[campnic]

This is completely loaded. Your implication is that the only viable test is a test which exercises all of the functionality of the most feature rich framework. How would that be a)viable and b) meaningful?

We know that there is a set of common features and the benchmarks goal is to test least common denominator stuff on the networks. Authentication and portability are not LCD. The argument that they are is capricious. What if we made the requirement be that the framework is a lisp? Now we've completely changed the intent.

[sequoia]

I meant to suggest that comparing php-raw to Rails is apples & oranges- not "you must benchmark in a way that benefits larger frameworks", just "please acknowledge that LCD tests like this inherently cast Railsy frameworks in a bad light."

It's like condemning a swiss army knife because it's not as efficient as a fixed blade at cutting apples. Well yeah that's true, but what about when you need to screw a screw or pull a cork? One is a multitool, it doesn't make sense to compare it to a specialized tool unless all you plan to do is cut apples.