I have a custom property.
On my code, I copy this as a template, and modify accordingly, without updating the original custom property.
Consider a reference dictionary with nested dictionary and color set to red.
When my code runs, I will copy the reference and modify the color values.
The shallow color will not get modified
The deep color will be modified. WTH!!!
This is simply how jython (and Python in general) handles object assignments. You get references instead of copies for anything that isn't a primitive type (number, string).
You can perform shallow copy by passing the original dictionary-like object to the dict() function (constructor, really). Note that you'll get a shallow copy--a new dict, but any objects within it will be references.
To truly break off your own modifiable dictionary, you need a deep copy. Python stdlib has a function that does this for pure python types.
See my integration toolkit for Ignition helpers. (Or do your own recursive function.)
I also encountered deepcopy throwing an error, so went to look another options.
But why do you need to deepcopy an object that was already encoded-decoded:
def deepcopy(node):
"""
Create a deep copy of nested data structures.
Recursively copies dictionaries, iterables, and primitive values to create
a completely independent copy of the original data structure. Handles nested
combinations of dicts, lists, and other iterables.
Args:
node: The data structure to copy (dict, list, tuple, or primitive value)
Returns:
A deep copy of the input node with the same structure but independent objects
"""
if hasattr(node, 'items'):
return {k: deepcopy(v) for k, v in node.iteritems()}
elif hasattr(node, '__iter__'):
return map(deepcopy, node)
else:
return node
Whenever I've used a naive deepcopy with a perspective array, it hands back a new array of the original QVs. That is, a shallow copy. So too with nested objects inside Perspective objects or arrays.
