Skip to content

Latest commit

 

History

History
270 lines (182 loc) · 12.3 KB

index.md

File metadata and controls

270 lines (182 loc) · 12.3 KB
Raw

Introduction

We welcome and encourage contributions of all kinds, such as:

  1. Tickets with issue reports of feature requests
  2. Documentation improvements
  3. Code (PR or PR Review)

In addition to submitting new PRs, we have a healthy tradition of community members helping review each other's PRs. Doing so is a great way to help the community as well as get more familiar with Rust and the relevant codebases.

You can find a curated good-first-issue list to help you get started.

Pull Requests

We welcome pull requests (PRs) from anyone from the community.

DataFusion is a very active fast-moving project and we try to review and merge PRs quickly to keep the review backlog down and the pace up. After review and approval, one of the many people with commit access will merge your PR.

Review bandwidth is currently our most limited resource, and we highly encourage reviews by the broader community. If you are waiting for your PR to be reviewed, consider helping review other PRs that are waiting. Such review both helps the reviewer to learn the codebase and become more expert, as well as helps identify issues in the PR (such as lack of test coverage), that can be addressed and make future reviews faster and more efficient.

Merging PRs

Since we are a worldwide community, we have contributors in many timezones who review and comment. To ensure anyone who wishes has an opportunity to review a PR, our committers try to ensure that at least 24 hours passes between when a "major" PR is approved and when it is merged.

A "major" PR means there is a substantial change in design or a change in the API. Committers apply their best judgment to determine what constitutes a substantial change. A "minor" PR might be merged without a 24 hour delay, again subject to the judgment of the committer. Examples of potential "minor" PRs are:

  1. Documentation improvements/additions
  2. Small bug fixes
  3. Non-controversial build-related changes (clippy, version upgrades etc.)
  4. Smaller non-controversial feature additions

Developer's guide

This section describes how you can get started at developing DataFusion.

Windows setup

wget https://az792536.vo.msecnd.net/vms/VMBuild_20190311/VirtualBox/MSEdge/MSEdge.Win10.VirtualBox.zip
choco install -y git rustup.install visualcpp-build-tools
git-bash.exe
cargo build

Protoc Installation

Compiling DataFusion from sources requires an installed version of the protobuf compiler, protoc.

On most platforms this can be installed from your system's package manager

$ apt install -y protobuf-compiler
$ dnf install -y protobuf-compiler
$ pacman -S protobuf
$ brew install protobuf

You will want to verify the version installed is 3.12 or greater, which introduced support for explicit field presence. Older versions may fail to compile.

$ protoc --version
libprotoc 3.12.4

Alternatively a binary release can be downloaded from the Release Page or built from source.

Bootstrap environment

DataFusion is written in Rust and it uses a standard rust toolkit:

  • cargo build
  • cargo fmt to format the code
  • cargo test to test
  • etc.

Testing setup:

  • rustup update stable DataFusion uses the latest stable release of rust
  • git submodule init
  • git submodule update

Formatting instructions:

or run them all at once:

Test Organization

DataFusion has several levels of tests in its Test Pyramid and tries to follow Testing Organization in the The Book.

This section highlights the most important test modules that exist

Unit tests

Tests for the code in an individual module are defined in the same source file with a test module, following Rust convention

Rust Integration Tests

There are several tests of the public interface of the DataFusion library in the tests directory.

You can run these tests individually using a command such as

cargo test -p datafusion --tests sql_integration

One very important test is the sql_integration test which validates DataFusion's ability to run a large assortment of SQL queries against an assortment of data setups.

sqllogictests Tests

The sqllogictests also validate DataFusion SQL against an assortment of data setups.

Data Driven tests have many benefits including being easier to write and maintain. We are in the process of migrating sql_integration tests and encourage you to add new tests using sqllogictests if possible.

Benchmarks

Criterion Benchmarks

Criterion is a statistics-driven micro-benchmarking framework used by DataFusion for evaluating the performance of specific code-paths. In particular, the criterion benchmarks help to both guide optimisation efforts, and prevent performance regressions within DataFusion.

