Re: Why change require_once? A brief explanation of motives

Matthew Weier O'Phinney wrote:

On 7/18/07, Pádraic Brady <padraic.brady@yahoo.com> wrote:

It's the RFC that needs editing - and external FAQ isn't as powerful as inline explanations backing up every RFC proposal with both pros and cons. We also need benchmarks to prove any assertions the RFC editors intend making - the whole APC issue looks very dodgy to me until someone comes up with numbers to prove one way or the other. It would be preferable to see benchmarks based on a real PEAR package. Same for using autoload in various circumstances/configurations - userland __autoload() only, multiple autoload SPL stack, etc. I think the discussion isn't really achieving anything until these are produced. My 2c ;).

I'm absolutely in agreement on this, and have been asking for these for a week now, with no response from any of the RFC authors. Greg, Arnaud, and whoever else helped create the draft: could you please either create such benchmarks and/or delegate this task to others? The benchmark code should be packaged so others may grab it and verify the results. Additionally, they should be done using the most recent versions of PHP and APC available; testing against old versions makes no sense if the performance issues have been fixed in recent versions.

Hi Matthew, the RFC will be updated, please allow some time. I did not answer last week as I was on vacation. Arnaud.