Author Topic: Hash tables for ultra fast dictionaries  (Read 4373 times)


  • Member
  • ***
  • Posts: 274
Re: Hash tables for ultra fast dictionaries
« Reply #60 on: February 08, 2022, 10:21:57 PM »
i'm comprehended. about the speed and the memory, we'll get both of them is impossible. either loss the speed or loss the memory.
1, Indices_Array is stored by the DictObjectIndex, Entries_Array is stored by the CollisionIndex, Collision_Array is stored by the CollisionIndex.  If the actual address is stored, this memory database backup and recovery cannot work.
2, Why did the hashtable still use comparison instructions? Due to the possibility of collision at any time, the target dictobject item must be confirmed by the szStrCmp.
3, When deleting a DictObject item, keep nextaddr in there. otherwise the openning address chain will break, the DictObject item in this openning address chain will be lost. If the nextdatas move a bit by bit, it will take a lot of time.
4, Replacing index zero with "- 1" is to distinguish between initialization "0" and written "0".

« Last Edit: February 10, 2022, 03:53:15 PM by six_L »
Say you, Say me, Say the codes together for ever.