Criterion integrates with Cargo's built-in benchmark support and a given benchmark can be run with

cargo bench --bench BENCHMARK_NAME

A full list of benchmarks can be found here.

cargo-criterion may also be used for more advanced reporting.

Parquet SQL Benchmarks

The parquet SQL benchmarks can be run with

 cargo bench --bench parquet_query_sql

These randomly generate a parquet file, and then benchmark queries sourced from parquet_query_sql.sql against it. This can therefore be a quick way to add coverage of particular query and/or data paths.

If the environment variable PARQUET_FILE is set, the benchmark will run queries against this file instead of a randomly generated one. This can be useful for performing multiple runs, potentially with different code, against the same source data, or for testing against a custom dataset.

The benchmark will automatically remove any generated parquet file on exit, however, if interrupted (e.g. by CTRL+C) it will not. This can be useful for analysing the particular file after the fact, or preserving it to use with PARQUET_FILE in subsequent runs.

Upstream Benchmark Suites

Instructions and tooling for running upstream benchmark suites against DataFusion can be found in benchmarks.

These are valuable for comparative evaluation against alternative Arrow implementations and query engines.

How to add a new scalar function

Below is a checklist of what you need to do to add a new scalar function to DataFusion:

  • Add the actual implementation of the function:
    • here for string functions
    • here for math functions
    • here for datetime functions
    • create a new module here for other functions
  • In physical-expr/src, add:
    • a new variant to BuiltinScalarFunction
    • a new entry to FromStr with the name of the function as called by SQL
    • a new line in return_type with the expected return type of the function, given an incoming type
    • a new line in signature with the signature of the function (number and types of its arguments)
    • a new line in create_physical_expr/create_physical_fun mapping the built-in to the implementation
    • tests to the function.
  • In core/tests/sql, add a new test where the function is called through SQL against well known data and returns the expected result.
  • In expr/src/expr_fn.rs, add:
    • a new entry of the unary_scalar_expr! macro for the new function.

How to add a new aggregate function

Below is a checklist of what you need to do to add a new aggregate function to DataFusion:

  • Add the actual implementation of an Accumulator and AggregateExpr:
    • here for string functions
    • here for math functions
    • here for datetime functions
    • create a new module here for other functions
  • In datafusion/expr/src, add:
    • a new variant to AggregateFunction
    • a new entry to FromStr with the name of the function as called by SQL
    • a new line in return_type with the expected return type of the function, given an incoming type
    • a new line in signature with the signature of the function (number and types of its arguments)
    • a new line in create_aggregate_expr mapping the built-in to the implementation
    • tests to the function.
  • In tests/sql, add a new test where the function is called through SQL against well known data and returns the expected result.

How to display plans graphically

The query plans represented by LogicalPlan nodes can be graphically rendered using Graphviz.

To do so, save the output of the display_graphviz function to a file.:

// Create plan somehow...
let mut output = File::create("/tmp/plan.dot")?;
write!(output, "{}", plan.display_graphviz());

Then, use the dot command line tool to render it into a file that can be displayed. For example, the following command creates a /tmp/plan.pdf file:

dot -Tpdf < /tmp/plan.dot > /tmp/plan.pdf

Specifications

We formalize DataFusion semantics and behaviors through specification documents. These specifications are useful to be used as references to help resolve ambiguities during development or code reviews.

You are also welcome to propose changes to existing specifications or create new specifications as you see fit.

Here is the list current active specifications:

All specifications are stored in the docs/source/specification folder.

How to format .md document

We are using prettier to format .md files.

You can either use npm i -g prettier to install it globally or use npx to run it as a standalone binary. Using npx required a working node environment. Upgrading to the latest prettier is recommended (by adding --upgrade to the npm command).

$ prettier --version
2.3.0

After you've confirmed your prettier version, you can format all the .md files:

prettier -w {datafusion,datafusion-cli,datafusion-examples,dev,docs}/**/*.md