[Haskell-cafe] Return of the revenge of the revisit of the extensible records, reiterated

[AntC]

> AntC <anthony_clayden <at> clear.net.nz> writes:
 
Actually, I think there's more wrong with that line than a typo:
> 
>     extendUnique :: (..., l :\r ) => ...
>                   -- s/b: r :\ l  ??
> 
(It's supposed to do renaming with non-duplicate labels?)

Talking of renaming, how does it go with duplicate labels?
The comment on `rename` says it can be expressed using the "above 
operations" (presumably restrict followed by extend with the new label, as 
per Gastar&Jones and Leijen).

If that's genuinely equivalent, then rename will 'unhide' any duplicate 
label. So presumably the implementation must split the HashMap into two 
keys, rather than changing the label on the existing Seq(?)


AntC