And that is (part of) why RDB "failed" to keep people even for use-cases where they should be better then NoSQL.
- To much hidden complexity which can leaf to bugs if you don't know about the hidden complexity.
- To messy and/or misunderstood concurrency.
- Many People having fundamental misunderstandings about how to do things and how far transactions are useful for given (often implicit default) options.
One thing I had to learn the hard way is that many programmers which supposedly "know" about RDMS/SQL don't even understand isolation levels properly.
Also is it just me or is this way easier in postgres (with on conflict ...).
- To much hidden complexity which can leaf to bugs if you don't know about the hidden complexity.
- To messy and/or misunderstood concurrency.
- Many People having fundamental misunderstandings about how to do things and how far transactions are useful for given (often implicit default) options.
One thing I had to learn the hard way is that many programmers which supposedly "know" about RDMS/SQL don't even understand isolation levels properly.
Also is it just me or is this way easier in postgres (with on conflict ...).