• [^] # Re: Peut-être aurait-il été utile de préciser que...

    Posté par (site web personnel) . En réponse à la dépêche IronPython : implémentation pour Mono/.NET. Évalué à 4.

    > ce que de petits malin ont affirmé en faisant une implentation bidon de Python

    Tu parles peut-etre de l'implementation realisee par Mark Hammond un des gourou de la communaute python, implementation financee par Microsoft ?

    Dans ce cas, je remballerai l'experssion "petits malins". Mark Hammond est extremement respecte dans la communaute python et il y a des raisons pour lesquelles son implementaiton n 'etait pas optimale, que Guido van Rossum detaille dans mon interview (http://www.freehackers.org/fosdem2002/guido.html(...)):

    <<
    Anyway, they did a Python port to .NET very early on, with some funding from Microsoft. I think the issue was mostly for Microsoft to show that Common Language Runtime really was a suitable target for multiple languages because they did the same thing with lot of language groups. [...] It's true that the Python .NET interpreter is very slow. I think part of that is just that they didn't have enough time to make it faster, there is always optimisations. The other thing was beacuase they did it so early on that they had to sign very strict Non Disclosure Agreements and Mark Hammond couldn't ask anyone for help because he was on NDA. He couldn't tell anybody even that he was working on this project until it was finished. So if he didn't know the compiler techniques, he was basically on his own. And there was also not much documentation about .NET available at that point, because .NET was basically still a Microsoft internal project. So I think now, this is based on a few things that Miguel told me actually. There are number of approaches to make it much faster.
    >>