Prototype PHP interpreter using the PyPy toolchain - Hippy VM View as Markdown

On Mon, Jul 16, 2012 at 12:04 PM, marius adrian popa mapopa@gmail.comwrote:

http://morepypy.blogspot.com/2012/07/hello-everyone.html

and the comments
http://news.ycombinator.com/item?id=4241921

http://en.reddit.com/r/programming/comments/wipf2/prototype_php_interpreter_using_the_pypy/

Hi,

interesting, but am I the only one that we could also boost the performance
of the Zend Engine, if we would throw out everything, except the "PHP 1.0"
features: functions, arrays, ints, floats and strings
not that we should do that, I just think that the numbers are somehow
misleading.
+1
thanks

--
Ferenc Kovács
@Tyr43l - http://tyrael.hu

--
Laruence Xinchen Hui
http://www.laruence.com/

To help us in this, there's the PHP Native Interface RFC. It might be
difficult to improve the Zend engine until extensions aren't so
tightly coupled with the Zend API.

On Mon, Jul 16, 2012 at 12:04 PM, marius adrian popa mapopa@gmail.comwrote:

http://morepypy.blogspot.com/2012/07/hello-everyone.html

and the comments
http://news.ycombinator.com/item?id=4241921

http://en.reddit.com/r/programming/comments/wipf2/prototype_php_interpreter_using_the_pypy/

Hi,

interesting, but am I the only one that we could also boost the performance
of the Zend Engine, if we would throw out everything, except the "PHP 1.0"
features: functions, arrays, ints, floats and strings
not that we should do that, I just think that the numbers are somehow
misleading.

--
Ferenc Kovács
@Tyr43l - http://tyrael.hu

--
Andrew Faulds (AJF)
http://ajf.me/

Hi!

interesting, but am I the only one that we could also boost the performance
of the Zend Engine, if we would throw out everything, except the "PHP 1.0"
features: functions, arrays, ints, floats and strings
not that we should do that, I just think that the numbers are somehow
misleading.

They also removed refcounting from the language - meaning, as I
understand from this:

The project's biggest deviation from the PHP specification is probably
that GC is no longer reference counting. That means that the object
finalizer, when implemented, will not be called directly at the moment
of object death, but at some later point.

that they do not destroy objects at all currently. So these benchmarks
don't mean much.

But the idea is interesting, if they ever take it to actually working
implementation. It's the kind of things that first 90% take 90% effort,
and last 10% take another 90% effort.

Stanislav Malyshev, Software Architect
SugarCRM: http://www.sugarcrm.com/
(408)454-6900 ext. 227