What Is the Observer Pattern? | 283However, by using a single call to the web service and then broadcasting to the mul-
tiple instances, you need far fewer service calls.
In addition to being more efficient, a central data source guarantees that every
instance gets the same information. Imagine a change in data from one web service
call to the next where a major change occurs. The first instance calls the service and
formats the data into a table, and the second instance calls the service to format the
data in a bar chat. The data in the chart does not reflect the data in the table, even
though it’s supposed to. Using an Observer pattern, a single call to the web service
always sends data from the same callto all subscribers. So the data sent to the table
formatting instance and that sent to the charts is guaranteed to be from the same set.
In applications where a high rate of data change occurs, the Observer pattern helps
to cut down on the bookkeeping. All data are sent to a central source and then dis-
tributed to subscribing instances. For example, in an action game, the score keeps
changing as many game conditions rapidly update. When an object in an action
game is “destroyed,” it no longer needs the information and should not keep gather-
ing in data. Likewise, objects that come into the game or are “resurrected” need to
start getting data. This can be a programming nightmare without some kind of sys-
tem to take in all data changes and then uniformly distribute those changes to the
different game elements while taking care of all subscription changes. Here the
Observer pattern comes to the rescue by handling all data collection and
distribution.
Key Features
The central feature of the Observer pattern is that state change is gathered in one
place and sent to all subscribing units. This one-to-many relationship allows devel-
opers to create loosely coupled classes and yet maintain information consistency.
Figure 8-1 shows the general relationship between the initial data source, the subject
class, and the observer classes:
Sending information in this manner to different classes is not only efficient, but also
allows for expansion. For example, suppose stock data is subjected to different types
of statistical models, each one encapsulated in a class. As new analytical models are
Figure 8-1. Observer information flow
State change SubjectObservers