The way to a faster python [was Python IS slow !]

[Christian Tismer]

Martijn Faassen wrote:
> 
> Markus Kohler wrote:
> [lots of my Swallow ideas]
> >
> > IMHO it's a great idea.
> 
> Thanks! Though I think the only new contribution (which doesn't seem to
> be new outside Python) is the idea of subsetting Python (a strict
> subset) to get a manageable variety for translation/compilation.
> 
> Now we should work out that subset. :)

How about starting small, very small.
Just let's try to write S-Functions.
The problem is how to set up the filters
and rules what might go how into an S-function
without damaging anything. Simple immutables need just
a typecheck. Collections need more work.
If something can exist in both worlds at the same time,
this may not give conflicts.
If an S-function receives a list which containes an
assured type of elements, they must be no way to break this.
The element must be passed completely,or in other words, the 
refcounts of all mutable parts which the S-function uses
must be one. After working with that, it gives it back.

This means there is no direct access to globals possible,
only by a local, probably deep, copy. Do you see it?

ciao - chris

-- 
Christian Tismer             :^)   <mailto:tismer at appliedbiometrics.com>
Applied Biometrics GmbH      :     Have a break! Take a ride on Python's
Kaiserin-Augusta-Allee 101   :    *Starship* http://starship.python.net
10553 Berlin                 :     PGP key -> http://wwwkeys.pgp.net
PGP Fingerprint       E182 71C7 1A9D 66E9 9D15  D3CC D4D7 93E2 1FAE F6DF
     we're tired of banana software - shipped green, ripens at home