Is Tags.WAVELENGTH still necessary?

[TomTomRixRix]

Currently there are two different tags: Tags.WAVELENGTHS and Tags.WAVELENGTH. While the former is used to define all wavelengths used for simulation or processing, the latter is used to specify the currently processed wavelength or the wavelength to be visualized or tested. Tags.WAVELENGTH is iteratively updated in the global settings in the simulate method while looping through all wavelengths.

💡 Idea:
Remove the Tags.WAVELENGTH and:

1. In examples and tests replace the Tags.WAVELENGTH with the last entry from Tags.WAVELENGTHS as this value will anyways be set during simulation in simulate().

2. In all simulation modules and processing components we could pass the currently used wavelength from the simulate method to each adapter. Currently everything happens in the run method of the adapter, however, this should maybe be independent of wavelengths. The wavelength is currently only used for reading and writing the required data for each adapter. Therefore, two new abstract methods could be introduced: load_data and save_data. We could either
   
   a) write something like this

   data = pipeline_element.load_data(wavelength)
   data = pipeline_element.run(digital_device_twin, data)
   pipeline_element.save_data(wavelength, data)

   where we return the loaded and processed data and give it to the next function via arguments, or

   b) use such a solution

   pipeline_element.load_data(wavelength)
   pipeline_element.run(digital_device_twin)
   pipeline_element.save_data(wavelength)

   where the data is stored in attributes of the adapter.

Pros and Cons

• (b) would be cleaner as the data doesn't leave the adapter
• (a) would provide the opportunity to run the adapter without loading the data with the load_data method of the adapter but providing your own data to the adapter, e.g.

   my_adapter = Adapter()
   my_adapter.run(my_digital_device_twin, my_data)
   ``

Things to consider

• How is the RAM usage affected be these options?

[leoyala]

I like the idea of removing redundancy in the code. I would just make the methods more explicit about what they do, such as load_adapter_data or so. I am more of a fan of saving the data as attributes, which would imply also creating a new method to set it, for example set_adapter_data in case the data wants to be passed to the pipeline manually.