Selecting a nonexistent field using record dot yields error about missing HasField instance

[MorrowM]

Using GHC 9.4.4:

λ> :set -XOverloadedRecordDot
λ> data Person = Person { name :: String, age :: Int } deriving (Eq, Show, Ord)
λ> john = Person "John" 34
λ> john.address

<interactive>:4:1: error:
    • No instance for (GHC.Records.HasField "address" Person a0)
        arising from selecting the field ‘address’
    • In the expression: john.address
      In an equation for ‘it’: it = john.address

While this is technically a "correct" error message I believe we can do better. For example

<interactive>:4:1: error:
    • Cannot select nonexistent field ‘address’
        from a record of type ‘Person’
    • In the expression: john.address
      In an equation for ‘it’: it = john.address

Whether to mention HasField at all is perhaps a question, but I believe that if a user is advanced enough to potentially want to mess with HasField then they don't need the error message to spell that out for them.

[dixonary]

Seconding this. It's severe enough that I really think it will slow down the take-up of record dot syntax, especially among new learners.

Anything of the general shape that Morrow suggests would be streets ahead.

[Kleidukos]

Agreed, we can score big time with this feature amongst new users and this needs to be a prime concern.

[MorrowM]

The relevant GHC issue:
https://gitlab.haskell.org/ghc/ghc/-/issues/22382

[massung]

I know this issue is old, but wanted to add that - as a hobbyist Haskell user - this error popped up and it took me way too long to figure out that the actual issue was I was missing an import; the record wasn't in scope.

No instance for ‘GHC.Records.HasField
                         "login" Persist.User.User String’
        arising from selecting the field ‘login’

Having the error message be as suggested in the issue - and perhaps adding a little hint that the record type may not be in scope (not imported or misspelled, as with other error messages), would be very helpful.