Python IS slow ! [was] Re: Python too slow for real world
Markus Kohler
markus_kohler at hp.com
Thu Apr 29 04:17:41 EDT 1999
>>>>> "William" == William Tanksley <wtanksle at dolphin.openprojects.net> writes:
[deletia]
William> Your data is correct (Python is slow for many things,
William> slower than it needs to be), but your conclusion is
William> wrong. Python isn't slow because its bytecode engine is
William> slow; actually, although there's a lot of room for
William> improvement, its bytecode engine is faster than many of
William> the others out there.
It seems to me that *most* of the byte code engine is ok.
William> The reason it's slow is its runtime model. _Every_
William> function call requires a lookup in a hash table, just on
William> the off-chance that the programmer changed the meaning of
William> the function.
That is the same in Smalltalk (squeak). You can build new classes and new
methods at runtime and everything still works as expected.
>> It seems to me that without a redesign of at least the bytecode
>> for function calls python's speed will not take off.
William> Bytecode won't help enough -- the whole calling model
William> needs to be examined. Fortunately, that's one of the
William> things the 2.0 design process will be looking at. Like
William> you, I hope that they consider Smalltalk as an example.
I hope so too. My point was exactly that the bytecode for doing the
function call seems to be the problem.
William> And Oberon (SlimBinaries), and Eiffel (typing and general
William> compile-time error catching), and ...
OPTIONAL types would be cool.
Markus
--
Markus Kohler mailto:markus_kohler at hp.com
More information about the Python-list
mailing list