https://github.com/athanclark/purescript-arraybuffer-class

Simple encoding / decoding classes for ArrayBuffers
https://github.com/athanclark/purescript-arraybuffer-class

arraybuffer generics purescript typeclass

Last synced: 2 months ago
JSON representation

Simple encoding / decoding classes for ArrayBuffers

• Host: GitHub
• URL: https://github.com/athanclark/purescript-arraybuffer-class
• Owner: athanclark
• License: bsd-3-clause
• Created: 2018-12-16T22:55:23.000Z (over 5 years ago)
• Default Branch: master
• Last Pushed: 2021-05-09T06:43:30.000Z (about 3 years ago)
• Last Synced: 2024-04-01T01:09:58.669Z (3 months ago)
• Topics: arraybuffer, generics, purescript, typeclass
• Language: PureScript
• Size: 53.7 KB
• Stars: 5
• Watchers: 3
• Forks: 2
• Open Issues: 1
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Lists

• awesome-purescript - purescript-arraybuffer-class - Typeclasses for ArrayBuffer encoding and decoding. (Binary Serialization)

README

        
# purescript-arraybuffer-class

[![Build Status](https://travis-ci.org/athanclark/purescript-arraybuffer-class.svg?branch=master)](https://travis-ci.org/athanclark/purescript-arraybuffer-class)

[![Pursuit](https://pursuit.purescript.org/packages/purescript-arraybuffer-class/badge)](https://pursuit.purescript.org/packages/purescript-arraybuffer-class)

Typeclasses for binary serialization in PureScript.

> Note: Although this library shares a lot of features similar to the [cereal library from Haskell](https://hackage.haskell.org/package/cereal), it is not yet compatible.

## Usage

It should feel fairly similar to [Argonaut](https://pursuit.purescript.org/package/purescript-argonaut), but with a

binary spin on things - everything is very `Effect`-ful, and will throw errors if something doesn't line-up as expected.

Just implement the `EncodeArrayBuffer` and `DecodeArrayBuffer` typeclasses to support binary serialization for

your data type.

## Notes

- The `Array` instance first packs its length as a 4-byte `Uint32BE` - I don't know of a perfect way around this

  without excessive testing.

- The `Char` and `CodePoint` instances seem to reflect utf-8 pretty decently - Woo Hoo! Pure utf-8 serialization!! But wait...

- The `String` instance just turns it into an `Array CodePoint`, so there's an initial word denoting the string's length.

  Right now, I don't know of a clean way around this, without making every instance have a length argument.

- The `Vec` instance actually sidesteps the length argument, because the length is assumed to be known at compile-time.

- The `ArrayBuffer`, `DataView`, and `ArrayView` instances also have a similar initial length argument. Writing

  is not as efficient as you'd hope.

- The `Record` instance first turns the record into an `Object ArrayBuffer`, which then becomes a

  `Array (Tuple String ArrayBuffer)` - it's not super efficient.

- The `Generic` instances treat every `Sum` as its own distinguished, isolated possability - for data types with

  multiple consecutive sum values, like `These` or `Ordering`, it's suggested to write your own instance to pack

  each sum constructor as its own flag in the byte range `0 - 255`. `Sum`s are treated like `Either` - only `0` and

  `1` are used.