Haskell records compare Standard ML

[atravers]

So the answer to the questions in your original post:

@AntC2

I can’t find any explanation/rationale for why Haskell’s records are so different vs SML. Does anyone know of any refs? Or can give some explanations?

…then must surely be something like:

It would improve the language to fix this. But this hasn’t risen to the level where it’s worth […] spending the cycles to fix the problem. That said, I’d welcome work in this direction.

Richard Eisenberg.

…as opposed to:

@AntC2

[…] GHC doesn’t care.

Q.E.D.