This is a library for manipulation data in a DB, in which all data structures and data flow are specified once declaratively, and then data updates are simply applied, which the library distributes over this structure in accordance with the declaration.
At the same time, it takes into account the statistics of the use of this data and, on its basis, can optimize the actual data flow, choosing the degree of aggressiveness of materialization of secondary data structures.
We have comments table with next user fields:
id- comment idtopic_id- id of the commented topicauthor_id- comment author id
And next automatic fields:
likes_cnt = comments_votes[ id = id, is_like = true ].count- likes count.dislikes_cnt = comments_votes[ id = id, is_like = false ].count- dislikes count.
Then, if there are much more writes into the comments_votes table than the reads from comments, then the automatic fields likes_cnt and dislikes_cnt will stop materializing, and will be calculated by a separate query to the comments_votes table at each reading.
And if there are much more reads from the comments table than the writes into comments_votes (which is most often the case), then the automatic fields likes_cnt and dislikes_cnt will be materialized. Each time comments_votes changes, comments will be updated.
The key feature of StateFlow will be the ability to recognize most effectively update policies for all data. Here we can use a lot of different interesting dataflow statistics analysis.
It is planned to implement a dataflow dashboard panel, in which it is convenient to graphically represent the dataflow and present the key statistics data.