[matthewmacleod]

That is a weird middlebrow dismissal of a response with essentially no value to anybody.

Use an ORM when appropriate; when using one, follow its conventions. Don't use an ORM if it's not appropriate. This is much better advice.

[hluska]

I'm north of 40 now, so while I both love and agree with your advice, I'd like to add a caveat. Even if you are lucky enough to be able to use an ORM for your entire career, you will still be well served by learning and truly understanding SQL.

In that sense, I'd argue that an ORM is only appropriate if your understanding of SQL and database design are strong enough that you can understand what's happening under the hood if/when everything goes to shit!! :)

[matthewmacleod]

I absolutely agree with that!

An ORM is an abstraction of a concrete system. For very simple use cases, that’s fine. For anything even slightly complex, the ORM becomes a tool that you can use to enhance code readability, or reliability, or modularity. But it’s essential to know what is going on underneath before using that abstraction, much like most other systems. There are too many times I’ve witnessed a less experienced developer build a shockingly expensive n+1 query using an ORM to doubt that :)

[nurettin]

Solving problems with Postgres is wildly different than solving them with mssql informix or Oracle. So I would say "there is no SQL".

[hluska]

I don't agree with that. Once you understand the basics of building performant databases/queries, that knowledge will persist across databases. There are small differences in syntax and features, but any remotely competent person should be able to pick those up while they go, provided they understand what happens under the hood when they execute SQL.

[irrational]

In what way are they wildly different? I spent 15 years working in Oracle and the last 2 years working in Postgres. They are slightly different (Postgres adheres to the standard much more closely than Oracle), but 95% of my knowledge of SQL transferred from one to the other.

[nurettin]

A simple example would be "how to select groupwise maximum" where Postgres would respond well to a correlated subquery and Oracle would respond well to a window function and derived table. The engines and optimizations are different.

[deburo]

>Use an ORM when appropriate; when using one, follow its conventions. Don't use an ORM if it's not appropriate.

It's just as vague. If you know of a good article that goes over the details, that'd be cool :^)