Redo Visualization

[HSMarieK]

Complete overhaul of the visualization for better human interface.

[windisch]

@HSMarieK: If I see correctly, line.setup_draw() is not needed any more, however, its still called from lineflow.simulation.environment.LineSimulation - do we have to adapt this?

[HSMarieK]

@windisch: That is the case, the line.setup_draw() Method does not have to be called there. I still have to take a closer look at lineflow.simulation.environment.LineSimulation to rule out any bugs.

[HSMarieK]

@windisch: Should I clean up the rest of the (now) redundant draw() Methods in the Objects?

[windisch]

@HSMarieK: This is just awesome work - thanks a lot for your contribution!

I refactored things a bit and gave the methods more speaking names. Moreover, I also reworked how the visualization-data is handled (formerly, a data list was send to methods and they appended, which I think is not required) - please have a careful look!

[windisch]

is it possible to have the initial center on the "average position"?

For instance, we could set it to None initially and than use the first data package to compute the average position of stationary objects?

[HSMarieK]

That could be done. Might be a really simple calculation. I will look into that.

[windisch]

The data structure looks a bit wired at first glance. At the line object, we stack everything in a big list and then we have to sort it out. Do not have a suggestions right now how to make this smoother - thus fine for now!

[HSMarieK]

If someone has a better idea, I am open for suggestions. This felt like the most obvious approach to me.

[windisch]

Instead of sending a list of dicts, we could send an object of a (yet to be implemented) class VisualizationData (or ConnectionData, but I like VisualizationData more, its what it is :-)) where everything can be accessed using attributes?

[HSMarieK]

Maybe. I went with dicts since I knew they could be serialized. I thought of using a dataclass instead, but I wanted it to work first.

[windisch]

We can refactor this later - should be no problem to serialize a own class if everything stays pure python.

[peerplexit]

Hey @HSMarieK,
I also tried out the new visualisation. Really cool features, I like it!
I noticed a small bug. When the simulation starts, I think the first state visualized is the state T=1.
This leads to the minor bug, that if the step from T=0 -> T=1 takes long. The pygame window stays white.
Also if you press Shift+H in this "slow mode" the simulation is not closing immediately. It first finishes the tick and then closes. Idk if this is the expected behaviour.

You can reproduce it by yourself if you turn up the factor parameter of a line (for example 20 -> 20 seconds until you see the line).

[HSMarieK]

@peerplexit Thanks. I will look into the visualized simulation states. But the halting process by pressing Shift+H is working as intended. This seems slow, since the simulation is sleeping and thus not reacting to user input until it steps again.