Extend validator rules

[jan-ladleif]

We should add additional validation rules, now that we can specify ItemDefinitions and such. What I am thinking of from the top of my head:

• Timer events can now be specified, and impose some restrictions on their context depending on whether they are absolute or relative. For example:

  For relative timers: Only the Participants involved in the Choreography Activity that immediately precedes the Event would know the time. The sender of the Choreography Activity that immediately follows the timer MUST be involved in the Choreography Activity that immediately precedes the timer. [p. 341 in standard]

• Exclusive gateway conditions can only reference specific message data:

  The data used for Gateway Conditions MUST have been in a Message that was sent prior to (upstream from) the Gateway. More specifically, all Participants that are directly affected by the Gateway MUST have either sent or received the Message(s) that contained the data used in the Conditions. [p. 344 in standard]

  Since there is no required format for those conditions and they can be any formal expression, we cannot be sure which messages are referenced. However, I think we can guess to some degree, e.g., if there is a formal expression order.price > 50 we may assume there must be a message order sent upstream. If not, we can issue a warning.

• ... and maybe much more other stuff hidden in the standard somewhere.

Proposed Methodology

1. Go through the standard systematically and collect all constraints.
2. Consolidate these constraints, look for overlaps, etc.
3. Discuss whether we can check these constraints.
4. Implement those which we do not have yet.

[jan-ladleif]

Timer event validation was added in #46.

However, the standard does not at all make sense: would allow diagrams such as this: But obviously, C does not know whether the upper path has been taken, and just starting c at the first of January would not always be compliant.

[jan-ladleif]

The standard is too vague, PR #47 is closed.