map: When removing a key, don't NULL the entry, but mark as deleted.

pfalcon · pfalcon · commit 4a088f4b6168 · 2014-04-05T05:11:12.000+03:00
When searching next time, such entry should be just skipped, not terminate
the search. It's known that marking techique is not efficient at the presense
of many removes, but namespace usage should not require many deletes, and
as for user dictionaries - well, open addressing map table with linear
rehashing and load factor of ~1 is not particularly efficient at all ;-).
TODO: May consider "shift other entries in cluster" approach as an
alternative.
diff --git a/py/map.c b/py/map.c
@@ -127,17 +127,20 @@ mp_map_elem_t* mp_map_lookup(mp_map_t *map, mp_obj_t index, mp_map_lookup_kind_t
                     mp_map_rehash(map);
                     // restart the search for the new element
                     pos = hash % map->alloc;
+                    continue;
                 } else {
                     map->used += 1;
                     elem->key = index;
+                    elem->value = NULL;
                     if (!MP_OBJ_IS_QSTR(index)) {
                         map->all_keys_are_qstrs = 0;
                     }
                     return elem;
                 }
-            } else {
+            } else if (elem->value == NULL) {
                 return NULL;
             }
+            // Otherwise it's just entry marked as deleted, so continue with next one
         } else if (elem->key == index || (!map->all_keys_are_qstrs && mp_obj_equal(elem->key, index))) {
             // found it
             /* it seems CPython does not replace the index; try x={True:'true'};x[1]='one';x
@@ -152,14 +155,15 @@ mp_map_elem_t* mp_map_lookup(mp_map_t *map, mp_obj_t index, mp_map_lookup_kind_t
                 retval->key = elem->key;
                 retval->value = elem->value;
                 elem->key = NULL;
-                elem->value = NULL;
+                // elem->key = NULL && elem->value != NULL means "marked deleted"
+                // assume value indeed never NULL
                 return retval;
             }
             return elem;
-        } else {
-            // not yet found, keep searching in this table
-            pos = (pos + 1) % map->alloc;
         }
+
+        // not yet found, keep searching in this table
+        pos = (pos + 1) % map->alloc;
     }
 }
 
