AWS JLL Auto-Update

This repository contains GitHub Actions workflows and utilities for automatically updating AWS CRT library JLL packages in the Julia ecosystem.

Strategy

Problem

Previously, when updating a library (e.g., aws_c_cal from version 0.9.0 to 0.9.1), the workflow would:

1. Update the library’s build_tarballs.jl
2. Update ALL other build_tarballs.jl files that depend on this library to use the new version

This caused a cascade of unnecessary rebuilds in Yggdrasil, since every change to a build_tarballs.jl file requires incrementing the version and rebuilding.

Solution

The new strategy focuses on updating dependencies within the library being updated, rather than updating dependents:

1. When updating a library: Update its own dependencies to use the latest available JLL versions
2. Maintain a JLL versions cache: Keep a jll-versions.json file with the latest known versions of all AWS-related JLL packages
3. Dynamic dependency resolution: When updating a library’s build_tarballs.jl, automatically fetch and update its dependencies to the latest versions

Components

1. JLL Versions Cache (jll-versions.json)

A JSON file containing the latest known versions of all AWS-related JLL packages:

{
  "last_updated": "2025-05-26T00:00:00Z",
  "versions": {
    "aws_c_auth_jll": "0.8.1",
    "aws_c_cal_jll": "0.9.1",
    "aws_c_common_jll": "0.12.3",
    ...
  }
}

This cache is updated periodically by the GitHub Actions workflow.

2. JLL Version Manager (jll-version-manager.jl)

A Julia utility script that:

• Queries Julia registries and GitHub for latest JLL package versions
• Updates the versions cache
• Parses build_tarballs.jl files and updates dependency versions
• Provides a CLI interface for version management

Usage:

julia jll-version-manager.jl update-cache                    # Update the versions cache
julia jll-version-manager.jl update-deps build_tarballs.jl   # Update dependencies in a file
julia jll-version-manager.jl get-version aws_c_common_jll    # Get latest version of a package

3. GitHub Actions Workflow

The workflow runs hourly and:

1. Updates the JLL cache with latest versions from Julia registries
2. Checks each AWS library for new releases
3. When a new release is found:
   • Updates the library version and SHA in its build_tarballs.jl
   • Updates all dependencies in that file to use the latest available versions
   • Creates a PR with the changes

Benefits

1. Fewer unnecessary rebuilds: Only the updated library gets rebuilt, not all its dependents
2. Always up-to-date dependencies: When a library is updated, it automatically gets the latest dependency versions
3. Cached version lookups: Fast dependency resolution using the maintained cache
4. Fallback to live queries: If cache is stale, automatically queries for latest versions
5. Comprehensive dependency management: Handles both Dependency and BuildDependency declarations

Example

When aws_c_common is updated from 0.12.2 to 0.12.3:

Old approach:

• Update aws_c_common/build_tarballs.jl to version 0.12.3
• Find and update ~10 other libraries that depend on aws_c_common_jll
• All ~10 libraries get rebuilt unnecessarily

New approach:

• Update aws_c_common/build_tarballs.jl to version 0.12.3
• Update any dependencies within aws_c_common to their latest versions
• Only aws_c_common gets rebuilt
• Other libraries will pick up the new version naturally when they’re next updated

This results in significantly fewer spurious rebuilds while keeping the ecosystem current.