On: PostgreSQL 11 and Just in Time Compilation of Queries

[allan_s]

honest and naive question:

What would be the difficulties to have these in PG, outside of "getting developer interested in implementing it" (i.e is there some current architecture decision in PG that prevent it, like it's the case for "why vanilla PG is not suitable as a column store")

[SigmundA → allan_s]

My understanding is gin/gist indexes do not store enough information to use better scoring and RUM indexes could change that:

https://github.com/postgrespro/rum