AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |
Back to Blog
Azure data studio sqlite11/25/2023 We have to agree that manual implementation of it is really a tedious task. I am only moving away from RDBMS only because of the sync issue. Follow your favorite evil corp's engineering guideline on how to make a piece of shit system.Whether the local data should be in an SQLite db depends on the data (but very likely the answer is yes). Being able to perform some operations locally and submit them as a batch sounds great to me.Don't try a direct sync of "this data from localdb into remotedb" it's only going to lead to trouble. Then apply them in an atomic transaction. Transfer an array of change operations from the client to the server.To answer the questions based on the approach above: Since the server is validating the lawfulness of the operations there can be no cheating. This also solves the issue of malicious client sending potentially fake state. You'll probably want client and server to share the logic of applying the state changes to avoid one of them updating state differently than the other. operations) from client to server and applying the changes in the server. If you go this direction the implementation of the sync becomes a matter of sending the logs (i.e. Of course this assumes that two clients cannot perform conflicting changes while offline (or that you're willing to fail one client's commit). If this is the case, it reminds me a lot of a log structured file system (sqllite changes) with serialized state as the starting position (state in azure): Correct me if I'm wrong here, but it sounds like your client can perform local operations (stored in sql lite) and at a later point commit those to the global state (azure db).
0 Comments
Read More
Leave a Reply. |