cassava VS witch

I use it for everything: tracking personal finance and tax data (https://hackage.haskell.org/package/hledger), small scripts to gather online information that I want to track (https://hackage.haskell.org/package/cassava), sending alerts to my mobile device, etc...there's too much to list.

cassava?

I personnaly use cassava which should have everything you need (even though it can be quite obscure some times). I also know about Frames which might reduce some boiler plate at the price of a step up in complexity (disclaimer I've never use it, but it's author is a serious guy so I'm sure this package as some benefits).

I have done a lot of CSV development! I usually use the cassava library, though I have my own library as well. The cassava library uses FromRecord, FromNamedRecord, ToRecord, and ToNamedRecord type classes for parsing and rendering records, and it uses FromField and ToField type classes for parsing and rendering fields. An identifier type like the UserName example above should declare instances for FromField and ToField in order to be used in CSV files. For types that have appropriate Render and Parse instances, I implement general functions named something like parseFieldWithTTC and toFieldWithTTC, which allows me to declare instances as follows:

Thanks for mentioning Witch! I'm obviously biased because I wrote it, but it's a great library :)

See this issue for some discussion: https://github.com/tfausak/witch/issues/2

The witch package provides type classes for converting between arbitrary types: text, numbers, and anything else that people create instances for. It is therefore much more broadly applicable compared to TTC, which only works with text. I really appreciate that both a From type class (for conversion without failure) and a TryFrom type class (for conversion that may fail) are used, and the implementation looks very convenient. I will remember this project and use it when I have a suitable application in the future.