Monthly Downloads: 69
Programming language: Haskell
License: BSD 3-clause "New" or "Revised" License
Tags: Data     Configuration     Configurator    
Latest version: v0.2.4

configurator-pg alternatives and similar packages

Based on the "configurator" category.
Alternatively, view configurator-pg alternatives based on common mentions on social networks and blogs.

Do you think we are missing an alternative of configurator-pg or a related project?

Add another 'configurator' Package


What is this?

This is a simplified version of the resting configurator-ng, aimed particularly to offer users of configurator-ng such as [PostgREST](postgrest.org) an easy path to migrate to a package that compiles with modern GHC versions and that continues to read existing configuration files.


configurator-pg skips some of configurator-ng's features, and changes the API in other places:

  • No configuration file reloading.
  • Simplified parsing API:
    • There is no type-class based value parsing; you need to supply explicit value parsers.
    • There's only load to read and evaluate a configuration file, and runParser to extract configuration values.
  • Simpler error handling and improved logging of parse errors.
  • Simplified handling of configuration subsets. There's subassocs and the unit tests pass, but the author didn't attempt to understand the original implementation fully.


The original configurator-ng is due to MailRank, Inc., Bryan O'Sullivan and Leon P Smith.

The low-level parser (Data.Configurator.Syntax) was initially mostly unchanged, but has since been rewritten with Megaparsec for better error messages. The evaluation (Data.Configurator.Load) is still close to the original. The high-level parser (Data.Configurator.Parser) is original.

File format

In short, the file format supports:

  • A simple but flexible configuration language that supports several of the most commonly needed types of data, along with interpolation of strings from the configuration or the system environment (e.g. $(HOME)).

  • An import directive allows the configuration of a complex application to be split across several smaller files, or common configuration data to be shared across several applications.

The format is more fully documented in the packages configurator and configurator-ng.

Here's an example:

# listen address
hostname = "localhost"
port = 8000

logdir = "$(HOME)/logs"
logfile = "$(logdir)/log.txt"
loglevels = [1, 4, 5]

users {
  alice = "[email protected]"
  bob   = "[email protected]"

# passwords.txt might contain
#   alice = "12345"
#   bob   = "sesame"
passwords {
  import "secrets/passwords.txt"


The following code can be used to parse the example above.

import Data.Configurator

data Settings = Settings
  { hostname  :: Text
  , port      :: Int
  , logfile   :: Maybe FilePath
  , loglevels :: Maybe [Int]
  , users     :: [(Text, Text)]
  , passwords :: [(Text, Text)]

settingsParser :: Parser Config Settings
settingsParser =
    <$> required "hostname" string
    <*> (Maybe.withDefault 1234 <$> optional "port" int)
    <*> optional "logfile" (pack <$> string)
    <*> optional "loglevels" (list int)
    <*> subassocs "users" string
    <*> subassocs "passwords" string

loadSettings :: IO Settings
loadSettings = do
  cfg <- load "settings.cfg"
  case runParser settingsParser cfg of
    Left err       -> die $ "reading config: " <> err
    Right settings -> return settings

Though note that for no apparent reason, subassocs returns the full key, whence the parsed list of users will be

    [ ("users.alice", "[email protected]")
    , ("users.bob", "[email protected]")