Travis

Athougies

Travis

Athougies

• Navigation

  • Home
  • About
  • Projects
  • Contact
  • Archive

• Tags

  • haskell (17)
  • hydroponics (9)
  • math (8)
  • sustainability (5)
  • gardening (4)
  • philosophy (3)
  • beam (3)
  • ai (3)
  • web (2)
  • type theory (1)
  • social criticism (1)
  • programming (1)
  • physics (1)
  • harvey mudd college (1)
  • hakyll (1)
  • food (1)
  • finance (1)
  • ethics (1)
  • education (1)
  • databases (1)
  • business (1)
  • agriculture (1)
  • academia (1)
  • HoTT (1)

Site proudly generated by Hakyll

This site kept updated by Travis CI

Beam

Summary

Beam is a Haskell interface to relational databases. Beam is written in a pure Haskell, and doesn't make use of Template Haskell or non-intuitive embedded domain specific languages

Beam is a Haskell interface to relational databases. Its key strengths are its type-safety and intuitive syntax. Beam is written in pure Haskell, and doesn’t make use of Template Haskell or non-intuitive embedded domain specific languages.

Documentation and Tutorials

Beam has comprehensive documentation on its website.

A series of tutorial for the current version of Beam can be found there as well.

You may also be interested in the API reference on hackage and in the GitHub sources.

Use

Beam is ready to be used in production. The Beam API is fairly stable and breaking changes follow the Haskell PVP.

Beam is available on both hackage and stackage.

Beam has several backends:

• beam-sqlite – Beam driver for the SQLite embedded database. On hackage and stackage.
• beam-postgres – Beam driver for Postgres. On hackage and stackage.
• beam-mysql – Beam driver for MySQL. Currently unreleased pending further tests, but available on github.

There are some independent backends as well:

• beam-firebird – Beam driver for FireBird. Available on github.

Beam has a somewhat active IRC channel on freenode at #haskell-beam and a google discussions group beam-discussion.

Comparisons to other libraries

persistent

• Beam does not use Template Haskell. Persistent requires the use of template haskell or very large manual instance declarations to create a database schema. The Template Haskell schema specification uses a non-Haskell DSL making projects using persistent hard to understand to those not familiar with persistent.
• Beam supports basic SQL features like JOINs or aggregates. Persistent supports simple queries, but it does not support database-native joins or aggregations. Instead users are forced to use third-party libraries like esqueleto.
• Beam also supports advanced features Beam has full support for window functions and compound data types, like arrays. The postgres backend supports many postgres-specific types like json and jsonb.

opaleye

• Beam has simpler data types. Opaleye requires that each table type is parameterized over each column in the table. This means that if your table contains 16 columns, it needs to have 16 type-level parameters. As you can imagine, this leads to incredibly long types. On the other hand, beam types are parameterized over only one type.
• Beam does not need Template Haskell to derive instances and simplify types. Hiding type definitions behind Template Haskell makes it difficult to understand what is going on with the code behind the scenes. Beam uses generics and regular Haskell types which are easy to reason about.
• Beam generates readable SQL. Opaleye’s SQL can be very difficult to understand (see the example output on GitHub). Beam’s SQL is easy to read and reason about. This makes it easier to debug.

HaskellDB

• Again, Beam does not use Template Haskell. The Haskell type system is at an advanced enough state that we do not need to resort to Template Haskell EDSLs to do what most languages can in a straightforward manner. HaskellDB requires that types be created using the HaskellDB template haskell quoters.
• To be continued…