Not-Yet-Awesome Rust

An (unofficial) sub-community of the Rust programming language that wants to close gaps you feel in Rust's ecosystem.

There are two major resources provided in this community at time of writing: the NYAR ecosystem list and issues in the NYAR tracker.

The List

This README is a list with a twofold purpose:

  • Enumerate specific use cases and their problems domains that do not yet have a robust ecosystem in Rust.
    • The definition of "specific" and "robust" for this list is yet to be determined!
  • Encourage the Rust community to approach gaps in the Rust ecosystem by providing this list as a point of collaboration!

You can jump right into editing this file here. See the contributing guide for information on what you can do to help or if you have questions about this list!

Issues

NYAR's issue tracker can help you get ecosystem gaps filled when you participate in:

  • Voting on your favorite issues to let people know you'd use something, optionally commenting and explaining your use case.
  • Placing bounties on issues you're particularly passionate about, since issues are where BountySource integrates! We use the bounty tag to indicate which issues have bounties associated.

Table of contents

Documentation

Stack Overflow

  • There are many older Rust questions on Stack Overflow that wouldn't work with today's Rust because of syntax that has changed since the release of 1.0, or that may have better solutions because of other Rust ecosystem developments.
    • See #4 for an SO query and a list of these known issues!

Character encodings

Computer Vision

  • Some work has been done to create OpenCV bindings in cv-rs as well as automatically generating bindings like opencv-rust. Neither are very complete.
  • Piston has imageproc based on image, but in their words, "This is very much a work in progress".

Data processing

  • DDS library wiki
  • HDF5 (see also Wikipedia and this Reddit post) A stable version of hdf5 crate has been released and is now available on crates.io.
  • A good stream processing pipeline with back pressure doesn't yet exist for an asynchronous data processing pipeline
    • RxRust is an older attempt to implement this according to the reactive streams model -- it currently seems closest to this use case.
    • tokio and futures may be interesting components to use when building this.
    • New features soon to come in Rust like impl Trait will probably make developing something like this easier to develop and use.
  • Bindings for pandoc A pandoc crate now exists with CLI bindings!
  • Bindings for git-annex

Data structures

  • A concurrent std::collections::HashMap-like structure has not been fully developed yet.

    • concurrent-hashmap is still missing methods like iter_mut, entry, drain, and clear from the original HashMap interface.
    • evmap is a different design around eventual consistency, and so departs from the normal HashMap interface.

    dashmap is intended to fit this use case exactly.

Dynamic programming analysis/instrumentation

  • Binary instrumentation tools like Intel Pin or DynamoRIO are useful for embedded applications. An ecosystem around them (which doesn't exist yet!) could include:
    • Bindings to the C APIs associated with existing applications (for instance, Intel Pin and DynamioRIO)
    • Libraries or applications implementing the features of the aforementioned in Rust

Embedded development

  • MINIX support is nonexistent! still a WIP.

Game development

  • Integration with existing mature game engines is poor at time of writing.
    • Unreal Engine has had some prior work accessible by making an Unreal account, but nothing has been made freely public at time of writing.
    • Unity3D currently has no published crates for Rust interop.
    • Godot Engine allows Rust interop via godot-rust. Authors' note on stability:

      The bindings cover most of the exposed API of Godot 3.2, and are being used on a number of projects in development, but we still expect non-trivial breaking changes in the API in the coming releases.

  • No mature game engine written in pure Rust exists yet.

Geometry

  • PCL equivalent - point clouds, essential 3D geometry functions
  • Voxel library, operations and representation of voxel data. There is a piston crate for rendering voxels but this isn't suitable for working on domains like medical data, more geared towards game development. To enable processing in scientific domains like medicine there needs to be processing functions such as being able to: convert to triangular mesh, thresholding, and morphology. This is because things like MRI data is expressed as voxels, thresholding can separate grey and white matter in the brain, morphology identifies shapes and structures inside the body etc.

Geospatial Information Systems

  • OGC standards - multiple crates for standards for encoding, sharing or manipulating geospatial data link. There's already a crate for GeoJSON but none of the others appear to have crates.
  • More complete GDAL wrapper (or pure rust alternative). rust-gdal is an incomplete wrapper so needs work.

