Bill Dortch wrote:
On 10/20/07, *Bill Dortch* <[EMAIL PROTECTED]
<mailto:[EMAIL PROTECTED]>> wrote:
Might it have something to do with the reference to self in the call?
...
This appears to be at least part of the story. If I change:
...
to:
def benchmark_attr_get
...
self.a; self.a; self.a; ...
...
end
...
100k * 100 attr gets (a;), 1 ivar 2.703000 0.000000
2.703000 ( 2.703000)
100k * 100 attr gets (self.a;), 1 ivar 5.234000 0.000000 5.234000
( 5.234000)
Yes, this seems in line; the reference to self is necessary in the attr
set test because "a = " is *always* a local variable set. attr sets
*must* be fully qualified.
Even stranger, the self. reference is slower than accessing an attr in a
separate object (x):
100k * 100 attr gets (x.a;), 1 ivar 3.922000 0.000000 3.922000
( 3.922000)
Curious, no?
Quite. I'm looking at it now.
- Charlie
---------------------------------------------------------------------
To unsubscribe from this list please visit:
http://xircles.codehaus.org/manage_email
