Check types of RunParameter to catch errors and give user feedback

[LeonardHd]

ADF only supports specific data types for what we call RunParameter and PipelineRunVariable.

We support python primitives that map to the data factory types.
For RunParameters this should be:

• str -> String,
• int -> Int,
• float -> Float,
• bool -> Bool
• list -> Array
• dict -> Object
• str -> SecureString

For PipelineRunVariable these should be:

• str -> String,
• bool -> Boolean,
• list-> Array,
• int-> Integer,

Currently we do not validate any value on RunParameter and PipelineRunVariable.

[yehiaelbehery]

Hi @LeonardHd, I can work on this.

[LeonardHd]

@yehiaelbehery Thanks for offering help. I will share some more details as this might help.
The example might be a bit too ambiguous, so I will collect some more details tomorrow 👍

[LeonardHd]

@arjendev I updated the issue. In fact, we do not validate anything at the moment in the framework.
I think simple guard clauses would be enough so only valid RunParameter and PipelineRunVariable could be instantiated. What do you think?

@yehiaelbehery let's see what Arjen's point of view here is, as he leads the API concepts.

[arjendev]

Sounds good!

@LeonardHd, perhaps we can also introduce the check to verify we pass in the right type of the output argument in activity.set_result and the output argument in state.add_activity_result. These should only accept dictionary and lists iirc. This was the place where you accidently pushed in a tuple and got an internal error, right?