PEP 266: Optimizing Global Variable/Attribute Access

[Skip Montanaro]

    gzeljko> It's not only about changes in bindings.  Given x.y.z, if any
    gzeljko> of objects (x, y or z) is obtained by function with side
    gzeljko> effects, using a(=x.y.z) simple is not the same semantic.

The PEP addresses this.  I updated it yesterday.  Read the response in the
Questions section.  It goes into some detail about how this will work.  In
essence, if you track x.y.z, the system will notice changes to the bindings
of x, x.y and x.y.z and respond accordingly.  In fact, the goal is that it
will work no matter how many dots your expression has, though the most
common case will be x.y.

Repeat the mantra:

    Python's dynamic semantics will not change
    Python's dynamic semantics will not change
    Python's dynamic semantics will not change

;-)

-- 
Skip Montanaro (skip at pobox.com)
http://www.mojam.com/
http://www.musi-cal.com/