Re: 4.0.2 RC1 php.exe performance

At 01:50 PM 8/26/00 -0400, Brad Atkins wrote:

Hi Andi, Are you using require() a lot? I think in big scripts that did lots of require()'s the old "fast" method ended up being substantially slower. This could explain the performance difference. Yes, we are using require()'s instead of includes, for reasons you know of. Was there a change with the require in 4.0.2?

Yes. require() is just like an include() now but is not allowed to fail. There are three main reasons for this: a) The less important one is that include_once() and require_once() now work together. b) Most importantly, require() contrary to what we thought ended up being slower than include(). c) The semantics of returning from a require() were different than from an include() which would have confused ppl a lot. We thought it would be best to make them consistent. d) The change allowed us to save quite a bit of memory. Well never mind the blah blah :) I think this is probably the main reason for the performance increase you are seeing now.

Also in case you were doing lots and lots of eval()'s there was also a related memory leak patch which could have also effected your servers performance. Yes, we use eval() a lot also, not lots and lots, but enough :)

Again, probably require() had more of an impact but there was a memory leak with eval() which could have caused PHP to take up more memory and thus slow down your scripts. Andi --- Andi Gutmans <andi@zend.com> http://www.zend.com/