Re: Fwd: What's up with Quercus?
| From: | Tom Samplonius | Date: | Tue, 31 May 2011 07:06:23 +0000 |
| Subject: | Re: Fwd: What's up with Quercus? | ||
| References: | 1 | Groups: | php.internals |
| Request: | Send a blank email to internals+get-52551@lists.php.net to get a copy of this message | ||
...
> Running PHP on the JVM doesn't mean making PHP more like Java. It
> just means running the PHP language on a platform with a lot of
> benefits and advantages, and given the differences in engineering
> resources dedicated to each, one that's likely to continue to
> improve a lot faster than the native PHP runtime.
I think you are assuming that the bit of code that converts PHP to JVM bytecode (and/or a Java PHP
interpreter) is an insignificant chunk of engineering. Quercus has lagged PHP5 development, and it
needs work in order to run common apps. Plus, an analog of the Ruby RSpec project will probably
have to be built for PHP (PHPSpec?). The PHP5 implementation is the spec right now, just like Ruby
(used to be).
Quercus is currently a component of a commercial Java application server. Can Quercus can be run
in another app server, like Jetty, or does it only work in Resin? I doubt that Caucho would be
interesting in making Quercus work stand alone. Quercus will probably need to be forked.
That is quite a lot of work. How many people did it take to create JRuby (Ruby on the JVM) and
the RSpec project? From the mailing list, it looks like it was about 5 to 8 core developers over
about 2 years. And then after all that, JRuby is still not the preferred deployment environment for
Ruby (it's mod_rails aka Passenger). Though, RSpec did end up making the original Ruby
implementation better.
I'm not opposed to PHP on the JVM, but it is not a small undertaking. You should probably
begin by extracting the Quercus code from from the Caucho subversion repository, and import it into
a new github repository, and then see if you can get Quercus to run without Resin.
> As far as HipHop, from what I've seen it holds zero interest for me
> right now. If you're FaceBook, and you're spending hundreds of
> millions of dollars on servers every year then squeezing 5% or 10%
More like 50% to 100%.
> more performance out of you code is probably worthwhile. But with
> only a few hundred servers running the PHP applications I'm
> responsible for, I'm much more interested in developing less-buggy
> applications faster, leveraging other people's frameworks instead of
PHP has shed loads of web frameworks too.
> reinventing the wheel (hence my interest in Java), improving
> monitoring and deployment, etc.
>
> -- Arnold
Tom