Missing validation for noerror on events

[joeseibel]

When a transition condition uses the keyword noerror, the standard's syntax definition shows that noerror should only be used with propagations and not with events. In section E.8, the standard has the error_condition_trigger grammar rule:

error_condition_trigger ::=
  error_behavior_event_identifier [ error_type_set ]
  | [ in ] incoming_error_propagation_point [ error_type_set_or_noerror ]
  | subcomponent_identifier . outgoing_error_propagation_point [ error_type_set_or_noerror ]

Note that the rule error_type_set_or_noerror only appears after an incoming propagation or an outgoing propagation in a subcomponent. It does not appear after an event, which instead uses the rule error_type_set.

However, the EMV2 grammar within OSATE permits noerror after an event. We should leave the grammar and meta-model as is and instead check this in the validator. The validator should mark an error in the following model:

package missing_validation
public
  system s
    features
      f1: in feature;
      f2: in feature;
    annex EMV2 {**
      use types ErrorLibrary;
      use behavior StandardIssue::machine1;

      error propagations
        f1: in propagation {CommonErrors};
        f2: in propagation {CommonErrors};
      end propagations;

      component error behavior
        events
          error1: error event {CommonErrors};
          error2: error event {CommonErrors};
        transitions
          --Valid
          transition1: state1 -[f1 {ServiceError} and f2 {noerror}]-> state2;
          --Not valid
          transition2: state1 -[error1 {ServiceError} and error2 {noerror}]-> state2;
      end component;
    **};
  end s;

  annex EMV2 {**
    error behavior machine1
      states
        state1: initial state;
        state2: state;
    end behavior;
  **};
end missing_validation;