chars = dict(sorted(crops.items(),key=lambda kv: kv[1].size, reverse=True)[:4])
I am creating a dictionary, chars, that maps count variables (any integer vals) to my four largest crop values from a previous dictionary, crops.
I am running the above line in both Google Colab and Pycharms and it gives me a different dictionary in each. In Colab, it gives me what I want, from the list seen below (Note: lots of lines were taken out of the list). It retains the order of the list { (6, crop1), (7,crop2) , (2, crop3) , (3, crop4) } .
However, in pycharms the output is { (2, crop3) , (3, crop4), (6, crop1), (7,crop2)} , where it is still getting the largest four crops, but by casting it to dict() it seems to reorder my list in the dictionary by the count variables.
Chars list:
[(6, array([[78, 76, 77, 78, 73, 81, 80, 81, 79, 72, 77, 78, 71, 73, 78, 81,
82, 79, 76, 74, 75, 85, 86, 78, 76, 84, 85, 80, 78, 73, 76],
..., (7, array([[89, 90, 87, 83, 82, 83, 77, 76, 75, 74, 71, 73, 74, 77, 77, 80,
77, 75..., (3, array([[ 79, 77, 81, 74, 80, 85, 83, 79, 74, 77, 74, 76, 78,
...]
I am not sure what is causing this and could definitely use some help. Maybe an issue is also for PyCharms the code is running realtime in a Gazebo simulation, so maybe that changes how dict() is casting my list?
>Solution :
Dictionaries are an inheritently unordered data structure – period.
Dictionaries in python (*) are insert sorted – the first item put into it is also the first printed.
(*) holds for python 3.7 and above – below 3.7 dicts may be randomly sorted or (implementation detail) be insert sorted as well.
If you get different sortings you either insert in a different order OR run wildly different versions of python.
How do I detect the Python version at runtime?
In case you use an earlier version of python >= 3.1 you can use collections.ordereddict