Does Python (or psyco) already do this optimisation?

[Mike C. Fletcher]

Thanks for the pointers Skip (and Michael and Terry), PEP 280 also 
appears to target the same problem.  The approaches are all fairly 
different, with the DLict approach being closest to what I was thinking 
of.  I'd not considered the locality-of-reference problem with tracing 
the pointers, though I would have thought that would be at worst a wash 
compared with hash-table lookups.

All-in-all, looks like wiser (and more C-savy) minds than mine are 
already on the task, so I guess I'll leave it to them.

Much appreciated all, can cross it off the list of ideas that might go 
somewhere now,
Mike

Skip Montanaro wrote:

>    Mike> Any idea (anyone) whether the outlined approach has been attempted
>    Mike> for global-variable access optimisation?  i.e. is this an explored
>    Mike> and deemed-dead path, or something new?
>
>Check PEPs 266 and 267.  I don't think it's necessarily dead-end.  It's
>proven to be a damned hard problem so far though.  Jeremy and I led a
>developer's day session on optimization at the February 2002 Python
>conference.  Jeremy's approach has come much farther than mine since then
>(as in I didn't do anything).  Check the python-dev archives for the
>February-April 2002 timeframe for more discussion on the subject.
>  
>
_______________________________________
  Mike C. Fletcher
  Designer, VR Plumber, Coder
  http://members.rogers.com/mcfletch/