Hackage 2 and acid-state vs traditional databases

Johan Tibell johan.tibell at gmail.com
Thu Sep 6 20:59:02 CEST 2012


On Thu, Sep 6, 2012 at 11:49 AM, Ian Lynagh <ian at well-typed.com> wrote:
> Someone pointed out that one disadvantage of traditional databases is
> that they discourage you from writing as if everything was Haskell
> datastructures in memory. For example, if you have things of type
>     data Foo = Foo {
>         str :: String,
>         bool :: Bool,
>         ints :: [Int]
>     }
> stored in a database then you could write either:
>     foo <- getFoo 23
>     print $ bool foo
> or
>     b <- getFooBool 23
>     print b

Using Bryan's mysql-simple library makes mapping between Haskell data
types and SQL records quite straightforward (you write one type class
instance).

> Has anyone else got any thoughts?

I've argued in the past that we should use a SQL database, for mostly
the same reasons as you gave above, with the addition that I don't
anything but old and battle-tested technology with data as important
as the Hackage data.

-- Johan



More information about the cabal-devel mailing list