[Stackless] The Future of Stackless

Bob Ippolito bob at redivi.com
Mon Dec 24 03:42:16 CET 2007


On 12/23/07, Travis Jensen <travis.jensen at gmail.com> wrote:
> Let me say, Stackless is awesome.  It takes a great language and moves into
> the world of concurrency very nicely.
>
> Unfortunately, Python doesn't play in the concurrent, multi-processor world
> very well.  I'm in the architecture process for a massively multiplayer
> game, and I love what Stackless Python offers for both the client and the
> server.  I'm concerned, though, as I watch the number of cores increase
> around me and I don't see anything happening in the Python world (including
> Stackless) to address this.
>
> So, this leads to my question:  Is there a vision for where Stackless is
> going, or has it gotten there and is being maintained?  Is there any thought
> of Stackless addressing the GIL-related issues of CPython?
>
>  I look at Stackless and see the possibility for a Python-based "Erlang",
> but I also recognize that I'm not the one donating the time to make it
> happen.  I also understand that there are huge potential problems around the
> GIL that have to be addressed as well.
>
> This community seems to have the most invested in concurrent programming.
> There are other projects, such as MP Python, but the Stackless model is SO
> much cleaner for small tasks that fit so nicely in the game world.
>
> Thoughts?  I'm really just trying to understand where Stackless is going, if
> anywhere, but other interesting thoughts are always appreciated, including
> pointing me in directions I might not have thought of.

Why not just learn Erlang, at least for the server? It's good at doing
all of these things *today* (actually, years ago) and it also scales
to multiple machines. It's also much better at doing TCP stuff than
Python is, in my real-world experience, which is pretty important.

-bob




More information about the Stackless mailing list