There’s been attempts to revise the SQL standard to accommodate various types of temporality as a “first party” feature.
I think that we won’t see traction at the RDBMS “kernel space” until it’s in the SQL standard. There are many valid and complex options to choose from, and there are successful solutions in user space that aren’t overly burdened, performance-wise, from being in user space.
FWIW, the “audit table” approach is the approach that people who study this field gravitate towards. Mainly because it maintains consistent ACIDity in the database, and maintains Postgres as the single point of failure (a trade off vs introducing a proxy/polling job).
I think that we won’t see traction at the RDBMS “kernel space” until it’s in the SQL standard. There are many valid and complex options to choose from, and there are successful solutions in user space that aren’t overly burdened, performance-wise, from being in user space.
FWIW, the “audit table” approach is the approach that people who study this field gravitate towards. Mainly because it maintains consistent ACIDity in the database, and maintains Postgres as the single point of failure (a trade off vs introducing a proxy/polling job).