SIG practices for approver and maintainers

This pages includes guidelines and some common practices used by approvers and maintainers.

Onboarding

If a contributor steps up to take on a role with more responsibility towards the documentation (approver, maintainer) they will be onboarded by existing approvers and maintainers:

• They are added to the docs-approvers (or docs-maintainers) group.
• They are added to the #otel-comms and #otel-maintainers and private in-team slack channels.
• They are asked to enroll for the calendar invites for SIG Comms meeting and maintainers meeting.
• They are asked to verify that the current meeting time for SIG Comms works for them and if not to collaborate with existing approvers and maintainers to find a time that suits everyone.
• They are asked to review the different resources available for contributors:
  • Community Resources, especially the document around Community Membership and the social media guide.
  • Contributing Guidelines As part of this, they will review those documents and provide feedback for improving them via issues or pull requests.

Additional valuable resources to review are

• Hugo documentation
• Docsy documentation
• Marketing guidelines, including the Linux Foundation’s branding and trademark usage guidelines. Those are especially valuable when reviewing entries to the registry, integrations, vendors, adopters or distributions.

Collaboration

• Approvers and maintainers have different work schedules and circumstances. That’s why all communication is assumed to be asynchronous and they should not feel obligated to reply outside of their normal schedule.
• When an approver or maintainer won’t be available to contribute for an extended period of time (more than a few days or a week) or won’t be available in that period of time, they should communicate this using the #otel-comms channel and updating the GitHub status.
• Approver and maintainer adhere to the OTel Code of Conduct and the Community Values. They are friendly and helpful towards contributors. In the case of a conflict, misunderstanding or any other kind of situation that makes an approver/maintainer feel uncomfortable they can step back from a conversation, issue or PR and ask another approver/maintainer to step in.

Triage

Issues

• Incoming issues are triaged by @open-telemetry/docs-triagers team.
• As a first step, a triager will read through an issue title and description and apply the following labelling:
  • Mandatory: A sig:*, lang:* or docs:* to determine (co)ownership of the issue:
    • A sig:* label if the issue is related to content or a question that is co-owned by a SIG (e.g. a question around the Collector will be labelled sig:collector).
    • A lang:* label if the issue is related to content or a question that is related to a specific localization.
    • A docs:* label if the issue is related to content or a question that is solely owned by the docs team (SIG Comms):
      • docs
      • docs:admin
      • docs:accessibility
      • docs:analytics-and-seo
      • docs:IA
      • docs:blog
      • docs:cleanup/refactoring
      • docs:upstream, docs:upstream/docsy
      • docs:javascript
      • docs:mobile
      • docs:registry
      • docs:ux
  • Mandatory: A triage:* label:
    • triage:accepted, triage:accepted:needs-pr
    • triage:deciding, triage:deciding:blocked, triage:deciding:needs-info
    • triage:rejected, triage:rejected:duplicate, triage:rejected:invalid, triage:rejected:wontfix
  • Mandatory: Set the “type” of the issue as follows:
    • issue type bug for bugs
    • issue type enhancement for feature requests
    • label type:question for questions
    • label type:copyedit for copy edits
    • move an the issue to “discussions” if it seems to be a non workable open-ended conversation
  • Optional: An estimate label if applicable:
    • e0-minutes
    • …
    • e4-months
  • Optional (and only set by maintainers): A priority label:
    • p0-critical
    • p1-high
    • p2-medium
    • p3-low
  • Optional: One of the following special tags:
    • good first issue
    • help wanted
    • contribfest
    • maintainers only
    • forever
    • stale
• Automation will mark an issue in triage:deciding with triage:followup for re-triage after 14 days of inactivitiy on an issue. A triage:followup label should be removed within 7 days. Pinging the participants and removing the label is sufficient activity.

PRs

• PRs must have a linked issue labelled triage:accepted with the following exceptions:
  • Automatic PRs
  • hotfixes by maintainers/approvers
• Automation will ensure that PRs are labelled and assigned to the appropriate co-owning SIG or localization team.
• PRs should have the same co-ownership labels as issues
• If the PR is co-owned by a SIG, this group is responsible for doing a first review to ensure that the content is technically correct.
• If the PR is co-owned by a language team, this group is responsible for ensuring that the translation of the content is correct.
• The main responsibility of docs team is to ensure, that the PR is in line with the overall goals of the project, is put in the right place within the structure and follows the style and content guides of the project.
• PRs which are missing something to be merged, should be labeled accordingly:
  • missing:cla
  • missing:docs-approval
  • missing:sig-approval
  • blocked
• Automation will mark a PR as stale to request a re-review after 21 days of inactivity. A stale label should be removed within 14 days. Pinging the participants and removing the label is sufficient activity.
• PRs are never auto-closed.

Code reviews

General

• If the PR branch is out-of-date with the base branch, they do not need to be updated continuously: every update triggers all the PR CI checks to be run! It’s often enough to update them before merging.
• A PR by non-maintainers should never update git sub modules. This happens by accident from time to time. Let the PR author know that they should not worry about it, we will fix this before merging, but in the future they should make sure that they work from an up-to-date fork.
• If the contributor is having trouble signing the CLA or used the wrong email by mistake in one of their commits, ask them to fix the issue or rebase the pull request. Worst case scenario, close and re-open the PR to trigger a new CLA check.
• Words unknown to cspell should be added to the cspell ignore list per page by PR authors. Only approvers and maintainers will add commonly used terms to the global list.

Co-owned PRs

PRs with changes to documentation co-owned by a SIG (collector, demo, language-specific…) should aim for two approvals: one by a docs approver and one by a SIG approver:

• Doc approver label such PRs with sig:<name> and tag the SIG -approvers group on that PR.
• After a doc approver has reviewed and approved the PR, they can add the label sig-approval-missing. This signals to the SIG that they need to handle the PR.
• If no SIG approval is given within a certain grace period (two weeks in general, but may be less in urgent cases), docs maintainer may use their own judgement to merge that PR.

PRs from bots

PRs created by bots can be merged by the following practice:

• PRs that auto-update versions in the registry can be fixed, approved and merged immediately.
• PRs that auto-update the versions of SDKs, zero-code instrumentations or the collector can be approved and merged except the corresponding SIG signals that merging should be postponed.
• PRs that auto-update the version of any specification often require updates to scripts for the CI checks to pass. In that case @chalin will handle the PR. Otherwise those PRs can as well be approved and merged except the corresponding SIG signals that merging should be postponed.

Translation PRs

PRs with changes to translations should aim for two approvals: one by a docs approver and one by a translation approver. Similar practices apply as suggested for the co-owned PRs.

Merging PRs

The following workflow can be applied by maintainers to merge PRs:

• Make sure that a PR has all approvals and all CI checks pass.

• If the branch is out-of-date, rebase update it via the GitHub UI.

• The update will trigger all CI checks to run again, wait for them to pass or execute a script like the following to make it happen in the background:

  export PR=<ID OF THE PR>; gh pr checks ${PR} --watch && gh pr merge ${PR} --squash