copy.deepcopy when inserting elements in a ring

[simoneliuzzo]

I have spent several hours tracking back this issue. Thanks a lot to @lcarver, who spotted the problem.

I tried to insert elements in the lattice, thin correctors.

I forgot to add copy.deepcopy(element) so all elements were being updated as a family in later computations.

I think I am not the only one that fell into this trap. Below a code snippet:

thin_cor = at.ThinMultipole('ThinCorrector', [0], [0], max_order=0)
ring_thin = copy.deepcopy(ring)

for c, ic in enumerate(ind_cor):
    ring_thin.insert(ic+c, thin_cor)     # <<-- FAILS TO DO WHAT IS GENERALLY INTENDED, adds a family of magnets 
    ring_thin.insert(ic+c, copy.deepcopy(thin_cor))  # <<-- DOES WHAT IS GENERALLY INTENDED, adds independent elements

I find that in pyAT the use of copy.deepcopy() is necessary very often in order. It is in fact a really frequent reason for unreasonable results in my code.

Is there a way to make more clear to the user that an element is not an individual but it is linked to many others?
Is there a way to 'un-link' all magnets in a lattice? (independent_ring = [copy.deepcopy(el) for el in ring])

[lfarv]

No, you should not have to use deep copy() so often. For instance, to add individual elements, you would write:

for c, ic in enumerate(ind_cor):
    ring_thin.insert(ic+c, at.ThinMultipole('ThinCorrector', [0], [0], max_order=0))

Keep in mind that in python everything is a reference.
Also, deep_copy is costly: it has to investigate recursively there contents of an object. Creating a new one is more efficient !