> My question is: why would anyone want to start something new with PHP in 2021? Is there anything that it does better than any other language/ecosystem?
Well, on Twitter the other day, I asked, only partly tongue-in-cheek:
> Even in 2021, any new server-side web app with even a vain hope of being installed by people who are tech savvy but not modern web nerds should probably still be written in PHP. True or false?
The "false" responses cited things like "it doesn’t matter what you write it in as long as it runs cleanly in a Docker container" and "Node is widespread enough I think it goes beyond 'modern web nerd' at this point," and they might be right -- but I'm not certain they are.
So, I guess I'd suggest that maybe the answer to your first question ("why start something new with PHP in 2021") doesn't actually depend on getting an unqualified "yes" answer to your second question ("is there anything it does better"). If I was starting a project that I wanted to have installed by the kinds of people who will not use the word "deploy" instead of "install," if you take my meaning, then I'd at least seriously entertain whether it should be in PHP.
Now, for the kinds of projects I think get talked about on HN (e.g., actual paid jobs), this is more of a fair question. I appreciate Laravel and Symfony, but PHP in 2021 has traded the feeling of being a cargo cult version of Perl for the feeling of being a cargo cult version of Java. If I do write a new project in PHP, it's probably not going to be with any existing framework, because even the lightest one feels like endless mazes of "get a Request object from a RequestFactory by instantiating a dependency injection container and passing it to a RequestFactoryFactory".
> I appreciate Laravel and Symfony, but PHP in 2021 has traded the feeling of being a cargo cult version of Perl for the feeling of being a cargo cult version of Java.
As a Symfony dev I agree. And the process to convert PHP into a Java-esqe Kotlin-esque language is continuing. Why not use a different language if you like it rather than change PHP?
> If I do write a new project in PHP, it's probably not going to be with any existing framework, because even the lightest one feels like endless mazes of "get a Request object from a RequestFactory by instantiating a dependency injection container and passing it to a RequestFactoryFactory".
I'll be interested to hear what you come up with. Two anecdotes:
1. I had to touch some PHP code I wrote many years ago based on the ColdFusion "Fusebox" architecture. It's missing some nice abstractions but unlike Symfony et al there was so little code I could understand what every line did within minutes. Parts of Symfony's internals are still magic to me (complicated by compiler passes).
2. I was asked to look at replacing a decade-old internal PHP app, the kind people look down on built using jQuery UI AJAX-powered datagrids and raw PHP. There were multiple problems like SQL injection but the app responded incredibly fast. I had to warn them that the replacement would feel slower unless they drastically increased the hosting, because a modern, engineered framework would've only finished booting up by the time this one was returning data.
"the process to convert PHP into a Java-esqe Kotlin-esque language is continuing"
My observation is also that PHP is growing too much in language size. Depending on one's viewpoint, this can be seen as either matching the feature and syntax bloat of other languages, or taming the language to make it "saner".
The beauty of PHP is that you can still choose how complex you want your application to grow - and that you can adapt if needs change.
You can do everything from a one-file hacked together shell scripts (which I love to do because bash is just plain annoying sometimes, especially for stuff that has to work on OS X and Linux with their not-so-small differences in coreutils) to enterprise-style architected software in Symfony or Laravel.
> As a Symfony dev I agree. And the process to convert PHP into a Java-esqe Kotlin-esque language is continuing. Why not use a different language if you like it rather than change PHP?
The audacity of PHP 5.3 developers. How dared they!
The audacity of PHP 5.4 developers. How dared they!
The audacity of PHP 5.5 developers. How dared they!
The audacity of PHP 5.6 developers. How dared they!
The audacity of PHP 7.0 developers. How dared they!
The audacity of PHP 7.1 developers. How dared they!
The audacity of PHP 7.2 developers. How dared they!
The audacity of PHP 7.3 developers. How dared they!
The audacity of PHP 7.4 developers. How dared they!
The audacity of PHP 8.0 developers. How dared they!
The audacity of PHP 8.1 developers. How dared they!
/s
(sorry, couldn't help. this particular argument is sooooo silly...)
> Why not use a different language if you like it rather than change PHP?
I don't disagree in principal, but I'm struggling to think of much that isn't totally optional. For example, you can still completely ignore the type system.
There have been some BC breaks lately, but it's mostly around footguns that should never have been like that in the first place.
Well, on Twitter the other day, I asked, only partly tongue-in-cheek:
> Even in 2021, any new server-side web app with even a vain hope of being installed by people who are tech savvy but not modern web nerds should probably still be written in PHP. True or false?
The "false" responses cited things like "it doesn’t matter what you write it in as long as it runs cleanly in a Docker container" and "Node is widespread enough I think it goes beyond 'modern web nerd' at this point," and they might be right -- but I'm not certain they are.
So, I guess I'd suggest that maybe the answer to your first question ("why start something new with PHP in 2021") doesn't actually depend on getting an unqualified "yes" answer to your second question ("is there anything it does better"). If I was starting a project that I wanted to have installed by the kinds of people who will not use the word "deploy" instead of "install," if you take my meaning, then I'd at least seriously entertain whether it should be in PHP.
Now, for the kinds of projects I think get talked about on HN (e.g., actual paid jobs), this is more of a fair question. I appreciate Laravel and Symfony, but PHP in 2021 has traded the feeling of being a cargo cult version of Perl for the feeling of being a cargo cult version of Java. If I do write a new project in PHP, it's probably not going to be with any existing framework, because even the lightest one feels like endless mazes of "get a Request object from a RequestFactory by instantiating a dependency injection container and passing it to a RequestFactoryFactory".