Speed: bytecode vz C API calls

[Jacek Generowicz]

Peter Otten <__peter__ at web.de> writes:

> See below for my ideas towards a generalized memoizing framework.

[I haven't the time to have a look at it in any detail right now, so
forgive me for not responding/commenting in full, just yet.]

> >   def foo(some_type):
> >       ...
> >       return something
> > 
> > 
> >   foo = memoize(foo)
> > 
> >   data = [int, float, my_class, his_class, str, other_class] * 10**6
> > 
> >   map(foo, data)  [+]
> 
> Do you discard the result for real?

No, sorry, of course not.

> > [+] Can you see why I don't want to mess with the call site ?
> 
> No.

It would mean taking it out of the map and that would mean losing the
C-coded loop that map provides.

> Now to my code. The idea is to make memoized functions available to
> different modules from a central registry, and to expose the cached results
> for hand-optimized hotspots. Otherwise it was all mentioned in this thread,
> if not the original post.

I'll make sure to find the time to study this.