add comments

proto/osmosis/twap/v1beta1/twap_record.proto

@@ -93,4 +93,4 @@ message PruningState {
   // last_key_seen is the last key of the TWAP records that were pruned
   // before reaching the block's prune limit
   bytes last_key_seen = 3;
-}
+}

x/twap/listeners_test.go

@@ -275,8 +275,6 @@ func (s *TestSuite) TestAfterEpochEnd() {
 		err = s.App.TwapKeeper.EpochHooks().AfterEpochEnd(s.Ctx, allEpochs[i].Identifier, int64(1))
 		s.Require().NoError(err)
 
-		//recordsAfterEpoch, err := s.twapkeeper.GetAllHistoricalTimeIndexedTWAPs(s.Ctx)
-
 		lastKeptTime := s.Ctx.BlockTime().Add(-s.twapkeeper.RecordHistoryKeepPeriod(s.Ctx))
 		pruneState := s.twapkeeper.GetPruningState(s.Ctx)
 

x/twap/store.go

@@ -80,6 +80,13 @@ func (k Keeper) StoreHistoricalTWAP(ctx sdk.Context, twap types.TwapRecord) {
 // So, in order to have correct behavior for the desired guarantee,
 // we keep the newest record that is older than the pruning time.
 // This is why we would keep the -50 hour and -1hour twaps despite a 48hr pruning period
+//
+// If we reach the per block pruning limit, we store the last key seen in the pruning state.
+// This is so that we can continue pruning from where we left off in the next block.
+// If we have pruned all records, we set the pruning state to not pruning.
+// There is a small bug here where we store more seenPoolAssetTriplets than we need to.
+// Issue added here: https://github.com/osmosis-labs/osmosis/issues/7435
+// The bloat is minimal though, and is not at risk of getting out of hand.
 func (k Keeper) pruneRecordsBeforeTimeButNewest(ctx sdk.Context, state types.PruningState) error {
 	store := ctx.KVStore(k.storeKey)