Machine Learning

  • Machine learning toolkit like scikit-learn in Python (both rust-learn and rusty-machine are insufficient). Rust-learn only supports classification, rusty-machines misses support for sparse data and serialization. Both of them miss quite common unsupervised techniques (like PCA).
  • Deep learning toolkit with GPU support a good flexibility (think Tensorflow or Chainer in Python). Most of the current libraries are either simplistic (you cannot do seq2seq network in them for example), or miss GPU support.

Mathematics

  • Sparse matrix libraries (SPRS library needs some love, the basics are there but advanced linear algebra is missing)
  • Designing low latency DSP algorithms suitable for embedded use (common filters, analysis functions)
  • Library for nonlinear dynamical or chaotic systems (solvers, numeric methods etc.)
  • A pure Rust libm implementation. These are required to get math functions on #![no_std] platforms
    • The m crate has made some headway here, though it hasn't been maintained recently.
    • The libm crate is a port of libm functionality from musl and newlib to Rust, with the aim of eventually being merged into core.

Native desktop application integrations

Microsoft Office

  • An interactive Visual Basic uses for scripting by using the COM interface, which I believe winapi supports.

Networking and Protocols

  • AMQP 1.0 - While there are several libraries for AMQP 0.9.1, the AMQP 1.0 spec represents a significant departure from the previous version. fe2o3-amqp implements the AMQP 1.0 protocol for both client side and server/listener side.

Parsers/Emitters

  • Ability to parse Registry.pol files from Windows machines -- implemented by this person!
  • Common office document formats are yet to have more mature solutions:
    • Excel/Calc spreadsheet deserialization seems available with calamine, but no serialization libraries seem available for them, let alone for the entire XML formats that the Office/OpenOffice suites themselves support.
    • Otherwise, OpenOffice and Microsoft Office
  • There is currently no library to convert between different office document formats.
  • The beancount data format has no parser or emitter libraries yet.
  • There is no pure-Rust solution for QR decoding. The only other crate that handles QR decoding is the quirc crate, which uses C bindings -- implemented by @Wanzenbug and by @piderman, the latter of which was announced on Reddit. See the PR.
  • The RData file format does not have a parser or emitter yet.
    • Currently, this formatted is implemented for the R language in the serialize.c module.
  • There are no SGML parsers or emitters on crates.io at all.
    • This is mostly a legacy markup language used for legacy type setting and supported in applications like APP (AKA 3B2).

Personal information management

  • Contacts via vCard have been implemented using vobject, but no "high-level interface" exists yet that uses it or an alternative.
  • iCalendar parsing has been implemented via several crates (i.e., vobject), but a higher-level API is missing.

Rust Toolchain

  • A stable Rust interpreter does not yet exist, which would made code exploration easier.
    • miri seems to be a step in the right direction -- it just needs some love!
  • No debugging experience offers integration with rustdoc, which would
  • A rustdoc backend for generating zeal/ dash docsets is missing.

User Interfaces

  • A mature framework for Windows native UI has yet to be established.
    • native-windows-gui at one point claimed to be approaching feature-completion, but is now unmaintained.
  • Abstractions over native UIs for each platform have yet to be available in pure Rust.
  • A good reactive UI library complete with event-driven state management comparable to Redux does not yet exist.
    • carboxyl looks like it may be a good fundamental building block for this.

Web bindings

Google API

  • Generation of Google bindings using googleapis and gRPC would be more performant than using JSON web requests to the Google API, as with google-apis-rs service.
  • There is room for more idiomatic APIs for Google in general. google-apis-rs uses the Google Discovery service to expose the vast majority of Google Services, but they can be difficult to grok for beginners or someone unfamiliar with Google APIs in general.

XML

There is yet to be a library that handles all of these:

  • General purpose DOM tree
  • Proper encoding handling
  • DTD handling
    • including the disabling of features with security implications like entity expansion bombs
  • XML Schema validation
  • XPath
  • XQuery
  • XSLT 1.0/2.0/3.0
  • Option to preserve input style
  • Is fast

For more feature and performance comparisons for existing Rust XML crates, see choose-your-xml-rs.