conclusion.conclusion.tex

   1 This thesis introduces a novel system for adding \gls{IoT} functionality to
   2 the \gls{TOP} implementation \gls{iTasks}. A new view for the existing
   3 \gls{mTask}-\gls{EDSL} has been created which compiles the program
   4 into bytecode that can be interpreted by a client. Clients have
   5 been written for several microcontrollers and consumer architectures which can
   6 be connected through various means of communication such as serial port,
   7 wifi and wired network communication. The bytecode on the devices is
   8 interpreted using a stack machine and provides the programmer with interfaces
   9 to the peripherals. The semantics for \gls{mTask} try to resemble the
  10 \gls{iTasks} semantics as close as possible.
  11
  12 The host language has a proven efficient compiler and code generator. The
  13 compilation is linear in the amount of instructions generated and is therefore
  14 also scalable. Moreover, compiling \glspl{Task} is fast because it is nothing
  15 more than running some functions native to the host language and there is no
  16 intermediate \gls{AST}.
  17
  18 The dynamic nature of the client allows the microcontroller to be programmed
  19 once and used many times. The program memory of microcontrollers often
  20 guarantees around $10.000$ write or upload cycles and therefore existing
  21 techniques such as generating \gls{C} code are not suitable for dynamic
  22 \gls{Task} environments. The dynamic nature also allows the programmer to
  23 design fail-over mechanisms. When a device is assigned a \gls{Task} but another
  24 device suddenly becomes unusable, the \gls{iTasks} system can reassign a new
  25 \gls{mTask}-\gls{Task} to another device that is also suitable for running the
  26 \gls{Task} without needing to recompile the code. It also showed that adding
  27 peripherals is not a time consuming task and does not even requires
  28 recompilation of clients not having the peripheral.