fix: the HTML report survives Rails' fork-parallel test runs (#258)

[pftg]

Closes #258.

The bug, measured

Under Rails' default parallelize(workers: N) the HTML report is never written and the
[snap_diff] N verified, N changed, N new line never prints. Minitest skips after_run in
forked children (allow_fork = false at minitest.rb:64, guard at :79), so the workers
holding the results never finalize, and the parent that does finalize recorded nothing.

Pass/fail is unaffected — failures marshal back over DRb — and every image artifact still
lands on disk. Only the report and the summary are lost, which is what makes it quiet:
ActiveSupport.test_parallelization_threshold defaults to 50, so a suite silently loses its
report the day someone adds the 51st test.

The fix

Rails runs run_cleanup_hooks inside the worker just before it exits
(parallelization/worker.rb:31). Each worker dumps its records to a JSON fragment there; the
parent merges every fragment in the Minitest.after_run it does reach, then writes one
report at the documented path. No application-side configuration.

• The summary counts are the merged totals. verified and changed come from the merged
  reporter state; new from the union of the workers' missing-baseline sets, so a name missing
  in two workers is counted once — same as serial.
• Fragments live under the system temp dir, keyed by the parent's pid, removed after the
  merge. Nothing is written inside the repository. Each is written to a .tmp name and renamed
  into place, so a worker killed mid-write leaves nothing the merge will read.
• Not the per-worker save_path route. save_path is also where baselines are read from
  (config.screenshot_area), so a per-worker value points every comparison at an empty
  baseline directory and records everything as new.
• Feature-detected. Registered only when ActiveSupport::Testing::Parallelization is
  defined, and retried from ActiveSupport.on_load(:active_support_test_case) because
  Bundler.require loads this gem before ActiveSupport::TestCase exists — the zero-require
  Rails path. A non-Rails Capybara suite is untouched.

Evidence

An 8-classes × 8-tests harness with committed baselines (8 of them deliberately stale), run
under parallelize(workers: 4) against the same gem with and without the three changed files:

	runs	result	report	summary line
before, workers: 4	5/5	64 runs, 8 failures	none	0 verified, 0 changed, 0 new. NOTHING WAS VERIFIED
after, workers: 4	5/5	64 runs, 8 failures	one, 8 failures	128 verified, 8 changed, 0 new
after, with: :threads	5/5	64 runs, 8 failures	one, 8 failures	128 verified, 8 changed, 0 new
after, serial	3/3	64 runs, 8 failures	one, 8 failures	128 verified, 8 changed, 0 new

The merged report names exactly the 8 stale screenshots, with all image paths resolved. Totals
are 128, not 512 — no worker's pre-fork state is counted more than once. With two baselines
left uncommitted, the merged line reads 126 verified, 8 changed, 2 new and names both across
worker boundaries. No fragment directory survives a run.

Each guard in test/unit/parallel_report_merge_test.rb was mutation-checked: breaking the
idempotence guard, the feature detection, the deferred on_load install, the atomic write, the
merge glob, the owner-pid guard, the missing-baseline merge, the total accumulation, and the key
symbolization each turns a specific test red.

Docs

docs/reporters.md and one row of docs/thread_safety.md said the report is not written under
forking parallelism. Updated. The manual parallelize_teardown { finalize! } workaround they
prescribed is no longer needed; a suite that still carries it keeps working (measured — the
merged report is written last, at the documented path, with every failure in it), it just also
prints a partial summary line per worker.

rake test:unit, rake test:canonical and standardrb lib test are green.

🤖 Generated with Claude Code

https://claude.ai/code/session_014BQJX6eWzBj2UTm5zQsjEs

Summary by Sourcery

Ensure Rails fork-parallel test runs produce one complete HTML screenshot report and accurate suite-wide summary.

Bug Fixes:

• Restore complete HTML reports and summary counts for Rails fork-based parallel test runs by merging screenshot results from all workers in the parent process.

Enhancements:

• Add Rails-specific, feature-detected parallel cleanup integration with temporary JSON handoff fragments, merged missing-baseline tracking, atomic writes, and cleanup after merging.
• Preserve existing behavior for serial, threaded, non-Rails, and legacy workaround configurations.

Documentation:

• Update reporter and thread-safety documentation to describe complete report support under Rails fork parallelism and remove the obsolete workaround guidance.

Tests:

• Add fork-based integration coverage for result merging, summary totals, fragment cleanup, interrupted writes, parent ownership, deferred Rails hook installation, duplicate registration, and Rails-free loading.

[sourcery-ai]

Sorry @pftg, you have reached your weekly rate limit of 500000 diff characters.

Please try again later or upgrade to continue using Sourcery

[coderabbitai]

Warning

Review limit reached

Next included review available in 36 minutes.

View limit details

Limit details: You’ve used the included review currently available.

You've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository.

Learn how review limits work.

Review configuration:

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: 16d1b74d-977c-4058-acfe-b2189faf2bf9

📥 Commits

Reviewing files that changed from the base of the PR and between de5b662 and 71d3ad4.

📒 Files selected for processing (6)

• docs/reporters.md
• docs/thread_safety.md
• lib/snap_diff/integrations/minitest.rb
• lib/snap_diff/reporters/html.rb
• lib/snap_diff/reporting.rb
• test/unit/parallel_report_merge_test.rb

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[sourcery-ai]

Reviewer's Guide

Fixes missing HTML reports and summary counts under Rails fork-based parallel test runs by dumping each worker’s reporter state through a Rails cleanup hook, merging atomic JSON fragments in the parent before finalization, and cleaning up the temporary handoff directory. The change is Rails-feature-detected and deferred when necessary, preserves serial/thread/non-Rails behavior, adds comprehensive fork and failure-safety tests, and updates the concurrency documentation.

