Quickguide

You definitely need to go through more docs but this is the rough outline of what happens when you write ecto code: you define Python/C++ cells that belong to a Python module. You then use that module to create a graph called plasm, and you execute it.

Ecto can be both pure C++ or pure Python but we’ll show you a combination of both as it provides the speed/tightness of the first one and the flexibility of the second one.

You need to go through the following steps:

• If you create C++ cells, you need to define your ecto module through an ectomodule CMake macro:

  ectomodule(my_ecto_module_name INSTALL
                                 DESTINATION ./here/different_name
                                 module.cpp
                                 awesome_file1.cpp
                                 awesome_file2.cpp
             )
  

• You need a module.cpp file that simply defines the Python module containing your cells

  #include <ecto/ecto.hpp>
  ECTO_DEFINE_MODULE(my_module) { }
  

• each cell needs to define 4 C++ functions

  static void
  declare_params(tendrils&)
  static void
  declare_io(const tendrils&, tendrils&, tendrils&)
  void
  configure(const tendrils&, const tendrils&, const tendrils&)
  int
  process(const tendrils&, const tendrils&)
  

• you create a Python script to create your graph called plasm and run it:

  #!/usr/bin/env python
  import ecto
  import my_awesome_cpp_ecto_module
  import my_awesome_python_ecto_module
  
  # create a plasm
  plasm = ecto.Plasm()
  
  # create some cells
  cell1 = my_awesome_cpp_ecto_module.MyAwesomeCell1(param1=whatever1)
  cell2 = my_awesome_python_ecto_module.MyAwesomeCell2(param2=whatever2)
  
  # connect those cells in the plasm
  plasm.connect(cell1['output'] >> cell2['input'])
  
  # execute the graph
  plasm.execute(niter=2)
  

Well, that seems like a lot of stuff to learn, really, what do I gain ? Well, let’s see from the above:

• you create a graph and execute it without recompiling code
• your mix C++ and Python transparently
• you have access to multithreading and scheduling transparently

Pretty good for a first start but ecto offers way, way more like:

• more complex graphs than DAGs (like condition/for loops)
• static introspection of cells
• introspection of cell inputs/outputs during execution
• runtime parameter change
• serialization
• pure C++ or pure Python code
• GUI’s
• ...

Keep on reading ...