https://github.com/danidiaz/streamy

Using Backpack to abstract over the main Haskell streaming libraries.
https://github.com/danidiaz/streamy

Last synced: about 1 month ago
JSON representation

Using Backpack to abstract over the main Haskell streaming libraries.

• Host: GitHub
• URL: https://github.com/danidiaz/streamy
• Owner: danidiaz
• License: mit
• Created: 2017-08-09T17:03:15.000Z (over 7 years ago)
• Default Branch: master
• Last Pushed: 2017-09-16T15:05:03.000Z (over 7 years ago)
• Last Synced: 2024-08-03T21:02:12.133Z (5 months ago)
• Language: Haskell
• Homepage:
• Size: 102 KB
• Stars: 34
• Watchers: 6
• Forks: 1
• Open Issues: 3
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• awesome-cabal - `streamy`

README

        
# streamy

## What's this?

As a learning exercise, I'm trying to use the

[Backpack](https://github.com/ezyang/ghc-proposals/blob/backpack/proposals/0000-backpack.rst)

module system to give a common interface to the main streaming libraries of

Haskell: [conduit](http://hackage.haskell.org/package/conduit), [pipes](http://hackage.haskell.org/package/pipes), [streaming](http://hackage.haskell.org/package/streaming). (Inspired by how

[str-sig](http://next.hackage.haskell.org:8080/package/str-sig) gives a common

interface to String-like types.)

## Structure of this project 

- **streamy-sig** is the abstract signature.

- **streamy-bytes-sig** expands *streamy-sig* with a module signature for byte streams.

- **streamy-pipes** is the pipes "bridge" package.

- **streamy-streaming** 

- **streamy-conduit** 

- **streamy-testsuite** test each of the implementations using a shared set of tests.

## Do I need all these packages?

Not really, even if you still want to use Backpack to abstract your streaming

library.

You can always define a self-contained signature in your own library, and the

users of your library can define their own "bridge" implementations to make the

types line up.

The advantage of these premade packages is not having to invent your own

signatures and bridge modules.

## Building instructions

Built using [cabal 2.0](https://www.haskell.org/cabal/).

> cabal new-build all --enable-tests

> cabal new-test streamy-testsuite

## Where can I find further information on Backpack?

Edward Z. Yang's [thesis](https://github.com/ezyang/thesis/releases) is quite readable.

[Designing the Backpack signature

ecosystem](http://blog.ezyang.com/2017/03/designing-the-backpack-signature-ecosystem/)

explains the usefulness of signature thinning (see also section 2.7 of the thesis).

[Backpack for deep learning](http://blog.ezyang.com/2017/08/backpack-for-deep-learning/).

The [str-sig](http://next.hackage.haskell.org:8080/package/str-sig) signature

package and its various implementations.

I wrote a few Backpack tips & tricks

[here](https://medium.com/@danidiaz/backpacking-tips-3adb727bb8f7) and [here](https://medium.com/@danidiaz/backpacking-tips-ii-47fa86e5bf2).