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
One of the first steps for adding full support for ARRAY columns (#2115) is the creation of an RFC for the KV encoding of ARRAYs. Possible considerations are:
Future-proofing. It'll be significant effort to change the encoding later
Efficient support for expected access patterns (see this comment on 2115 for some ideas)
Multi-dimensional arrays
cc @jordanlewis@knz for other thoughts regarding what should go into the RFC
The text was updated successfully, but these errors were encountered:
Note that the kV indexing of arrays and that of array indexes are two fundamentally different problems and I suspect there will be not one but multiple index types for array columns depending on the queries that need to be satisfied.
--
Sent from my Android device with K-9 Mail. Please excuse my brevity.
@knz I believe the focus of the RFC is encoding ARRAYS into values, not keys (which would be necessary to use an ARRAY column in an index).
A big question to answer is if an ARRAY is encoded into a single KV or if it can be split across multiple KVs. The former would place limits on the size of the ARRAY. The latter introduces complexity and questions about how we specify (or heuristically determine) where the breakpoints are performed.
One of the first steps for adding full support for ARRAY columns (#2115) is the creation of an RFC for the KV encoding of ARRAYs. Possible considerations are:
cc @jordanlewis @knz for other thoughts regarding what should go into the RFC
The text was updated successfully, but these errors were encountered: