Skip to content

JuliaDeps.md

Latest commit

 

History

History
166 lines (117 loc) · 4.2 KB

JuliaDeps.md

File metadata and controls

166 lines (117 loc) · 4.2 KB

Dependency Management in Julia

Julia has a built-in package manager called Pkg that provides sophisticated dependency management out of the box.

Core Concepts

1. Project.toml

The manifest file that specifies your project's direct dependencies:

name = "MyProject"
uuid = "12345678-1234-1234-1234-123456789abc"
version = "0.1.0"

[deps]
DataFrames = "a93c6f00-e57d-5684-b7b6-d8193f3e46c0"
Plots = "91a5bcdd-55d7-5caf-9e0b-520d859cae80"

[compat]
DataFrames = "1.3"
Plots = "1.25"
julia = "1.6"
  • Lists direct dependencies with their UUIDs
  • [compat] section specifies version requirements
  • Human-readable and version-controlled

2. Manifest.toml

The lock file that captures the entire dependency graph with exact versions:

[[DataFrames]]
deps = ["Compat", "DataAPI", "Tables"]
uuid = "a93c6f00-e57d-5684-b7b6-d8193f3e46c0"
version = "1.3.4"

[[Tables]]
deps = ["DataAPI"]
uuid = "bd369af6-aec1-5ad0-b16a-f7cc5008161c"
version = "1.7.0"
  • Auto-generated, should not be edited manually
  • Ensures reproducible environments
  • Includes all transitive dependencies with exact versions

Using Pkg

Basic Commands

REPL mode (press ] to enter):

add DataFrames          # Add a package
add DataFrames@1.3      # Add specific version
rm DataFrames           # Remove a package
update                  # Update all packages
status                  # List installed packages

Script mode:

using Pkg
Pkg.add("DataFrames")
Pkg.instantiate()       # Install from Manifest.toml

Environments

Julia supports isolated environments similar to Python's virtual environments:

Project-Level Environments

# Create a new project
Pkg.activate("path/to/project")
Pkg.add("DataFrames")

This creates Project.toml and Manifest.toml in the project directory.

Global Environment

Pkg.activate()  # Switch to default environment

The default environment is located at ~/.julia/environments/v1.x/.

Dependency Resolution

Julia's package manager uses a sophisticated resolver that:

  • Handles semantic versioning automatically
  • Resolves version conflicts across the dependency tree
  • Finds compatible versions of all packages
  • Updates Manifest.toml with resolved versions

Version Compatibility

The [compat] section uses standard version specifiers:

[compat]
DataFrames = "1.3"           # >=1.3.0, <2.0.0
Plots = "1.25.0 - 1.30.0"    # Range
CSV = "^0.10"                # Caret (>=0.10.0, <0.11.0)

Registries

Julia packages are published to registries (default: General Registry):

  • Packages are identified by name and UUID
  • UUIDs prevent naming conflicts
  • Multiple registries can be used simultaneously

Adding Packages from URLs

# From GitHub
Pkg.add(url="https://github.com/user/Package.jl")

# From local path
Pkg.develop(path="path/to/Package")

Development Workflow

Developing a Package

Pkg.develop("MyPackage")  # Clone package for local development
Pkg.free("MyPackage")     # Return to using registry version

Creating a New Package

Pkg.generate("MyPackage")

Creates a package structure with Project.toml, src/, and test/.

Key Features

  1. Automatic Dependency Resolution: No need for separate tools like pip-tools or Poetry
  2. Reproducible Environments: Manifest.toml ensures exact version reconstruction
  3. UUID-based Identification: Prevents naming conflicts
  4. Built-in: No external tools required
  5. Per-Project Isolation: Easy environment switching
  6. Precompilation: Packages are precompiled for faster loading

Best Practices

  • Commit both files: Version control both Project.toml and Manifest.toml
  • Use [compat]: Specify compatibility bounds to avoid breaking changes
  • Run Pkg.instantiate(): On new machines to recreate exact environment
  • Separate environments: Use project-specific environments, not the global one
  • Test compatibility: Run Pkg.test() to verify package functionality

Key Difference from Python: Julia's dependency management is fully integrated into the language from the start, providing a unified experience without needing third-party tools like Poetry, Pipenv, or conda.