David Cournapeau wrote:
On Sun, Jan 17, 2010 at 4:17 AM, John Nagle <na...@animats.com> wrote:
Nobody wrote:
On Fri, 15 Jan 2010 12:34:17 -0800, John Nagle wrote:
Actually, no. It's quite possible to make a Python implementation
that
runs fast. It's just that CPython, a naive interpreter, is too primitive
to do it. I was really hoping that Google would put somebody good at
compilers in charge of Python and bring it up to production speed.
Look at Shed Skin, a hard-code compiler for Python
A hard-code compiler for the subset of Python which can easily be
compiled.
Shed Skin has so many restrictions that it isn't really accurate to
describe the language which it supports as "Python".
Hardly any real-world Python code can be compiled with Shed Skin. Some of
it could be changed without too much effort, although most of that is the
kind of code which wouldn't look any different if it was implemented in
C++ or Java.
The monomorphism restriction is likely to be particularly onerous: the
type of a variable must be known at compile time; instances of subclasses
are allowed, but you can only call methods which are defined in the
compile-time class.
If you're writing code which makes extensive use of Python's dynamicity,
making it work with Shed Skin would require as much effort as re-writing
it in e.g. Java, and would largely defeat the point of using Python in the
first place.
http://shedskin.googlecode.com/files/shedskin-tutorial-0.3.html
If you want a language to have comparable performance to C++ or Java, you
have to allow some things to be fixed at compile-time. There's a reason
why C++ and Java support both virtual and non-virtual ("final") methods.
My point is that Python is a good language held back by a bad
implementation. Python has gotten further with a declaration-free syntax
than any other language. BASIC and JavaScript started out declaration-free,
and declarations had to be retrofitted. Python has survived without them.
(Yes, there are hokey extensions like Psyco declarations and "decorators",
but both are marginal concepts.)
There are efficient implementations of dynamic programming languages
which do not rely on declaration (if by declaration you mean typing
declaration), even when available:
http://strongtalk.googlecode.com/svn/web%20site/history.html
See also:
http://www.avibryant.com/2008/05/those-who-misre.html
Yes, that's my point.
Psyco was a good first step. The big win with Psyco is that it
generally can recognize when a variable is an integer or floating
point number, and generate hard code for that. It doesn't do much
for the rest of the language. Psyco is really a kind of JIT compiler.
Those are useful, but in some ways limited.
To go beyond that, global analysis is needed. A big bottleneck
in Python is that too much time is spent doing dictionary lookups
for things that could be bound at compile time. So the next big
win is figuring out which classes definitely don't have any hidden
dynamism. A global check is needed to see if any external code
messes with the attributes of a class or its functions from outside
the function. Most of the time, this is the case. Once that's
been done, the class's module can be analyzed for optimization.
If the class doesn't use "setattr", etc. to add attributes to
itself, then the class can be "slotted", with a C++ like structure for
the class members and functions.
Global analysis also has to determine the class hierarchy; what inherits
from what. It may be necessary to implement "object" as an abstract class
with a huge number of virtual functions, so that "duck typing" will work.
That's a space cost, but not a time cost.
Caller/callee type inference is useful to determine the potential types
of parameters. Often, analysis of all the calls to a function will determine
the types of many of the paraeters. Then, those parameters can be hard-typed
at compile time.
You can go this far without the restrictions Shed Skin imposes, such as
the restriction that lists must be homogeneous. If you do impose that
restriction, array processing becomes much faster. Type inference for
array elements is hard when arrays are computed from other arrays, so
that's a huge simplification.
Yes, you can't use "eval" to get at existing variables. But in Python,
you don't really need to.
John Nagle
--
http://mail.python.org/mailman/listinfo/python-list
