-
Notifications
You must be signed in to change notification settings - Fork 69
Description
Seems there is no support of the non-equiwidth partitioning, i.e. we know we have say 100M items per partition and we know start, end of a bigint value but there's no interval as such. This happens quite often in scientific domains (i.e. keys based on healpix indexing in astronomy, scarcely sampled time domains, molecules encoding etc), everywhere where uneven sampling is common.
add_range_partition could look like a saver, but it also relies on interval type.
Could you please add something like:
create_partitions_from_range( relation REGCLASS, attribute TEXT, start_range ANYARRAY, end_value ANYARRAY, partition_data BOOLEAN DEFAULT true)
where start_, end_range are arrays of boundaries of partitions?
or even arrays based on existing range types, int8range[] etc?
This would be a very welcomed addition.
Do you have any plans to include it?