Healthy backlog initiative Pipeline Authoring

Pipeline Authoring Backlog Refinement - completed

Overview

Work Completed

✅ Feature Issues (type::feature)

• Reviewed: All feature issues in the Pipeline Authoring backlog ~1700 issues
• Labeled:
  • 143 backlogprospective
  • 1376 backlogno-commitment
• Closed: ~250
• Strategic decisions:
  • Catalog related items - backlogprospective
  • Inputs enhancements - backlogprospective
  • Variable expansion - backlogno-commitment

💡 Key insights

• Our CI Inputs investment delivered several requested features without requiring per-key variable expansion implementation. Users now get automatic variable resolution across all YAML keys through inputs, which are a secure and preferred method to pass parameters to a pipeline Here are a couple of examples
  • Backend: Use CI/CD UI variables to define paral... (gitlab-org/gitlab#11549 - closed) 225 👍
  • CI variable expansion in `cache:key:files` (gitlab-org/gitlab#118466 - closed) 65 👍
  • Gitlab CI variable expansion in needs (gitlab-org/gitlab#255311 - closed) 45 👍

✅ Maintenance Issues (type::maintenance)

• Reviewed: All maintenance issues in the Pipeline Authoring backlog ~350 issues
• Labeled:
  • ~100 backlogprospective
  • ~200 backlogno-commitment
• Closed: ~50
• Strategic decisions:
  • Pipeline creation performance - backlogprospective
  • Database growth - backlogprospective
  • Inputs/Components/Catalog related - backlogprospective
  • Unclear impact/benefit - backlogno-commitment
  • "Nice to have" items - backlogno-commitment

💡 Key insights

• Pipeline creation performance and database growth are things we will have to commit to, the former because we're already facing internal pressure (from a DevEx perspective), the latter due to physical realities.
• Outside of these two key aspects, I focussed on issues related to our investment areas as outlined by Dov above where a tangible impact on maintainability was presented in the issue.

✅ Bug Issues (type::bug)

• Reviewed: All bug issues in the Pipeline Authoring backlog ~250 issues
• Labeled:
  • ~60 backlogprospective
  • ~90 backlogno-commitment
• Closed: ~100
• Strategic decisions:
  • Inputs/Components/Catalog related - backlogprospective
  • Pipeline creation - backlogprospective
  • !reference tag - backlogprospective (gitlab-org&18767)
  • Accessibility - backlogprospective
  • Pipeline variables / job variables - backlogno-commitment
  • Avoidable problems (workaround available, or very specific requirements needed to trigger) - backlogno-commitment

💡 Key insights

• Pipeline creation (not performance as above) has pitfalls we need to (and already plan to via investing in pipeline inputs) address.
• Catalog/Components/Inputs is our key investment area, so we need to polish it. The !reference tag is symptomatic for the inconsistent and confusing state of our YAML keywords, and we need to start getting this into order – people will keep opening bug issues for things we consider feature enhancements due to our MVP-approach and frequent focus shifts, and LLMs will stumble over these inconsistencies and unexpected behaviors as well.
• Accessibility has a bunch of quick wins for us once we're back at full frontend capacity.
• Variables remain a complicated topic – we can divest from pipeline and job variables, but general variable bugs may remain that we need to address if they can't be resolved via inputs.