\subsection{Framework}
-Systems built with support for \gls{mTask} are often following the same design
+Systems built with support for \gls{mTask} often follow the same design
pattern. First the devices are created --- with or without the interaction of
the user --- and they are then connected. When all devices are registered, the
\gls{mTask}-\glspl{Task} can be sent and \gls{iTasks}-\glspl{Task} can be
peripherals. The following program shows a system containing two devices. The
first device --- the sensor --- contains a temperature sensor that measures the
room temperature. The second device --- the actor --- contains a heater,
-connected to the digital pin \CI{D5}. Moreover, this device contains a led to
-indicate whether the heater is on. The following code shows an implementation
-for this. The code fully uses the framework. Note that a little bit of type
-twiddling is required to fully us the result from the \gls{SDS}. This approach
-is still type safe due to the type safety of \CI{Dynamic}s.
+connected to the digital pin \CI{D5}. Moreover, this device contains an
+\gls{LED} to indicate whether the heater is on. The following code shows an
+implementation for this. The code makes use of all the aspects of the
+framework. Note that a little bit of type twiddling is required to fully use
+the result from the \gls{SDS}. This approach is still type safe due to the type
+safety of \CI{Dynamic}s.
\begin{lstlisting}[caption={Thermostat example}]
thermos :: Task ()
>>= \stm-> sendTaskToDevice "sensing" sensing (nod, OnInterval 1000)
>>= \(st, [t])->sendTaskToDevice "acting" acting (stm, OnInterval 1000)
(\(BCValue s)->set (BCValue $ dynInt (dynamic s) > 0) (shareShare nod a))
- >>| treturn ()
+ >>* [OnAction (Action "Shutdown") $ always $ deleteDevice nod >>| deleteDevice stm >>| shutDown 0]
where
dynInt :: Dynamic -> Int
dynInt (a :: Int) = a
\subsection[Lifting mTasks to iTasks-Tasks]%
{Lifting \gls{mTask}-\glspl{Task} to \gls{iTasks}-\glspl{Task}}
-If the user does not want to know where and when a \gls{mTask} is actually
-executed and is just interested in the results it can lift the \gls{mTask} to
+If the user does not want to know where and when an \gls{mTask} is actually
+executed and is just interested in the results, it can lift the \gls{mTask} to
an \gls{iTasks}-\gls{Task}. The function is called with a name, \gls{mTask},
device and interval specification and it will return a \gls{Task} that finishes
if and only if the \gls{mTask} has returned.
\end{lstlisting}
The factorial function example from Chapter~\ref{chp:mtaskcont} can then be
-lifted to a real \gls{iTasks}-\gls{mTask} with the following code:
+lifted to a real \gls{iTasks}-\gls{Task} with the following code:
\begin{lstlisting}[caption={Lifting the factorial \gls{Task} to \gls{iTasks}}]
factorial :: MTaskDevice -> Task BCValue
factorial dev = enterInformation "Factorial of ?" []
and oxygen saturation sensor add-on is a \textsc{PCB} the size of a fingernail
with a red \gls{LED} and a light sensor on it. Moreover, it contains an
\textsc{I2C} chip to communicate. The company producing the chip provides the
-programmer with example code for \gls{Arduino} and \textsc{mbed}. The sensor
-emits red light and measures the returning light intensity. The microcontroller
-hosting the device has to keep track of four seconds of samples to determine
-the heartbeat. In the \gls{mTask}-system, an abstraction is made. The current
-implementation runs on \textsc{mbed} supported devices.
+programmer with example code for \gls{Arduino} and \gls{mbed}. The sensor
+emits red light and measures the intensity of the light returned. The
+microcontroller hosting the device has to keep track of four seconds of samples
+to determine the heartbeat. In the \gls{mTask}-system, an abstraction is made.
+The current implementation runs on \gls{mbed} supported devices.
\subsubsection{\gls{mTask} Classes}
First, a class has to be devised to store the functionality of the sensor. The
heartbeat sensor updates four values continuously, namely the heartbeat, the
-validity of the reading, the oxygen saturation and the validity of it. For
-every value a function is added to the new \CI{hb} class. Moreover, the
-introduced datatype housing the values should implement the \CI{mTaskType}
-classes. The definition is as follows:
+oxygen saturation and the validity of the two. For every value a function is
+added to the new \CI{hb} class. Moreover, the introduced datatype housing the
+values should implement the \CI{mTaskType} classes. The definition is as
+follows:
\begin{lstlisting}[caption={The \texttt{hb} class}]
:: Heartbeat = HB Int
| BCValidHB
| BCGetSP02
| BCValidSP02
- | ...
instance hb ByteCode where
getHb = tell` [BCGetHB]
The bytecode instructions are added but still the functionality needs to be
added to the device interface to be implemented by clients. The following
addition to \CI{interface.h} and the interpreter shows the added instructions.
+When adding a peripheral, the devices not having the peripheral do not need to
+have their code recompiled. New instructions always get a higher bytecode
+number if added correctly. The peripheral byte in the device specification by
+default shows a negative flag for every peripheral. Only the peripherals added
+will be flagged positive.
\begin{lstlisting}[caption={Adding the device interface}]
// interface.h
repositories / msc-thesis1617.git / blobdiff