Popularity
6.6
Declining
Activity
0.0
Stable
13
7
1

Monthly Downloads: 7
Programming language: Yacc
License: BSD 3-clause "New" or "Revised" License
Tags: Database     PostgreSQL     Unclassified    
Latest version: v0.3

preql alternatives and similar packages

Based on the "Unclassified" category.
Alternatively, view preql alternatives based on common mentions on social networks and blogs.

Do you think we are missing an alternative of preql or a related project?

Add another 'Unclassified' Package

README

PreQL

Before you Post(gres)QL, preql.

A Haskell SQL library.

  1. Quickstart
  2. Vision

Current Status

preql provides a low-level interface to PostgreSQL and a quasiquoter that converts inline variable names to SQL parameters. Higher-level interfaces, checking SQL syntax & schema, are planned.

Quickstart

    import Preql -- The `Preql` module re-exports the interface useful to typical applications.
    import Control.Monad.Trans.Reader (runReaderT)

    main :: IO ()
    main = do
        conn <- connectdb "" -- Get a database connection with default connection string

        -- You can write a SQL instance to replace this ReaderT with
        -- your own application state, logging, error handling
        flip runReaderT conn $
            -- A simple query with no parameters
            cats <- query [sql| SELECT name, age FROM cats |]
            for_ cats \(cat :: (Text, Int)) -> print cat

            -- A query with parameters
            let minAge = 10
            oldCats <- query [sql| SELECT name, age FROM cats where age > ${minAge}|]
            for_ oldCats \(cat :: (Text, Int)) -> print cat

            -- A query that doesn't return rows
            query_ [sql| UPDATE cats SET age = 0 where age < 1 |]

            -- Two queries in a transaction
            moreOldCats <- runTransaction $ do
                maxAge <- V.head <$> query [sql| SELECT max(age) FROM cats |]
                query [sql| SELECT name FROM cats WHERE age = ${maxAge} |]
                -- Just an example; you could make this one DB roundtrip
            traverse_ putStrLn moreOldCats

Vision: Parsing SQL in Haskell Quasiquotes

Can GHC check that a query is sytactically correct at compile time, without having Postgres run it?

I want to

  • write standard (Postgres) SQL
  • support all of the Postgres features that I depend on in a single library
  • catch SQL syntax errors during Haskell type checking
  • catch SQL - Haskell type mismatches during Haskell type checking

To hedge my bets between the goal of supporting Postgres's extensive feature set & providing meaningful type errors, I plan to implement 3 quasiquoters:

  • parameter (and result column?) antiquotes, no attempt at interpreting the SQL
  • SQL syntax checking, without requiring a schema
  • with a schema provided, check that queries, parameters, and results are type-correct

I see several reasons to support literal SQL syntax:

  • many people already know it; learning it will be useful for a long time
  • non-Haskell developers on the project are also likely to know it
  • efficient interactive development of the SQL query, before parameterizing & inserting in Haskell code
  • easy adoption from existing code bases using postgresql-simple or hasql (or other SQL-string libraries)

Prior Art

Haskell libraries that I've used in production seem to fall into 3 categories:

  • SQL as strings, with syntax & types only checked when the query runs
  • simple non-SQL interfaces that are convenient for CRUD, but don't cover joins, aggregates, and other SQL features
  • complex EDSLs that cover much of SQL, with a new syntax & lots of fancy type hackery

Each approach addresses some of the goals under Vision, but none lets me both write literal SQL & provides a reasonable degree of type checking.

The most similar design that I've found in Haskell is hasql-th.

postgresql-typed also occupies a similar point in the design space. It provides a quasiquoter, and connects to Postgres at compile time to validate the queries. I haven't tried it out yet, but it seems like a very promising approach.

If you know of other libraries with similar ambitions, please let me know.