Thanks Greg, both references are very interesting. If I understand correctly, the DSH approach is to convert Haskell programs into SQL and run them inside the database. This seems a good solution when the program objective is to change data in the database according to some business logic or to preprocess relational data to create a non-relational temporary resultset which is later used by an external procedure. The persistent way, from this perspective, appears to be more "conventional" - data are obtained from the database to be processed by an external program and presented on a website, for instance. Similar to Takusen, the main objective here appears to be compile-time guarantees on database types. Indeed, in a functional program, the string-based embedded SQL is the weakest link in the chain. persistent strengthens this link. On 07/01/2011 08:37 PM, Greg Weber wrote:

Hi Tobias,

Have you seen DSH [1]? You might also be interested in Persistent [2], but it sounds like it has different goals than what you are after.

[1] http://hackage.haskell.org/package/DSH [2] http://www.yesodweb.com/book/persistent