The complainer lacks a full understanding or appreciation of set and relational theory

The problem with SQL is that it requires "a full understanding or appreciation of set and relational theory" to use it properly, but offers itself to the world through it's "friendly" declarative syntax.

It encourages people to think that they are competent because their queries return the results they expect. What they often do not understand is the costs that it went through to produce them, nor how much their query was optimised for them under the covers (in the case of the more sophisticated implementations--usually the commercial ones).

Perl has a similar problem in that it's accessibility make it easy to 'get something to go' for the beginner without grounding, but without the grounding, they do not appreciate the redundancies and inefficiencies in how they get the results.

SQL as a language is ill-fitting to the operations of set theory, with multiple, verbose, and often awkward and ambiguous ways of specifying relationships that would be concise and unambiguous in set notation. Many years ago I was trying to optimise a complex select across a 7-way join of some pretty large tables (for the time!). Consulting a DB2 expert from IBM he showed us an APL statement that did the equivalent selection; it consisted of about 15 to 20 characters. In total.

If you think I'm kidding, go here and scroll down to table under the heading "How Does APL Compare with other Languages".

I never learnt APL, but I've often wondered what could be achieved with if RDBMSs would accept it as an alternate syntax. Given that Unicode now supports the APL character set, maybe it's time for APL's resurrection :)