Hi, how do I create a partial trusted sandbox of an embedded scripting engine (IronPython for exampl

microsoft.public.dotnet.framework

6 Jun 2006 3:01 PM

Li-fan Chen

Hi,

I'd love to enable the writing of custom scripts by users. The main
purpose is to do smart merge tags for a mail merge. Each such smart
merge subroutine will add an additional dimension in transforming the
looked-up substitution value.

For example, one such sub routine would derive a suitable default
naturalized salutation in Spanish when the looked-up value is a DBNull.

It would be nice to cache and host the scripting engine and serve the
execution via web services.

The web service would take script snippets matching a certain call
signature. Verify it
    * doesn't try to get out of the sandbox;
    * makes use of a small selection of safe libraries;
    * enjoys completion within a certain amount of time, free of infinite
loops

The web service would then pre-compile the snippet and cache it for
calling on the fly. Script updates would prompt recompiles.

The web service will then field directives to call these subroutines.
These web service calls will provide the parameter values. The web
service is then responsible for loading the embedded scripting engine
environment, calling the script, and then returning the value.

The web service may cache the scripting environment to eliminate
unnecessary initialization.

I have been made aware that we should investigate CRL concepts such as
AppDomains with partial trust domains. And to study recent articles
embedding IronPython into MDbg. But has anyone had to do something like
this? With IronPython or another scripting language engine?

Any help or assistance would be greatly appreciated, thank you ahead of
time!

Best regards,
-- Li-fan