Hi All,

To follow up on the discussion in the Architecture Committee meeting yesterday, we're trying to make better use of GitHub labels.

# TL;DR

Take a look at https://github.com/kata-containers/tests/pull/1466.

(The reason the PR is in the "tests" repo by the way is that this repo is always cloned when running CI tests for any _other_ repo. That allows us to add additional CI tests on the labels database and labels on issues/PRs. See below for further info if you're interested).

__________

# The problem

The Kata GitHub-hosted repositories currently have a variety of labels that can be added to GitHub issues and PRs. However, the current labels suffer from a number of problems:

- Some labels refer to the same concept.
- Only a few of the labels have descriptions.
- The meaning of some of the labels is unclear.
- The overall set of labels is not adequate to make intelligent queries on the pool of open issues.
- Labels are not applied consistently to issues.

# Plan

Establish a small set of labels that are succinct, intuitive, documented and which allow us to capture the many dimensions of an issue/PR.


# Approach


- Create a YAML database of labels that will be used consistently across all Kata repos.

- Allow individual repos to supplement the base list of labels by providing their own YAML snippet.

- Add tooling to check and view the YAML database files.

- Get agreement on the labels to use.

- Apply the labels defined in the YAML labels database to all Kata repos.

- Add CI checks to validate the labels database.

- Review the backlog of issues + PRs across all repos and manually apply labels.

- Start to improve our process by:

- Creating GitHub issue templates to auto-label issues.

- Make the bug triage slightly more formal by having the team review each new issue and apply a few labels (things like priority).

- Consider using Planning Poker to "size" issues and break them down into manageable chunks.

- Auto-add labels to PRs showing their relative sizes (by SLOC).

- Start using GitHub projects to manage upcoming releases.

  - ...


The basic mantra is "automate and organise" - we don't want to impose heavy process, but we do want to make better use of our resources. By applying a few well-chosen labels (automatically where possible), we can start to get a clearer picture of the backlog, which will hopefully allow us to clear it down more quickly.


Labels will allow us to categorise issues and PRs quickly rather than having to trawl through screens of log output, etc. Also, labels will allow teams to focus on particular groups of issues and PRs rather than having to fall back on the dreaded "github ping" to get attention on particular areas.


Please comment on https://github.com/kata-containers/tests/pull/1466 so we can start moving this forward.


Cheers,

James
---
Open Source Technology Center
Intel Corporation (UK) Ltd. - Co. Reg. #1134945 - Pipers Way, Swindon SN3 1RJ.