Optimize factory usage for build action entity (!229044) · Merge requests · GitLab.org / GitLab

What does this MR do and why?

This MR replaces create with build_stubbed in spec/serializers/build_action_entity_spec.rb to optimize the factory usage as recommended in our handbook. This spec is part of the list .rubocop_todo/rspec/factory_bot/avoid_create.yml. It was verified that database persistence is not needed using the factory doctor command FDOC=1 bin/rspec spec/serializers/build_action_entity_spec.rb.

Factory Usage Optimization Results

Before optimization:

Factory time: 3.492s (16.32% of total time)

After optimization:

Factory time: 0.081s (0.43% of total time)

🚀 Improvement:

Factory time reduced by: 3.411s
Factory time percentage reduced by: 15.89%
Overall factory usage improvement: 97.7%

Factory Doctor Output

[TEST PROF INFO] FactoryDoctor enabled (event: "sql.active_record", threshold: 0.01)
Run options: include {:focus=>true}

All examples were filtered out; ignoring {:focus=>true}
==> Using precompiled Gitaly binaries from cache

Test environment set up in 1.3719100000162143 seconds
......[TEST PROF INFO] FactoryDoctor report

Total (potentially) bad examples: 3
Total wasted time: 00:02.523

BuildActionEntity (./spec/serializers/build_action_entity_spec.rb:5) (21 records created, 00:02.523)
  contains original job name (./spec/serializers/build_action_entity_spec.rb:21) – 7 records created, 00:01.812
  contains path to the action play (./spec/serializers/build_action_entity_spec.rb:25) – 7 records created, 00:00.356
  contains whether it is playable (./spec/serializers/build_action_entity_spec.rb:29) – 7 records created, 00:00.354




Finished in 17.42 seconds (files took 15.21 seconds to load)
6 examples, 0 failures

[TEST PROF INFO] Time spent in factories: 00:03.638 (16.55% of total time)

Factory time before the update

[TEST PROF INFO] Time spent in factories: 00:03.492 (16.32% of total time)

Factory time after update

[TEST PROF INFO] Time spent in factories: 00:00.081 (0.43% of total time)

AI Agent Output

**Changes made:**

- **Line 6:** `create(:ci_build, ...)` → `build_stubbed(:ci_build, ...)` — eliminates ~6 associated DB records (pipeline, project, namespace, author, etc.) that `ci_build` factory creates transitively.
- **Line 8:** `create(:user)` → `build_stubbed(:user)` — eliminates the user record. The user is only accessed via `request.current_user`, which is already a stubbed double.

**Why this works:**
- `build_stubbed` provides a fake integer ID, so `jobs/#{job.id}/play` path assertions in line 25 still work.
- `job.playable?` (line 29) is computed from in-memory attributes (status, when, etc.) — no DB query needed.
- `name` attribute (line 21) is purely in-memory.
- The `when job is scheduled` context overrides `let(:job)` with its own `create(:ci_build, :scheduled)`, so those tests are unaffected.

**Performance improvement:** ~7 DB records eliminated per example × 3 examples = ~21 fewer DB insertions, saving roughly 2.5s based on the FactoryDoctor report.

This MR also adds the missing feature category in this spec.

References

Relates to: #378910.

MR acceptance checklist

Evaluate this MR against the MR acceptance checklist. It helps you analyze changes to reduce risks in quality, performance, reliability, security, and maintainability.

Optimize factory usage for build action entity