Again and again, I have seen people here and everywhere advocate for preventing intolerable extensions of the SQL language, this is the latest example. I recall only one article about what I'm going to say, and I no longer have this link.
Did you really benefit from writing portable SQL and rejecting your own dialect tools / syntax?
I have never seen a case when someone is struggling to create a complex mysql application and then says that you know it will be just peachy? Let me switch to (PostGreSQL | Oracle | SQL Server)!
The shared libraries in -say-PHP do the abstract subtleties of SQL, but at what price? Ultimately, you cannot use efficient designs and functions for the supposed glimpse of portability that you are likely to never use. It sounds like a YAGNI tutorial to me.
EDIT: Perhaps the example I mentioned is too shy, but I think the thing remains: if you plan on moving from one DBMS to another, you are likely to reverse engineer the application anyway, or you won’t do it at all.
source
share