Right, I was thinking in terms of maintenance of the compiler itself. From an end-user perspective, I don’t really care if the language has this special syntax for struct updates. If I don’t like it I just won’t use it. But IIUC from the prior discussions in the GH issue that spawned this thread, the type checking on the struct update syntax is special-cased in the compiler itself. That’s what concerns me. But again, I’m not a compiler engineer, so only the Elixir core team can actually confirm or deny if this is really a problem for maintenance.
I just know that in the Typescript world, these kinds of early decisions didn’t cause problems until people started building very large enterprise apps. IIRC, Typescript has a three-pass type checker (don’t quote me on that), which is where a lot of this slowness comes from. Based on the prior discussions, I believe an additional pass was added to the compiler to accommodate that warning, so if that’s the only reason for that additional pass, I think it should be removed. 
