Re:Prefer duplication over the wrong abstraction
by unabst on Saturday July 09, 2016 @01:38PM
(#12062411)
Is it just me or did the author just provide the perfect argument for pattern matching in function calls as seen in Elixir? One can easily split/fork the functionality of the same abstraction depending on the arguments. This preserves the names (which preserves the abstraction), but extends the functionality to adapt to the new use case in parallel, without any dependencies.
Also, by "wrong abstraction", the author is arguing for abstractions to be backed by reasoning, not circumstances a coder might find themselves within code. Duplicate code is not itself reasoning, but a circumstance. Though it can be abstracted away, elimination cannot be the only reason.
This also highlights why more languages need features to help the coder who finds themselves in these situation. Circumstance is a context, and is driven by dependency. Again, dependency is the enemy. We need tools and code that fight it, not cause or encourage it.
Also, by "wrong abstraction", the author is arguing for abstractions to be backed by reasoning, not circumstances a coder might find themselves within code. Duplicate code is not itself reasoning, but a circumstance. Though it can be abstracted away, elimination cannot be the only reason.
This also highlights why more languages need features to help the coder who finds themselves in these situation. Circumstance is a context, and is driven by dependency. Again, dependency is the enemy. We need tools and code that fight it, not cause or encourage it.