You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I have the situation where my dataset is partitioned in S3 based on date, in the format YYYY/MM/DD. My query backend is AWS Athena.
My data has the following noteworthy (for the purposes of this issue) columns:
creation_datetime (UTC ISO-8601 timestamp)
year (partition column)
month (partition column)
day (partition column)
other data columns
When querying data over Superset, I can add a time-range filter to most of my charts and dashboards. I can also, manually, add a filter for my partition columns (year, month, day).
If these partition columns are not explicitly given, then Athena does not submit the query using the 'expected' filtering leading to:
higher costs
longer runtime
due to it not actually using the available partition metadata.
It would be rather tedious for every single user to input these partitioning filters every time they wish to query a dataset, and it is error prone.
My time ranges have millisecond granularity, so it would not be feasible to partition on the time range itself. In fact, the partition metadata is derived from the time range at write-time.
Is it possible to create derived filters from the "time-range" filter, so that whenever the time-range filter is selected, custom filters (in this case year, month, day) populated from the selected time range are also set?
I attempted to do this for a dashboard, but I was unable to select the "Time Range" filter as a parent filter for the parent-child filter functionality described e.g. here (the Time Range filter is simply not an option when selecting which filter to depend on). But ideally, I would like to be able to set this filter extension for an entire dataset (or in some other, global fashion). Is this possible?
The text was updated successfully, but these errors were encountered:
Thanks for the quick response. I assume you are talking about #23319.
While this is a step in the right direction, I don't think this would allow me to be able to dynamically populate my partition filters using just this feature, so I'm probably out of luck for the foreseeable future (at least for a straightforward solution).
I'm not sure if this is still an issue in current versions of Superset (3.x). Dependent filters have come a long way, but I haven't tried this exact use case in recent memory. If it is still an issue, we can re-open this if it's a "bug" but I'd point you to GitHub Discussions (using an Ideas thread) if this is more of a feature request. Closing this for now since it's not an acute bug in 3.x or newer.
I have the situation where my dataset is partitioned in S3 based on date, in the format YYYY/MM/DD. My query backend is AWS Athena.
My data has the following noteworthy (for the purposes of this issue) columns:
When querying data over Superset, I can add a time-range filter to most of my charts and dashboards. I can also, manually, add a filter for my partition columns (year, month, day).
If these partition columns are not explicitly given, then Athena does not submit the query using the 'expected' filtering leading to:
due to it not actually using the available partition metadata.
It would be rather tedious for every single user to input these partitioning filters every time they wish to query a dataset, and it is error prone.
My time ranges have millisecond granularity, so it would not be feasible to partition on the time range itself. In fact, the partition metadata is derived from the time range at write-time.
Is it possible to create derived filters from the "time-range" filter, so that whenever the time-range filter is selected, custom filters (in this case year, month, day) populated from the selected time range are also set?
I attempted to do this for a dashboard, but I was unable to select the "Time Range" filter as a parent filter for the parent-child filter functionality described e.g. here (the Time Range filter is simply not an option when selecting which filter to depend on). But ideally, I would like to be able to set this filter extension for an entire dataset (or in some other, global fashion). Is this possible?
The text was updated successfully, but these errors were encountered: