This remains a major pain point for me in Haskell. I love the language and find that it's often one of the most powerful languages I've ever used, especially for highly concurrent programs safely utilizing multiple cores, but I'm sort of in awe of how shoddy the record syntax is.

For what it's worth, lenses can avoid some of these issues. For instance, they can allow you to have the same field name in multiple records:

    data RecordA { _a_name :: String }
    data RecordB { _b_name :: String }
    makeLensesWith underscoreFields ''RecordA
    makeLensesWith underscoreFields ''RecordB

You can now use something like `view name x` or `x^.name` for values of type `RecordA` or `RecordB`. Within the record definitions themselves, the field names still have to be disambiguated. The naming convention I chose here is pretty arbitrary. The first argument to `makeLensesWith` tells it how to parse the field names to get the "true" name.