This subthread is about the future, not the past. 8 years ago, PG didn't have json support so your point is moot.
I'm very curious about what people think about the future of Mongo, independently and particularly in comparison to Postgres. However every time that comes up, people keep bringing up that Mongo was a buggy piece of crap in some irrelevant past. So what?
> This subthread is about the future, not the past. 8 years ago, PG didn't have json support so your point is moot.
8 years ago PG did have replication though, so not sure why it not having feature X 8 years ago makes my point moot.
People keep bringing up that it was a buggy piece of crap because its the icing on the cake and pretty much something you never want your database to be, past or present. Not that software configured by default to eat your data and not persist it can be called a database mind you.
I don't see this as a problem. It takes years for any software project to mature, a DBMS even more so. I'm sure that I can go back to the 1980s and find gamebreaking bugs in the original POSTGRES. It has been years for MongoDB to approach maturity.
Of course I would prefer Postgres when I can use it, and I can generally use it basically all the time, but NoSQL still has its use cases.
Synchronous replication was added in 9.1 and much improved in 9.6? pglogical[0] works pretty well for me under 10 but I have no production experience with bdr[1].
IMO, sure but it's far from seamless. (I also looked at pg's quorum commits, but the same applies.)
In general Postgres was not designed at its core for a distributed world. Even now, replication feels like an afterthought in the grand scheme of things, and sharding nonexistent without extensions.
Tell me again how Postgres can seamlessly do horizontal scaling and synchronous replication?