Proposal: support custom index repositories

[ahmetb]

Krew should support customizable index repositories.

This would allow users to distribute plugins without requiring them to make a pull request to the centralized index repo like "krew-index".

Similarly, this would allow vendors to distribute their plugins that are too specific or one-off to be in the centralized index repo.

Work here is probably:

• Determine/propose a user experience journey
• Propose a design
• Implement CLI commands to tap/untap into other index repos
• Refactor update/upgrade paths to keep taps in mind
• Write user/developer documentation on working with custom repos.

@lbb has showed interest to work on this.

[ahmetb]

There's a design at https://docs.google.com/document/d/1fG3ST0QbadQGcxNiFAZ9GhX_RZ8pm3piT99UK3FbT88/edit?ts=5b75a2a6 if anyone wants to review.

[ahmetb]

/cc @juanvallejo @soltysh

[melaniecebula]

@ahmetb @juanvallejo @soltysh any update on this issue? any way to help? I'm excited to try this out for Airbnb but support for custom index repositories is likely a requirement for us.

[juanvallejo]

@melaniecebula I've started a draft here to begin gathering thoughts on how to begin implementing this. Feel free to contribute or leave feedback on there

[ahmetb]

@juanvallejo the link is not available for public access

[juanvallejo]

@ahmetb Here's a clone of that document with public access: https://docs.google.com/document/d/1ofz884w9pQo4d1aolgPdYdLWSi1H5mtisoKAgmTW2kg/edit?usp=sharing (not sure I'm able to make original doc publicly accessible with that account)

[nulltrope]

hey, wondering what the next steps are for pushing this forward?

we've been writing many kubectl plugins internally which are specific to our stack, and being able to distribute them with krew would be a huge improvement over the manual process we use now.

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

[ahmetb]

/remove-lifecycle stale
/priority important-soon
/priority backlog

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

[corneliusweig]

/remove-lifecycle stale

[ahmetb]

/lifecycle frozen

[ahmetb]

We are moving ahead with the concrete proposal at #483.