+1 for this feature, which we've needed for a decade or more. :-)
It would be nice to add some (brief is fine) commentary about things enabled by the new kind of index, including queries against heterogeneous fields (e.g., searches for price = 3.99 or price = "3.99" with one index under the hood), nested-loops index joins (without having to type-cast to make an index usable), and also index-only query plans.
Cheers, Mike On 10/22/24 9:51 PM, Shahrzad Haji Amin Shirazi wrote:
Initiating the discussion thread to propose the creation of the Heterogeneous Index in AsterixDB. *Feature:* Heterogeneous Index *Details:*In the current version of AsterixDB, creating a secondary index on a field requires that the field's type be either defined within the data type or explicitly specified in the index DDL. We are planning to introduce the Heterogeneous Index in AsterixDB, which will allow you to create an index on a field without the need to define its type. This means the field’s type will no longer need to be specified in either the data type or the index creation DDL. *Changeset*:https://asterix-gerrit.ics.uci.edu/c/asterixdb/+/18522 *APE*: https://cwiki.apache.org/confluence/display/ASTERIXDB/APE+10%3A+Heterogeneous+Indexes
