Purescript Protobuf Versions Save

• Development: purs v0.15.7, node-streams-aff v5.0.0, assorted improvements.

Full Changelog: https://github.com/xc-jp/purescript-protobuf/compare/v4.2.0...v4.3.0

• Add doc comments from the .proto messages and enums to the generated .purs code.
• Improved dev environment for purescript-language-server and ide-purescript with PURS_IDE_SOURCES.
• Use spago2nix_nativeBuildInputs.

Full Changelog: https://github.com/xc-jp/purescript-protobuf/compare/v4.1.0...v4.2.0

Full Changelog: https://github.com/xc-jp/purescript-protobuf/compare/v4.0.0...v4.1.0

• Nix flake. With one-step spago2nix.
• Upgrade dependencies.
• Delete mkUint8Array. Upstreamed this to purescript-arraybuffer.
• Delete manyArray, parseMaybe, parseEither, parseExceptT. Upstreamed these to purescript-parsing.
• Refactor manyLength. Get rid of the FFI, and a slight speedup in the benchmarks.
• Delete Runtime.positionZero.

Upgraded to PureScript 0.14.3.

For byte fields we’re now using the DataBuff type from arraybuffer-builder v2.1.0 which allows us to avoid array copies, and so may speed up encoding and decoding in some cases.

Created a Protobuf.Prelude module to eliminate many import warnings in the generated code.

Breaking changes

The Bytes field type was formerly a wrapper for ArrayBuffer, and now it is a wrapper for DataBuff, which may be either ArrayBuffer or DataView. This changes the API and will require some changes in dependent code.

Bugfixes

Fixed a bug which sometimes incorrectly errored with “index out of bounds” when decoding packed repeated fields of double, float, fixed32, or sfixed32.

Fixed a bug which would cause a compilation failure in generated code for enums with negative values.