Various fixes to PR

erezsh · erezsh · commit 1838dd3cd1e3 · 2022-05-28T11:50:59.000+02:00
diff --git a/data_diff/diff_tables.py b/data_diff/diff_tables.py
@@ -120,7 +120,10 @@ def count_and_checksum(self) -> Tuple[int, int]:
         )
         duration = time.time() - start
         if duration > RECOMMENDED_CHECKSUM_DURATION:
-            logger.warn(f"Checksum is taking longer than expected ({duration:.2f}s). ")
+            logger.warn(
+                f"Checksum is taking longer than expected ({duration:.2f}s). "
+                "We recommend increasing the bisection-factor."
+            )
 
         # TODO Handle None TODO
         return count or 0, checksum
@@ -243,14 +246,14 @@ def _bisect_and_diff_tables(self, table1, table2, level=0, max_rows=None):
     def _diff_tables(self, table1, table2, level=0, segment_index=None, segment_count=None):
         logger.info(
             ". " * level
-            + f"Diffing segment {segment_index}/{segment_count}, key-range: {table1.start_key:x}..{table2.end_key:x}"
+            + f"Diffing segment {segment_index}/{segment_count}, key-range: {table1.start_key}..{table2.end_key}"
         )
 
         (count1, checksum1), (count2, checksum2) = self._threaded_call("count_and_checksum", [table1, table2])
 
-        if not (count1 or count2):
+        if count1 == 0 and count2 == 0:
             logger.warn(
-                "Uneven distribution of keys detected. "
+                "Uneven distribution of keys detected. (big gaps in the key column). "
                 "For better performance, we recommend to increase the bisection-threshold."
             )
             assert checksum1 is None and checksum2 is None
diff --git a/dev/prepare_db_gaps.pql b/dev/prepare_db_gaps.pql
@@ -71,6 +71,10 @@ print "Create more tables"
 const table rating_gap1_update0001p = rating_gap1
 const table rating_gap2_update0001p = rating_gap2
 const table rating_gap3_update0001p = rating_gap3
+create_indices(rating_gap1_update0001p)
+create_indices(rating_gap2_update0001p)
+create_indices(rating_gap3_update0001p)
+
 rating_gap1_update0001p[random() < 0.000001] update {timestamp: timestamp + 1}
 rating_gap2_update0001p[random() < 0.000001] update {timestamp: timestamp + 1}
 rating_gap3_update0001p[random() < 0.000001] update {timestamp: timestamp + 1}
