I’m trying to use the new Custom Datasource API, however, I’m still finding limitations that are making my project difficult to achieve. Specifically, it doesn’t support custom aggregations so I can’t utilize all the features of my datasources and APIs. Right now I’m working on Percentiles, but later I’ll likely need more aggregations.
Here is what i want things to look like in one of my simple pivot tables:
– Date (year/month/day)
Values: min, p50, p99, max
Then I would obviously like to graph this, so I can see percentile changes over time. I also sometimes will want to have “domain” and then “subdomain” both in rows as a second pivot table.
Thank you for reaching out to us.
Add the custom server-side logic that would scan the
uniqueName property of each received measure. In case the
uniqueName matches the measure that needs to be aggregated in the custom way, apply the desired aggregation.
Also, we recommend modifying the response on the
/fields request in the way the measure mentioned earlier does not have any available aggregations (
"aggregations": ["none"]). It is required in order to restrict the user to change the aggregation through UI (the chosen option will be ignored anyway). Please note that such a workaround limits the number of possible aggregations to the one specified on the server.
Detailed information about the
/fields request and the structure of the response on such request can be found in our documentation.
Our team hopes it works for your case.
Please let us know in case you have any additional questions on this point.