Sequence diagram for Rails fork-parallel report merging

sequenceDiagram
    participant Parent
    participant Worker as Rails worker
    participant TempDir as Temp fragment directory
    participant Reporter

    Parent->>Worker: fork test process
    Worker->>Reporter: run tests and record results
    Worker->>Worker: run_cleanup_hooks
    Worker->>Reporter: dump_state
    Reporter-->>Worker: reporter totals and failures
    Worker->>TempDir: write JSON .tmp fragment
    Worker->>TempDir: rename fragment to JSON
    Worker-->>Parent: worker exits
    Parent->>TempDir: read JSON fragments
    Parent->>Reporter: merge_state!
    Parent->>TempDir: remove fragment directory
    Parent->>Reporter: finalize!
    Reporter-->>Parent: write merged HTML report and summary

Loading

Flow diagram for feature-detected Rails parallel hook installation

flowchart TD
    Load["snap_diff minitest integration loads"] --> Install["install_parallel_hooks!"]
    Install --> RailsPresent{"ActiveSupport::Testing::Parallelization defined?"}
    RailsPresent -->|Yes| Register["run_cleanup_hook { dump_parallel_fragment }"]
    RailsPresent -->|No| OnLoad["ActiveSupport.on_load(:active_support_test_case)"]
    OnLoad --> Install
    Register --> Fork["Rails fork workers dump fragments"]
    Fork --> ParentHook["Minitest.after_run"]
    ParentHook --> Merge["merge_parallel_fragments!"]
    Merge --> Finalize["finalize!"]

Loading

File-Level Changes

Change	Details	Files
Adds Rails fork-parallel result handoff and parent-side report merging.	Registers a Rails cleanup hook with deferred ActiveSupport on-load installation and idempotent process ownership. Serializes worker reporter state and missing-baseline names as atomically renamed JSON fragments under the system temp directory. Merges worker fragments in the parent before finalization, symbolizes failure keys, unions missing baselines, and removes the fragment directory. Keeps non-Rails, serial, and thread-parallel behavior unchanged.	lib/snap_diff/integrations/minitest.rb lib/snap_diff/reporting.rb lib/snap_diff/reporters/html.rb
Adds focused coverage for forked report merging, cleanup, safety, and integration compatibility.	Exercises real fork workers and verifies merged totals, failures, new-baseline counts, report generation, and fragment cleanup. Checks atomic-write behavior, ignored temporary fragments, owner-PID isolation, no-fork no-op behavior, idempotent hook registration, and deferred Rails installation. Probes loading the Minitest integration without ActiveSupport.	test/unit/parallel_report_merge_test.rb
Updates concurrency documentation to describe complete Rails fork-parallel reports.	Marks Rails fork parallelism as fully supported with one merged report at the documented path. Documents temporary fragment handling and removes the old per-worker workaround while explaining compatibility caveats. Updates the thread-safety support matrix.	docs/reporters.md docs/thread_safety.md

Assessment against linked issues

Issue	Objective	Addressed	Explanation
#258	Make HTML reports and summary counts survive Rails fork-based parallelization by collecting each worker's records, merging them in the parent, and cleaning up the temporary fragments.	✅
#258	Register the fork-parallel integration only when Rails parallelization is available, while preserving existing serial, threaded, and non-Rails behavior.	✅
#258	Document the automatic fork-parallel report support and replace the prior manual workaround guidance, including the restriction against using per-worker save paths.	✅

Possibly linked issues

• Fork-parallel reports should need zero developer setup #258: The PR directly implements the issue checklist: feature-detected worker hooks, atomic fragments, parent merging, cleanup, tests, and documentation.
• Fork-parallel reports should need zero developer setup #258: The PR directly fixes issue Fork-parallel reports should need zero developer setup #258 by collecting worker results and merging them in the parent.

---

Tips and commands

Interacting with Sourcery

• Trigger a new review: Comment @sourcery-ai review on the pull request.
• Continue discussions: Reply directly to Sourcery's review comments.
• Generate a GitHub issue from a review comment: Ask Sourcery to create an
  issue from a review comment by replying to it. You can also reply to a
  review comment with @sourcery-ai issue to create an issue from it.
• Generate a pull request title: Write @sourcery-ai anywhere in the pull
  request title to generate a title at any time. You can also comment
  @sourcery-ai title on the pull request to (re-)generate the title at any time.
• Generate a pull request summary: Write @sourcery-ai summary anywhere in
  the pull request body to generate a PR summary at any time exactly where you
  want it. You can also comment @sourcery-ai summary on the pull request to
  (re-)generate the summary at any time.
• Generate reviewer's guide: Comment @sourcery-ai guide on the pull
  request to (re-)generate the reviewer's guide at any time.
• Resolve all Sourcery comments: Comment @sourcery-ai resolve on the
  pull request to resolve all Sourcery comments. Useful if you've already
  addressed all the comments and don't want to see them anymore.
• Dismiss all Sourcery reviews: Comment @sourcery-ai dismiss on the pull
  request to dismiss all existing Sourcery reviews. Especially useful if you
  want to start fresh with a new review - don't forget to comment
  @sourcery-ai review to trigger a new review!

Customizing Your Experience

Access your dashboard to:

• Enable or disable review features such as the Sourcery-generated pull request
  summary, the reviewer's guide, and others.
• Change the review language.
• Add, remove or edit custom review instructions.
• Adjust other review settings.

Getting Help

• Contact our support team for questions or feedback.
• Visit our documentation for detailed guides and information.
• Keep in touch with the Sourcery team by following us on X/Twitter, LinkedIn or GitHub.