Re: [PHP4BETA] Trouble with Oprimizer

Bytecode is a nice buzzword, but writing a byte code generator, which obviously also means you have to write a complmenting virtual machine, is several man-years of work. And, since it's not suitable for realtime execution, it would have to be in addition to the existing facilities inside PHP/Zend. In short, that's unlikely to ever happen. However, generating bytecode has little to do with the problem Chris raised. If you could save the compiled representation (which is not bytecode) of a PHP script in memory, that would solve your problem. Zend Tech has already announced the Zend Cache, you can read about it on the web site. Zeev On Sat, 1 Apr 2000, Andreas Pour wrote: > Chris Chabot wrote: > > > > My main question would be, how large is this script? ive found that in many > > cases, loading & preparsing a large code base is the main bottleneck, and > > not the script execution time. > > As im sure uve read in the description of the optimizer, its a _multi > > parse_ optimizer, so the script load time only increases. Ive got the same > > problem here with a 140Kb script :) > > What would be awesome is a facility to convert php scripts to bytecode. > It's an avoidable waste of resources to be parsing the same large > scripts every time someone hits a page. Anyone have an idea know how > hard it would be to add a bytecode generator/parser into the php > frontend (it could even automatically check timestamps to make sure the > bytecode is newer than the script)? Probably the better place to put > the bytecode generator is in Zend engine (post-optimization), but the > license seems to be unfavorable. > > Ciao, > > Andreas > > -- Zeev Suraski <zeev@zend.com> http://www.zend.com/