Abstract Syntax
The abstract syntax of the DSL captures its
syntax and
static semantics.
The requirements of the railway language are listed below:
- A railway system consists of the infrastructure with the trackwork, the signalling equipment, and the stations, as well as the rolling stock with cars and locomotives.
- A railway network consists of a number of interconnected railway segments (on which trains may be present). The types of segments your language needs to support are listed below:
- Straights--the most trivial segment, allows a train to move straight. Has one incoming and one outgoing segment.
- Turnouts--allows a train to go straight, or take a diverging route to another track connected to the segment. Has one incoming segment and two outgoing segments. The direction of a turnout can be changed--it's either in straight mode, meaning the train will take the straight route, or in diverging mode, meaning the train will take the diverging route.
- Junctions--joins two segments. Has two incoming segments and one outgoing segment. The direction of a junction can be changed--it's either in straight or diverging mode, selecting the train which can enter the segment.
- Stations--like straights, but can also be at the end or beginning of a track. Has zero or one incoming and zero or one outgoing segment, but is always connected to at least one track. Each station has a unique name, that starts with a single upper case letter followed by zero or more lower case letters, and ends with zero or more numbers.
- Although an undesirable property at runtime, your metamodel should allow for more than one train to be present on a railway segment.
- A railway network does not allow loops (i.e., it should never be possible to start at a station, move in one direction, and end up at that same station). Tip: in metaDepth, use EOL's closure function, as you have to write it as one line of code. In AToMPM, use the available Javascript functions.
- The signalling equipment of the railway system consists of lights, one light is associated with each segment. These lights control the traffic on a segment and can be in two modes--red means traffic on that segment needs to wait, green means the traffic can pass. Segments only allow one-way traffic.
- Trains have a unique identifier, that start with an upper case letter, followed by four numbers. A train can be on at most one segment.
- Each train has an associated schedule. This schedule is modelled in a second domain-specific language. Its requirements are listed below:
- A schedule is associated with a train by referring to the name of the train.
- The schedule of a train tells it where to go--it contains a sequence of consecutive steps. The 'start' step contains the name of the station where the train will start. It has no incoming steps and one outgoing step. The 'end' step contains the name of the station where the train will stop. It has one incoming step and no outgoing steps. The start and end steps are mandatory. In between, the train can be instructed to take the diverging or straight route when it encounters a turnout. These 'in-between' steps (so-called "waypoints") have exactly one incoming and one outgoing step.
- Each train needs exactly one schedule.
- Each schedule needs exactly one train.
Operational Semantics
In this part of the assignment, you will model the semantics of a "control
room" that regulates the flow of trains on the railway network.
The goal is to get each train from its start station to its end station
according to its schedule, and this in a safe manner.
Safety is defined as "no two trains shall be on a given segment
during any step (i.e, at any given time)". This means that the control
room needs to make sure two trains can never be on the same track, as well
as switching turnouts and junctions to the correct position. More detailed
rules are listed below:
- The simulation is broken up into a number of "steps". In each simulation step, the control room first sets all lights to the correct "mode" and switches the direction of turnouts and junctions. After that, all trains are moved to the next segment, if allowed.
- In the initial simulation step, all trains are placed in their start station. If this initial step leads to an invalid state, such as having more than one train on a single (station) segment, the simulation shall halt with an appropriate error message.
- A train is allowed to move to the next segment if the light on its current segment is set to green. If the light is red, the train has to wait.
- If a train is on a turnout, it moves in the direction the turnout is set to. In other words, a train has no access to its schedule and follows the directions set by the control room.
- The control room iterates over all segments that contain a train. If no train is present on the segment the train wants to move to, it sets the light of the segment the train is currently on to green.
- If a train is on a turnout, the control room switches the direction to where the train wants to go, by looking at the next step in the schedule of the train. It is an error if there is no step (the schedule has reached the end) when a train wants to leave a turnout.
- In case of two trains wanting to enter a junction at the same time, the control room chooses one randomly (make sure your simulation is repeatable by choosing an appropriate seed for your random number generator).
- A train can only enter a junction if the direction of the junction is set correctly.
- When a train reaches its end station, it is removed from the model.
- The simulation ends when all trains have reached their end station, or no more moves are possible (the end station is not reachable). The simulation should end with an appropriate message in both cases.
An example of an execution/simulation trace is shown here.
For a more detailed explanation on railway operation and control, take a look at the book by Joern Pachl: http://www.joernpachl.de/roc.htm