alexellis / arkade-get

Get all the CLIs you need from arkade for a GitHub Action

I create independent tools for developers

Hear a bit about my background and why I'm on this journey:

• Video: Hear my OpenFaaS story over 60 months from this Gophercon keynote
• Podcast: Independent Open Source, with Alex Ellis - Kubernetes Podcast by Google

Follow my journey and support my Open Source work

As a GitHub Sponsor, you'll hear from me every week by email - covering my journey building independent software and open source projects. I share about new features, updates, events, tutorials and often share what I've learned from reading and executing every day.

• ⭐️ Sponsor me on GitHub
• 🤓 Send me a book for inspiration from my Amazon wishlist

My books

I've been self-publishing books for a year.

Fast track your Go journey

My best-selling book!

• Learn unit testing and how to isolate dependencies
• Make lovely CLIs for internal tools and open source projects
• Monitor your services with Prometheus metrics
• Ship your code in a container with Docker
• Learn patterns and pitfalls for using Goroutines

Fast track your Go

📚 My first eBook - Serverless for Everyone Else

Learn the use-cases for serverless functions through practical exercises. You'll deploy your own self-hosted functions, write code in Node.js and using examples will start to call HTTP APIs, add npm modules and understand how to monitor your code with Grafana.

Get the eBook on its own, or upgrade to the video workshop for my expert overview.

• Learn how to apply functions for your own workflow

Need to contact me?

• 📫 Need to email me?
• 💬 Want to book a 1:1 with me?

arkade-get's Issues

`jq` repo change not fixed as this is dependent on the update-tools action

@alexellis is it possible to trigger the update-tools action manually so it pulls in the latest arkade with the JQ fix?

Tools can be cached in _work/_tool

Today I learned from @rodrik, that if this setup-arkade action places its tools within the _work/_tool directory instead of $HOME/.arkade/bin/ then these will be cached by the runner instead of being downloaded one each run.

Most pre-existing setup-xyz actions that use the GitHub JavaScript library already do this.

Flags:
      --arch string      CPU architecture for the tool (default "x86_64")
  -h, --help             help for get
      --os string        Operating system for the tool (default "Linux")
  -o, --output string    Output format of the list of tools (table/markdown/list)
      --progress         Display a progress bar (default true)
      --quiet            Suppress most additional output
      --stash            When set to true, stash binary in HOME/.arkade/bin/, otherwise store in /tmp/ (default true)
  -v, --version string   Download a specific version

We may need an additional --path folder, or to download first to the usual location, then to move the files.

I'm open to a PR to arkade to add the flag for a custom path or ENV-var to redirect where binaries are downloaded from arkade get

Tools won't get cleanup or get reused

The action fills up our runner when run couple of time. It creates always new /tmp/arkade-* and then the storage fills up.
What can we do against that? Maybe something with #35?

Download arkade in this action so that setup-arkade isn't required

Before:

    - uses: alexellis/setup-arkade@v2
    - uses: alexellis/arkade-get@master
      with:
        kubectl: latest
        faas-cli: 0.14.10
    - name: check for faas-cli
      run: |
        faas-cli version

After:

    - uses: alexellis/arkade-get@master
      with:
        kubectl: latest
        faas-cli: 0.14.10
    - name: check for faas-cli
      run: |
        faas-cli version

Bear in mind that the bash script installer for arkade is OS and arch aware, and this needs to be kept for the arkade-get action too. At least for Linux aarch64 and amd64.

https://github.com/alexellis/arkade/blob/master/get.sh