4 ms·

This is not the first attempt. This seems to be focused on extending PHP's performance and capabilities. Personally, I'm more interested in porting PHP applic

by hapless 13y ago

This is not the first attempt. This seems to be focused on extending PHP's performance and capabilities. Personally, I'm more interested in porting PHP applications away from PHP. Caucho Resin offers 100% php compatibility on the JVM, so that you can host your entire app in a servlet container, and port features one by one. http://www.caucho.com/resin-3.1/doc/quercus.xtp http://www.caucho.com/resin-3.1/doc/quercus.xtp The PHP-to-Scala migration helper uses Resin/Quercus to provide the standard library to translate PHP to Scala. The output is a Scala codebase, so that you never have to touch the legacy PHP again. https://code.google.com/p/php-to-scala-migration-helper/ https://code.google.com/p/php-to-scala-migration-helper/

camus2 13y ago
looks great!Any php "optout" solution is valuable.Or sometimes you just want to run a quick CMS for a client;) It's not clear though which PHP version it does support.
- pritambaral 13y ago
  They don't support any specific PHP version outright, rather individual features (appropriately tagged with PHP version).
aeontech 13y ago
Caucho deserves to be its own post. Looks very interesting, I had no idea something like that was available.
ampersandy 13y ago
Are there any unbiased comparisons of Resin and other app servers? The only source for their claims of better performance is their own benchmark of serving what seem to be static files. [0] A discussion of it pointed out that the comparison is totally moot, since Resin caches 1k files (and thus outperforms Nginx for that metric), but otherwise is pretty similar for larger files. [1] [0] http://www.slideshare.net/billdigman/resin-outperforms-nginx http://www.slideshare.net/billdigman/resin-outperforms-nginx [1] https://www.ruby-forum.com/topic/4404987 https://www.ruby-forum.com/topic/4404987
- bhauer 13y ago
  We use Resin as the Java Servlet container for plain Servlet and most of the Servlet-derived frameworks in our benchmarks [1]. Incidentally, benchmarking static delivery of larger files on modern hardware is a bit uninteresting since it's relatively trivial to saturate gigabit Ethernet. [1] http://www.techempower.com/benchmarks/ http://www.techempower.com/benchmarks/
- jmspring 13y ago
  Used Quercus (the actual php -> java bridge) years ago. I can't recall exact numbers, but generally was about 50-80% faster than the same php based site using a well tuned xcache install (ie, both avoiding the file load/compile). Ran it using resin, tomcat, glass fish, and jboss. No real difference by app server. That said, the biggest issue is always site design. Page too big, too many queries, too many assets loaded, etc. will not benefit much from Quercus or any similar technology.
EGreg 13y ago
How does it compare to HipHop or fb's newest offering?
- Terr_ 13y ago
  IIRC: 1. Normally, PHP becomes opcodes for the PHP interpreter to run. 2. Quercus takes PHP code and generates .java source files from it, which then run in the JVM like other Java code. 3. The earlier HipHop (HPHPc) compiles PHP to C++ source 4. The more-recently-released HipHop Virtual Machine (HHVM) is an alternative to PHP's typical interpreter.

by hapless 13y ago

camus2 13y ago

looks great!Any php "optout" solution is valuable.Or sometimes you just want to run a quick CMS for a client;) It's not clear though which PHP version it does support.

pritambaral 13y ago
They don't support any specific PHP version outright, rather individual features (appropriately tagged with PHP version).

aeontech 13y ago

Caucho deserves to be its own post. Looks very interesting, I had no idea something like that was available.

ampersandy 13y ago

Are there any unbiased comparisons of Resin and other app servers? The only source for their claims of better performance is their own benchmark of serving what seem to be static files. [0] A discussion of it pointed out that the comparison is totally moot, since Resin caches 1k files (and thus outperforms Nginx for that metric), but otherwise is pretty similar for larger files. [1] [0] http://www.slideshare.net/billdigman/resin-outperforms-nginx http://www.slideshare.net/billdigman/resin-outperforms-nginx [1] https://www.ruby-forum.com/topic/4404987 https://www.ruby-forum.com/topic/4404987

bhauer 13y ago
We use Resin as the Java Servlet container for plain Servlet and most of the Servlet-derived frameworks in our benchmarks [1]. Incidentally, benchmarking static delivery of larger files on modern hardware is a bit uninteresting since it's relatively trivial to saturate gigabit Ethernet. [1] http://www.techempower.com/benchmarks/ http://www.techempower.com/benchmarks/
jmspring 13y ago
Used Quercus (the actual php -> java bridge) years ago. I can't recall exact numbers, but generally was about 50-80% faster than the same php based site using a well tuned xcache install (ie, both avoiding the file load/compile). Ran it using resin, tomcat, glass fish, and jboss. No real difference by app server. That said, the biggest issue is always site design. Page too big, too many queries, too many assets loaded, etc. will not benefit much from Quercus or any similar technology.

EGreg 13y ago

How does it compare to HipHop or fb's newest offering?

Terr_ 13y ago
IIRC: 1. Normally, PHP becomes opcodes for the PHP interpreter to run. 2. Quercus takes PHP code and generates .java source files from it, which then run in the JVM like other Java code. 3. The earlier HipHop (HPHPc) compiles PHP to C++ source 4. The more-recently-released HipHop Virtual Machine (HHVM) is an alternative to PHP's typical interpreter.