '<date_start>' and '<date_end>'
    GROUP BY id_user, <dimension1> 
) t
GROUP BY dim;

Users could pass in parameters at query time.

date_start: date
date_end: date
dimension1: string

[yejianye]

Very interesting idea, how will that work with 'group by', 'metrics' parameters in various viz types?

[Jparks2532]

The "query parameters" would be completely independent of the typical "group by" and "metrics" fields.

[yejianye]

Since almost every type of visualization requires 'groupby' and 'metrics'
parameters, how could we visualize those query-based slices without them?

Jparks2532 notifications@github.com于2016年9月28日周三 上午10:28写道：

The "query parameters" would be completely independent of the typical
"group by" and "metrics" fields.

—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
#1203 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/AADkrK-d9bPzxaOcg1AnVekRwPphhCnxks5quqPUgaJpZM4KH2PG
.

[Jparks2532]

Please don't get too hung up on the example that I provided. It's mainly there to show how you can pass in parameters into an arbitrary query.

[shivaachari]

is this feature implemented in Superset?

[mistercrunch]

Yes pretty much, by following the "Visualize" button in SQL lab you're essentially running the explore view against your SQL statement as a subquery, and this SQL statement can be templated and receive url parameters.