[joestr87]

I think the main thing you’ve got to ask yourself when abstracting duplicated code is: are these things very similar because they are intrinsically linked or is it slightly coincidental. I.e if we ever have to change B will we always want to change A in the same way?

If the answer is that you’re not sure, it’s probably better to duplicate, and see what the case is when it needs to change again - and then decide whether to abstract or not.

[kazinator]

The main thing you have to ask yourself, will these duplicates ever need to be special-cased, and how hard will it be? Will someone have to write monolithic handling for that case from scratch, or are there ways of hooking or overriding individual behavioral choices?