This title kind of implies that PG or MySQL aren't modern or modern enough which i think is is very wrong. Look what they bring in every update. I think they are quite modern!
Not sure if you knew this or not, but Phil Eaton (the OP) writes databases for fun[1]. This post isn't saying Postgres and MySQL aren't modern, but is more trying to get ideas for what to play around with next.
Postgres is very modern and MySQL is also... em... also modern in a different sense. ;-)
I understood the question as:
> What can those newer database systems that are not beholden to RDBMS conventions do for us that Postgres and MySQL (and Oracle and SQL Server and DB2) can not?
The title seems fair to me. Anything that's actually used has certain commitments it made earlier in its lifecycle from which it now can't deviate, even if later developments made the commitments problematic.
Sure. There are probably also things they wish they could deprecate, but realistically can't. User expectations they wish they could replace and can't. I think that's the charitable interpretation of things asked for.
I think it's unfair. "Modern" is in no way equivalent to what the thread is really asking, which is what could you do if you built a new database today with all of the knowledge but none of the existing commitments of a large mainstream database.
It’s fair because these databases made trade offs for older ideas presumably for good reasons. The question isn’t what could the have done differently up to now, but rather what new ideas might designers have included if they existed back then.
Modern software development practices and knowledge are orthogonal to feature set / compatibility guarantees. The title asks about the former and the text asks about the latter, and I don't agree it's correct to conflate the two
Although most of the responses have been for ideas that did exist back then, but have significant trade-offs, both then and now. They're mostly just different approaches to db, not something fundamentally "modern".