From a “purely philosophical, purely language-design, purely theoretical” standpoint ... (hey, it’s alcohol-time in my time-zone, so kindly hear me out)   ;-) ...

A significant notion in all(!) of these various arguments just might be:   “just how much language-functionality ought to be ‘in the language™?’ ”

The ls command provides some insight:   the perl executable is many times smaller than either python or php.   (In fact, of the three, perl is positively tiny.)

With perl, the basic approach is:   “use it (whatever ‘it’ is on your system ...), or, it’s not there.”   A statement such as use DBI; is, in fact, utterly dependent on precisely which version of DBI is currently installed upon this particular computer system, such that it might will have an altogether-different functional meaning on another machine.

Per contra, both python and php took more-or-less a different approach.   There is more “built-in,” such that “exactly what is ‘built-in’” is more or less consistent between “language versions.”

Which of these approaches is “right” vs. “wrong?”   Obviously, this is not the right question.   Obviously, both approaches work.   But I think that it does serve to point out that, when we compare these various programming languages (and, programming approaches ...) to one another, we are truly (and so, irrelevantly ...) comparing apples to oranges.   There’s more than one way to do it.™   And, all three of them await our bidding at the command-line.