> The way it currently works is arguably the way it should work, irrespective of the spec. The spec is how it should work, regardless of developer's opinions.

> The way it currently works is arguably the way it should work, irrespective of the spec. The spec is how it should work, regardless of developer's opinions. > Why should [a] evaluate differently than [b]? In my opinion (and I'm guessing in the opinion of the spec authors, though I can't be sure), saying "x += y" means "add y to x", so I'd expect the language to: 1) Figure out what "x" is once 2) Add "y" to it Your opinion seems to be that saying "x += y" means "desugar to x = x + y" and you expect the language to: 1) Figure out what "x" is as a value 2) Add "y" to the value of "x" 3) Figure out what "x" is as a location (which may differ from 1) 4) Store the result into "x" To me step 3 is very surprising, that "x" may differ, which is why I think the spec's definition is the more intuitive one (plus, it doesn't require knowledge of "desugaring" things).