Like many, I've grown up with PHP, and grown to dislike it big time over the years. However, it has its virtues. Now, my rule is: use PHP iff it feels like the entire script could comfortably fit inside one small PHP file. The moment my very first design feels like it's better to, say, separate the view from the controller, use a real language. Sometimes this estimation is wrong, and I end up rolling my yet-another-PHP-microframework over again, but usually this is a pretty decent rule of thumb. I don't want to have to go through setting up a VPS and a rails environment for something that'd take me the same time to coe and deploy in PHP.
Really, PHP is excellent for what it was originally designed for: little server-side scripts that do relatively simple things. I know that its authors would currently like it to be used for more than that, but, well, don't.
As long as you're not making a massive application, PHP's massively filled global scope, decent batteries-included approach, documentation-with-examples, and its support on virtually every web host (and then some) make it an excellent choice.
Therefore, I disagree with the author's notion that it's bad for everything. It's excellent for little scripts, which is was mr Lerdorf originally intended it to be for anyway.
I used to concede that PHP is appropriate for little scripts, but I changed my mind over time, because so few of those scripts stay "little". Where is the line drawn? Do we find we need a new feature one day, decide that's enough to tip the scale, and set out to rewrite the thing—wasting all the time purportedly saved by writing it in PHP in the first place? We're lazy, so probably not.
I tried writing a little one-off Python thing in Flask last year and was actually surprised by how little effort it took. I spent more time learning Flask (and Flask is really, really simple) than I did actually deploying the thing to a VPS with spare cycles. It could stand to be slightly more turnkey, but running a tiny app server isn't nearly the nightmare it's made out to be.
Really, PHP is excellent for what it was originally designed for: little server-side scripts that do relatively simple things. I know that its authors would currently like it to be used for more than that, but, well, don't.
As long as you're not making a massive application, PHP's massively filled global scope, decent batteries-included approach, documentation-with-examples, and its support on virtually every web host (and then some) make it an excellent choice.
Therefore, I disagree with the author's notion that it's bad for everything. It's excellent for little scripts, which is was mr Lerdorf originally intended it to be for anyway.