[plt-scheme] How best to create a "sandbox" environment?

I see, interesting!

Would another approach here be to define your own module "language"
(not sure if this is the right terminology) where you extend mzscheme
and only export functions you deem to be safe? Would that require a
painful amount of work to do so?

On 7/9/07, Erich Rast <erich at snafu.de> wrote:
> >
>
> Hi,
>
> > What would be an approach to preparing a sandbox that disallowed IO?
> >
>
> Check out this entry in the Scheme cookbook:
>
> http://schemecookbook.org/Cookbook/DynamicUntrustedEval
>
> AFAIK nobody has used this in a public service and the security
> features of MzScheme haven't been under intensive attack so far, so I
> wouldn't depend my life on the security of it.
>
> Greetings,
>
> Erich
>