[Twisted-Python] does twisted.* support mobile code?
Andrew Straw
strawman at astraw.com
Sat Nov 5 20:34:06 MST 2005
Jean-Paul Calderone wrote:
> On Sat, 05 Nov 2005 14:21:25 -0800, Andrew Straw <strawman at astraw.com>
> wrote:
>
>> Hi,
>>
>> Does any part of twisted support mobile Python code? Something like
>> Pyro's Mobile Code objects?
>> http://pyro.sourceforge.net/manual/7-features.html#mobile
>
>
> I wrote out a big long screed in response to this, but then decided
> it was probably overly harsh.
Well, we're all grown-ups here and I think I can take it, but it's still
interesting enough to know that you think it's the wrong thing.
Translation -- I'll try not to be offended! :)
The bottom line for me is that I noticed an easy way to do some
parallelization of "embarrassingly parallel" problems which would have a
nice match with the Pythonista mindset: instead of map( func, vars ), we
could have a map_distributed( func, vars ), which gets mapped over many
processes, potentially over many CPUs. I'm thinking now solely of a
single multicore box or clusters behind firewalls, so security issues
didn't seem relevant. I've got some rudimentary Pyro code working, but I
was wondering what was available with Twisted.
> No, Twisted does not support this. The lack of support is intentional.
> There are a huge number of security-related issued with doing what Pyro
> does, so Twisted doesn't do it.
OK.
> In fact, PB goes in the other direction: even if the code defining the
> class of an instance passed over PB is available on both sides, it _still_
> might not be accepted by the recipient, if it has not been explicitly
> marked as safe by the recipient.
OK, thanks. I just wanted to make sure I wasn't missing something in the
spread docs...
More information about the Twisted-Python
mailing list