Add to Chrome
Add to Edge
Add to Firefox
Add to Opera
Add to Brave
Add to SafariWhen comparing Apiary vs Servant, the Slant community recommends Servant for most people. In the question“What are the best Haskell web frameworks for building RESTful web services?” Servant is ranked 3rd while Apiary is ranked 11th. The most important reason people chose Servant is:
Routes are defined end-to-end with type declarations. This allows your API to be introspective, serving its own documentation and derived query code, via jQuery for Javascript and `servant-client` for Haskell.
Question Ranking
Pros
Pro Automatically generates API documentation
Apiary generates comprehensive HTML documentation based on your API routes.
Pro Type-level routes
Apiary provides type modeling, down to URL parameters, ensuring type safety.
Pro Automatic documentation and Haskell/JS query generation
Routes are defined end-to-end with type declarations. This allows your API to be introspective, serving its own documentation and derived query code, via jQuery for Javascript and servant-client for Haskell.
Pro Excellent documentation
We Haskellers are accustomed to flying completely blind when using many libraries, luckily with Servant this isn’t the case. It features comprehensive tutorials, literate haskell examples for many use cases and great Haddock documentation for everything else.
Pro No Template Haskell
Everything is done within the confines of the existing type system, which helps keep compile times down.
Pro Nice ecosystem
Fairly large ecosystem of related modules for additional functionality.
Cons
Con Mediocre documentation
Users must rely primarily on the example project and auto-generated API documentation (Haddocks). The author of Apiary is not a native English speaker and openly acknowledges this issue.
Con Complicated internals
Extending Servant's existing functionality can be made difficult by the complexity of its type level machinery. Lucky most use cases are able to avoid this.
Con Route specifications and implementations are only connected by their position in a large type list
You actually have to count the index of the entry where you changed the specification, and then go and change the entry at the same index in the list of implementation methods. There is no other indication that the two are connected. This along with complex and verbose route definitions, makes it very hard to safely make changes to an API.
Con Route definitions are more verbose and complicated than other options
You are required to define a number of separate complicated types and their implementations which are usually spread out over a number of files. This makes it hard to figure out the API.
