Why does pgsync need to both pull from db and read from the replication logs?

[jeremykhon]

I'm just reading these two lines from the pull function in pgsync/sync.py and just wonder why isn't the self.sync function enough? why would there be anything missed if self.sync will query everything from txmin to txmax? Why do we need to call self.logical_slot_changes to sync everything from txmin to txmax again?

def pull(self) -> None:
        """Pull data from db."""
        txmin: int = self.checkpoint
        txmax: int = self.txid_current
        logger.debug(f"pull txmin: {txmin} - txmax: {txmax}")

        # forward pass sync
        self.es.bulk(self.index, self.sync(txmin=txmin, txmax=txmax))

        # now sync up to txmax to capture everything we may have missed
        self.logical_slot_changes(txmin=txmin, txmax=txmax, upto_nchanges=None)

        self.checkpoint: int = txmax or self.txid_current
        self._truncate = True

Thank you for working on this package

[toluaina]

logical_slot_changes is necessary because if pgsync was stopped for some time and during that period, changes were made to the database.
in this case, there is no way for pgsync to know e.g what deletion operations were made