PHP 8.0 Ready To Ship With Many New Features, Even Better Performance

[phoronix]

Phoronix: PHP 8.0 Ready To Ship With Many New Features, Even Better Performance

PHP 8.0 is scheduled for release tomorrow on the US Thanksgiving day. PHP 8.0 brings with it many new language features on top of the opt-in JIT compiler support. Here is a look at some of the PHP 8.0 changes along with a quick look at the near final performance of PHP 8.0 on an AMD EPYC Linux server...

http://www.phoronix.com/scan.php?page=news_item&px=PHP-8.0-Features-Benchmark

[baryluk]

Any important removals? I know there were many deprecated methods and functions.

[schmidtbag]

To my understanding, the JIT isn't architecture-specific, which is great news to me. Nextcloud can be pretty CPU intensive so I'm excited when it gains 8.0 support.

[ssokolow]

That list of features leaves me feeling, more than ever, that PHP is the C++ of scripting languages. Forever shackled to a flawed start, but piling on every feature it can to try to make up for it, turning into a giant grab bag.

[Redfoxmoon]

Still waiting for cross-compilation support in PHP :^)

[ed31337]

That list of features leaves me feeling, more than ever, that PHP is the C++ of scripting languages.

If it's got a JIT now, isn't PHP beyond a "scripting language" and more akin to VM languages like Java or .NET?

[caligula]

If it's got a JIT now, isn't PHP beyond a "scripting language" and more akin to VM languages like Java or .NET?

Still refuses to work with the established VM ecosystems, this is a totally new NIH platform for PHP only. Anyway this is pretty awesome from the end user's point of view - you'll get more speed by doing almost nothing. It's like the Herb Sutter's Free Lunch speech.

[ssokolow]

If it's got a JIT now, isn't PHP beyond a "scripting language" and more akin to VM languages like Java or .NET?

I tend to draw that particular line based on PHP's lack of a mandatory separate compilation step and focus on dynamic typing as default, same as with PyPy and LuaJIT.

(Also, JavaScript has JITs, but I'd never lump it in with Java and C#.)

[szymon_g]

Why bother with PHP when we have the python?

[gggeek]

Any important removals?

Not that many that I can think of, but a _lot_ of changes that will need adjustments in existing codebases and extensive app revalidation, such as f.e. changes in error reporting (many functions are much stricter now), subtle changes in string comparison semantics, etc...

Forever shackled to a flawed start, but piling on every feature it can to try to make up for it, turning into a giant grab bag.

Have to agree with this one :-D

Still refuses to work with the established VM ecosystems, this is a totally new NIH platform for PHP only

The Parrot project tried to do a multi-language VM (interpreter?) for scripting languages, and it got nowhere.
Graalvm, while being touted as giving revolutionary perfs, seems to be mostly a big jump in perfs for Scala, which was penalized by existing Java VMs being optimized for... Java! So a generic VM/JIT framework is not such a common thing? Last but not least, the PHP JIT is based on LUAJIT iirc. More details on the choices made can be found at https://wiki.php.net/rfc/jit - including "we tried LLVM, but its code generation speed was almost 100 times slower, making it prohibitively expensive to use"

Why bother with PHP when we have the python?

Fun fact: I never really learned Python because I have PHP