Even those reasons are really stretching for it these days.
There are numerous ways of dealing with hierarchical or otherwise structured data using relational DBs. Support for recursive CTEs is becoming more common, which makes such querying quite clean and simple. And with many relational systems offering support for storing and manipulating XML and JSON data, storing more structured data becomes a non-issue.
I think that SQL managing to hide a lot of the underlying implementation details is one of its strongest points. In many situations the defaults are more than sufficient, saving developers and users a significant time. Yes, there are situations where more control over the exact handling is needed, but it gives ways of managing this. At least a simplified, high-level abstraction is offered; many NoSQL systems immediately force you to take a very low-level approach for even the simplest of queries.
And I don't see the ability to intentionally lose data as something that's good, either. The database should be concerned with doing everything in its power to protect and ensure the integrity of the data it has been given. If it's fine to lose some of that data, then whatever or whoever is providing that data should just not provide it to the DB in the first place.
So while some people may present those as possible disadvantages of SQL, I just don't think the arguments really hold up upon further scrutiny.
There are numerous ways of dealing with hierarchical or otherwise structured data using relational DBs. Support for recursive CTEs is becoming more common, which makes such querying quite clean and simple. And with many relational systems offering support for storing and manipulating XML and JSON data, storing more structured data becomes a non-issue.
I think that SQL managing to hide a lot of the underlying implementation details is one of its strongest points. In many situations the defaults are more than sufficient, saving developers and users a significant time. Yes, there are situations where more control over the exact handling is needed, but it gives ways of managing this. At least a simplified, high-level abstraction is offered; many NoSQL systems immediately force you to take a very low-level approach for even the simplest of queries.
And I don't see the ability to intentionally lose data as something that's good, either. The database should be concerned with doing everything in its power to protect and ensure the integrity of the data it has been given. If it's fine to lose some of that data, then whatever or whoever is providing that data should just not provide it to the DB in the first place.
So while some people may present those as possible disadvantages of SQL, I just don't think the arguments really hold up upon further scrutiny.