> This is needless complexity engineered by people who insist on relearning schemas from scratch, badly, rather than trusting the database people who say "you need to be explicit about the schema, do data modelling up front".
The reason is with some projects/data it's hard to be explicit about the schema which is why NoSQL had it's popularity phase.
Now most applications don't have either entirely structured or entirely unstructured data, they will have a mix - so it's absolutely brilliant for one tool to do both. If they didn't support JSON I have a strong suspicion that they wouldn't have had some of the growth we have seen for Postgres across the last few years.
The reason is with some projects/data it's hard to be explicit about the schema which is why NoSQL had it's popularity phase.
Now most applications don't have either entirely structured or entirely unstructured data, they will have a mix - so it's absolutely brilliant for one tool to do both. If they didn't support JSON I have a strong suspicion that they wouldn't have had some of the growth we have seen for Postgres across the last few years.