Hacker News Reader

Comment by rp1 | Hacker News Reader

I haven’t used PHP in over a decade. I can never forgot how much PHP got wrong in terms of how application development on the internet would happen. The ergonomics of using it were just… bad compared to other frameworks that came on the market. I remember when rails came out and it was so mind blowing compared to PHP. The language was certainly useful for a time, and I’m sure the language has progressed since, but switching to PHP still feels like it would be a huge step backwards.

View on HN

Comparing Rails to PHP is unfair given that one is a framework and one is a programming language. Compare PHP and Ruby or Rails and Laravel to keep things apples to apples and oranges to oranges.

A lot of people 10+ years ago were building websites by interpreting PHP and using the output as HTML. The language (at the time) seemed to encourage this. I agree that PHP is not a framework, but it seemed to have some opinion on how HTML was to be generated, which most languages don’t have. Also rails is 7 years older than Laravel, so my point still stands that PHP was behind the curve.

If your standard is rails then literally everything in every language before rails was "behind the curve". Currently PHP has frameworks that have caught rails, but Ruby's perf is still garbage.

Django was released shortly after Rails. Ruby and Python developers _needed_ to build something like Rails and Django because those languages didn’t promote the same type of script-as-html pattern that PHP did. Had more thought gone in to serving websites, PHP might enjoy more popularity today. PHP had the first mover advantage after all, but frameworks built in other languages were purely better in almost all categories.

10+ is not that long time ago. There was already Zend Framework. And for some basic API using framework may be overkill anyway.

I'd be interested to know what you think PHP got wrong with respect to web development? The majority of the web is running on PHP to this day, so at least in terms of popularity it seems people don't feel this way generally. Furthermore, if one were philosophically-minded, it could be argued that PHP's "shared nothing/one thread per page" execution model is the very quintessence of HTTP.

It‘s also the essence of "serverless" so that model seems almost prescient.

Except Erlang (and nowadays Elixir as well) has that for 30+ years already and it's done much better -- one green thread per request, and you can have 200,000+ of them at the same time on some fairly modest hosts, without any of them stealing run time from the others (as much as the hardware allows, of course).

PHP's "prescient" model demanded one OS process per request which is frankly absurd and I don't get how anyone views PHP in serious light because of this single fact alone.

Now think about how many companies/products need the power of Erlang vs PHP where PHP has more tooling, better ecosystem and far more available talent to choose from. Just because Erlang can perform better doesn't mean it is the right tool for the job. Performance is one aspect and PHP is good enough for lot of use cases while it has tons of other advantages that Erlang doesn't.

Not performance per se. It's a much more robust model of work that in addition is sipping hardware resources more efficiently (so DoS attacks from one user to all others are hard).

"How many companies need X" is not a discussion, it's an exchange of opinions and won't ever go anywhere, so I refuse to start it.

I was merely responding to the claim that PHP had "prescient" ideas. It didn't.

I’m curious what metric you’re using to determine that a majority of the web is running on PHP.

Surveys generally indicate that of sites where it is possible to determine the language, somewhere between 70% and 80% of them are running PHP. Of course, that figure will include a lot of WordPress sites.

What percentage of websites is it possible to determine the language? My guess is not many. Your metric self-selects for recognizing PHP sites because of Wordpress.

What alternative language do you propose? Not that asking for a source is unreasonable, but you seem to doubt that PHP has majority market share. Do you suppose that Node is more popular? Rails? Go? Flask?

Php has developed a lot in a decade.

Lots of new good things in php7/php8. The typecasting is way better than before, but it still allows you to be more "dynamic" if you want to.

   $x = [1,null,'xxxx', collect([])];

gettype($x) is "array". you can annotate a type as being an `array`. How does that help me exactly?

Since $x is an array, it will get rejected at runtime by `function f(int $a)`. So this `array` type is limited but useful.

You can also add annotations like `/* @var []int */`. External tools (psalm, phpstan) use annotations in their static analysis of the code and will raise an error if $x elements are not integers.

Of course, it's far from Haskell, but my experience with types in PHP is smoother than in Python. Though it was 2 years ago in Python, and the environnement has probably matured.

In my opinion, a worse problem with PHP is that classes properties are dynamic. `class A {}; $a=new A; $a->x=1;` is perfectly valid and will add a property to the object that does not exist in the class. There's no simple way to forbid this, even at runtime (hacking the magic `__set()` creates other pain points).

> In my opinion, a worse problem with PHP is that classes properties are dynamic.

There's a current RFC that is aiming to deprecate this 'feature'.

https://wiki.php.net/rfc/deprecate_dynamic_properties

Reply on news.ycombinator.com