The whole "no build step" was the most underrated factor for its success, because it was such a low-skill tool to pick up.
The tightness of that edit-test-move loop was why I used it, even if the "edit" included "sftp it to a vhost".
That bit is what made it successful & I can't think of anything I've used in the last decade which felt as easy as PHP for that particular style of self-taught programmer.
I have a hard time with people joking about PHP because while it wasn’t perfect back in the day, it’s probably the culminating point of the non elitist computing. PHP was accessible to anyone who had the basics of creating an HTML website (which was a skill any random person could acquire in a day or two).
Also, the people making those jokes are most of the time working on projects with micro services and SPAs everywhere for a boring business app with CRUD and forms. Which I find pretty ironic since a single PHP monolith solved this exact use case 20 years ago.
Nowadays we tend to bring up a whole virtual OS.
An interesting direction in this area might be GraalOS, which might allow running multiple applications in a single OS process, doing the necessary isolation itself. (Not affiliated)
PHP has libraries for compiling SCSS and minifying JS, and filemtime() is very fast. So if you wire them together in just the right way, you can edit any script or stylesheet under the project hierarchy and have them compiled, minified, and bundled together on the fly. Cache the bundle until you edit one of them files again, and automatically bust the cache once you do.
No need to build anything whatsoever, both on the frontend and on the backend. It's like REPL over HTTP.
If you do compiling and minifying, that's a build step.
Web browsers can't understand SCSS natively.
To be truly extreme (performance), devs should consider tree-shaking unused CSS and JavaScript. This may require using Vite (Rolldown is in the works/Rollup) if there are no other options or it makes no sense to serve bloated resources. That's the reason, I chose Astro as I care about users, not prioritse the developers' minimalistic workflows.
I prefer Tailwind because it allows me to avoid duplicating classes (mine and vendors), which is an advantage over other approaches.
I did dogfooding in Go development.
I use this in production for a system with 3k requests per second that runs in a single Apache+PHP web server and a single 80GB MySQL database. Never needed to optimize anything.
You have no idea how happy it makes me to see someone else was bold enough to try this and it worked for them too.
Simple, maintainable, instant SCSS and JS build pipelines with all the benefits and no drawbacks! Plus any junior can understand and change it in minutes.
While it was always easy to debug java based apps and the debug support is great (remote connection, drop frames, all types of breakpoints etc.) people struggled with this in php.
I showed a professional a few years back how to debug with php, after 2 weeks he stoped using it...
To build anything useful you quickly discover that it's hard to build from source (both the lang and extensions), some parts unmaintained for years. There are many counterintuitive idiosyncrasies you must deal with to be productive. The ecosystem suffers from the overuse of indirection and reflection. Php only looks easy, but the rabbits hole goes deeper than most other langs.
Can you give examples of counterintuitive idiosyncrasies you must deal with to be productive?
Sure but the important part here is it’s also easy to deploy to the internet for free or a few bucks.
All you said is right but still, I don’t think there is anything comparable nowadays and if I had to explain to my son how to create a basic non static website without using proprietary PaaS, I’d have a hard time (without PHP).
Was it the best option when you were a company building a product ? idk but at least it was a viable option. Was it the best option for any kiddo wanting to learn creating things on the internet ? Totally and we never got anything better for this since PHP.