Connects polytropos open records to the standard Haskell CSV library cassava without intermediate nominal conversions: ToNamedRecord serializes field names and values straight from the flat Rep payload, and FromNamedRecord looks up each field label in the CSV header map in \(O(1)\).

Architectural Overview & Rationale

Heterogeneous Row Support

Unlike satellites that require homogeneous rows, polytropos-cassava is genuinely heterogeneous: the AllToField and AllFromField row constraint walks require the CSV field classes (ToField / FromField) at each field's own type, so a row mixing String, Int, and Double columns encodes and decodes directly.

Decoding is strict about field presence (a missing header column fails), but ignores extra columns in the CSV input, matching cassava's own generic behaviour.

When to Use This Module

• When reading or writing CSV files, spreadsheets, or tabular datasets using open records.

Worked Example

{-# LANGUAGE DataKinds #-}
{-# LANGUAGE OverloadedStrings #-}
{-# LANGUAGE TypeApplications #-}
{-# OPTIONS_GHC -fplugin=Polytropos.Plugin #-}

import qualified Data.Csv as Csv
import Polytropos
import Polytropos.Cassava

type UserRow = Record '["username" ':= String, "score" ':= Int]

row :: UserRow
row = REC { username = "Alice", score = 95 }

-- Encodes to CSV named record:
namedRec = Csv.toNamedRecord row