> ...their interest in implementing PHP using invoke-dynamic on the JVM

This is very interesting.

> The main seismic event will be nothing less than the complete removal of interpretors from main stream general purpose programming.

This is hyperbolic drivel.

Makes sense - the JVM is fast and stable. It comes with a huge array of tooling and its own built in debugging system. It is well supported and cross platform. Why would you not want to take advantage of all that good stuff?

Doesn't make that much sense to me - the JVM is complex and has disadvantages (slow startup times etc.), so if you can compile PHP to native code, why compile to interpreted/JIT-compiled code instead? You're just wasting cycles again.

The advantages of the Java ecosystem can be accessed through Thrift if necessary, but the C/C++ world is far from being dead.

It might be interesting for the security and debugging aspects of the JVM, but that's a bit meagre for such an effort.

And some people might not like to hear it but Oracle has really provided a new injection of excitement in the JVM with their G1 GC and JRockit additions.

And with clarity around the future roadmap and the ever growing list of non-Java languages the platform has never looked better IMHO.

Oracle moving on from the 1.6 doldrums was a huge plus for Java and the JVM. Now the buzz is around 1.8 with lambdas and enhancements to performance of invoke dynamic.

And what's great is that since Apple has handed over responsibility of OSX Java over to Oracle it means everyone can start to use the 1.8 features.

Feels like everyone has been stuck on 1.5/1.6 for too long.

It is a static language after all. </joke>

Source? Or anything to show that this should be trusted?

And even if the post is correct and they actually are testing out an run-php-on-the-jvm solution, it's not the same as moving all their stuff onto it.

There is already a PHP for JVM:

http://www.caucho.com/resin-3.1/doc/quercus.xtp

Wouldn't facebook benefit from just starting to rewrite parts of their infrastructure in a language more suited to their scale, like twitter did with scala? I can't help but feel all this effort into bending php into corners it was never meant for is not helping them in the long run.

Sure they could but where's the geeky fun in that?

As a developer, I'm just happy that the Facebook devs can do projects like this. They have a real shot at making life better for the rest of the devs on the planet.

When you are facebook's scale, there is no off-the-shelf solution. Whatever they choose to do, it's most likely going to be their own effort. There is nothing that bad about the PHP language and the fact that all their developers know it, it makes sense that they keep the language, just change what runs it.

PHP is fine and dandy for Wordpress sites but to think that Facebook got this far on it is nothing short of amazing.

If this article proves true it'll be a good move for them.

Facebook's PHP is not normal PHP: https://github.com/facebook/hiphop-php

Will Facebook contribute to Quercus, or doing their own thing? http://quercus.caucho.com/

Based on what we've seen from Facebook, Twitter, Google etc they will likely do it on their own.

As we've seen with the various Memcache and MySQL implementations often tradeoffs are made in favour of performance and scalability.

As I said only yesterday[1] there is no solid evidence that this is happening.

None.

This is, at best, a tertiary source.

[1]: http://news.ycombinator.com/item?id=4374216

first twitter, now facebook.