Now that we have the is operator for type checking the only missing piece for declarative schema checks is a way to throw compile errors declaratively.
Ideally we would check some boolean value and provide a useful compile message. So the declarative operator should probably take a tuple payload that contains both a message similar to the assert keyword. The main difference is that the new syntax should abort with an error as expected.
fail "Expected string field foo! in struct";
If we introduce a declarative compile failure then we inevitably want to ask if we should be able to trap those failures somehow. My gut says no but it's possible that we will have a good use case for this in the future sometime. If we we can look at introducing that feature then.
Now that we have the
isoperator for type checking the only missing piece for declarative schema checks is a way to throw compile errors declaratively.Ideally we would check some boolean value and provide a useful compile message. So the declarative operator should probably take a tuple payload that contains both a message similar to the assert keyword. The main difference is that the new syntax should abort with an error as expected.
If we introduce a declarative compile failure then we inevitably want to ask if we should be able to trap those failures somehow. My gut says no but it's possible that we will have a good use case for this in the future sometime. If we we can look at introducing that feature then.