There is something that bothers me with that text, I can't get to grasp what it is... 2011/7/1 Tobias Schoofs

** I am studying Haskell database libraries. In particular I am looking at haskelldb and the Takusen Database.Enumerator library.

In haskelldb, there are very good ideas aiming to represent database objects as (typeable) Haskell objects, instead of embedding SQL in plain strings. This is really an interesting approach that overcomes the hiatus between the database there and the processing logic here.

Unfortunately, the library cannot deal with some obvious issues related to this concept; conflicts arise, for instance, when Haskell keywords are used as names of attributes or tables or when the same name is used for an attribute and for a table (which is perfectly legal in SQL). Also, haskelldb cannot cope with peculiarities of popular database tables, such as ISAM tables - which are case sensitive.

Another, more conceptual issue, is the lack of database cursors. Database programs, usually, do not just issue isolated SQL statements, but implement a processing logic with nested queries and DML statements. A DB library should provide control structures to ease this kind of processing.

A library that tackles this issue, is the Takusen Database.Enumerator, again, a library with very strong ideas. Unfortunately, in Takusen there is nothing like the haskelldb DSL.

Concerning, cursors, Takusen uses iterators to stream query results. This makes a lot of sense. The iterators, however, cannot be used to implement the nesting of queries.

The user has to provide an additional action which is responsible for all the cursor-related logic. The tools provided to deal with cursors resemble very much the instructions found in other database-oriented languages, such as PL/SQL. This leads to verbose code where cursor control is implemented in a quite "imperative" way. The user, however, should be concerned with getting her queries right, not with the "technical" details of fetching from cursors.

In summary, what I miss is a database library that - integrates haskelldb approach with - a "functional-style" cursor concept

Is anybody working on something like this?

Thanks,

Tobias

_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe

Sorry for the awful message format in the first try! I am studying Haskell database libraries. In particular I am looking at haskelldb and the Takusen Database.Enumerator library. In haskelldb, there are very good ideas aiming to represent database objects as (typeable) Haskell objects, instead of embedding SQL in plain strings. This is really an interesting approach that overcomes the hiatus between the database there and the processing logic here. Unfortunately, the library cannot deal with some obvious issues related to this concept; conflicts arise, for instance, when Haskell keywords are used as names of attributes or tables or when the same name is used for an attribute and for a table (which is perfectly legal in SQL). Also, haskelldb cannot cope with peculiarities of popular database tables, such as ISAM tables - which are case sensitive. Another, more conceptual issue, is the lack of database cursors. Database programs, usually, do not just issue isolated SQL statements, but implement a processing logic with nested queries and DML statements. A DB library should provide control structures to ease this kind of processing. A library that tackles this issue, is the Takusen Database.Enumerator, again, a library with very strong ideas. Unfortunately, in Takusen there is nothing like the haskelldb DSL. Concerning, cursors, Takusen uses iterators to stream query results. This makes a lot of sense. The iterators, however, cannot be used to implement the nesting of queries. The user has to provide an additional action which is responsible for all the cursor-related logic. The tools provided to deal with cursors resemble very much the instructions found in other database-oriented languages, such as PL/SQL. This leads to verbose code where cursor control is implemented in a quite "imperative" way. The user, however, should be concerned with getting her queries right, not with the "technical" details of fetching from cursors. In summary, what I miss is a database library that - integrates haskelldb approach with - a "functional-style" cursor concept Is anybody working on something like this? Thanks, Tobias