Hello MW,
It's certainly possible to execute multiple scripts sequentially within a given script engine instance, and no unloading is necessary in between. However, a malicious or buggy script could easily corrupt the environment for subsequent scripts – by deleting or replacing built-in objects, leaving unexpected objects in the root namespace, etc.
Therefore a careful evaluation of script trust level is always a good idea, combined with an assessment of the impact of a corrupted script environment. Some applications such as web browsers rely on the ability to execute multiple untrusted scripts within the same context, and have robust mechanisms for detecting and discarding contexts that are stuck or otherwise problematic.
The bottom line is that if it's important that all scripts start out in a pristine environment, and you cannot assume that each script will clean up after itself, then you should create a new context (script engine instance) for each one.
On a separate note, if you need to expose a host object in a way that allows it to be replaced later, consider using dynamic property assignment instead of
Good luck!
It's certainly possible to execute multiple scripts sequentially within a given script engine instance, and no unloading is necessary in between. However, a malicious or buggy script could easily corrupt the environment for subsequent scripts – by deleting or replacing built-in objects, leaving unexpected objects in the root namespace, etc.
Therefore a careful evaluation of script trust level is always a good idea, combined with an assessment of the impact of a corrupted script environment. Some applications such as web browsers rely on the ability to execute multiple untrusted scripts within the same context, and have robust mechanisms for detecting and discarding contexts that are stuck or otherwise problematic.
The bottom line is that if it's important that all scripts start out in a pristine environment, and you cannot assume that each script will clean up after itself, then you should create a new context (script engine instance) for each one.
On a separate note, if you need to expose a host object in a way that allows it to be replaced later, consider using dynamic property assignment instead of
AddHostObject():engine.Script.foo = new Foo(); // or engine.Script["foo"] = new Foo();
