Custom Resource Types

A resource can grant its units by a rule of your own. This page shows how to write one.

---

What It Is

Some shops grant their units by a rule of their own, such as sharing a packer between customer classes by weight. A custom resource type replaces the rule by which a resource grants its units, and keeps the capacity that follows the registry.

---

How to Use It

To change how a resource grants units, subclass ResourceBase and name the class in type. ResourceBase is built on the hooks of SimPy's BaseResource, as a subclass of SimPy's Resource is. A subclass can change these:

• Override _do_put(request) to decide when a request is granted, and _do_get(release) to decide what a release gives back. An override calls the base method, which wakes when_busy() and when_idle(), so a breakdown that counts busy time still works.
• Set PutQueue to the class of the waiting line. It can be SimPy's SortedQueue, written staticmethod(functools.partial(SortedQueue, maxlen=2)). On a store or container, a type that sets its own queues refuses get_queue and put_queue.
• Set request to BoundClass(MyRequest), for a request class that subclasses DynamicResourceRequest, or set release the same way, for a release class that subclasses DynamicRelease. Both base classes are in desflow.resources. A request or release class calls the base constructor, which publishes the on_change reading.

This resource keeps how long each request waited:

YAMLBuilderLow-level

resources:
  # A subclass of ResourceBase that keeps how long each request waited
  packers: {current_cap: 1, max_cap: 4, type: !python shop_logic.LoggedPackers}

The class is in shop_logic.py, beside the blueprint:

class LoggedPackers(ResourceBase):
    """A resource that keeps how long each request waited before it was granted."""

    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)
        self.waits = []

    def _do_put(self, request):
        super()._do_put(request)
        if request.triggered:  # the request was granted just now
            self.waits.append(self.env.now - request.time)

app = SimulationContext(sim_id="Shop", factor=0.0)
app.add_resource("packers", current_cap=1, max_cap=4, type=LoggedPackers)

The class is the one the Low-level tab shows.

class LoggedPackers(ResourceBase):
    """A resource that keeps how long each request waited before it was granted."""

    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)
        self.waits = []

    def _do_put(self, request):
        super()._do_put(request)
        if request.triggered:  # the request was granted just now
            self.waits.append(self.env.now - request.time)

env = DynamicRealtimeEnvironment(factor=0.0)
env.registry.register_sim_parameter(
    SimParameter(
        sim_id="Shop", resources={"packers": CapacityConfig(current_cap=1, max_cap=4)}
    )
)
packers = LoggedPackers(env, "Shop", "packers")


def order(env):
    with packers.request() as request:
        yield request
        yield env.timeout(2.0)

---

How It Works

DynamicResource is ResourceBase with nothing changed. The constructor is DynamicResource's, because the resource reads its capacity from the registry. A subclass that defines __init__ therefore takes *args, **kwargs and passes them on. A PutQueue of your own sets its own order and length, so it takes no queue_limit. The tests run each of these beside the same SimPy subclass in all three APIs.