rspec-tracer 1.0.5 → 1.1.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 67309438f6000ac9913b4d3f48b3e7ba66c56e09674b2bf191b75d0d8134401c
-  data.tar.gz: e10e5253987938b6467920a7adda07dbbce152849373b1c8a4355e771007c196
+  metadata.gz: b7f283b4a6ab3d0d2616f7dca4bac721906d36b4296ba1d64446937db1503dd5
+  data.tar.gz: 2922f96420e344603a472b4e67bb5f2c58c16e8bf798e13098eec8f8096434c5
 SHA512:
-  metadata.gz: 80b3e75b07083b15e179fa4d108afbf31271ee15c3b87d48625ccfa481a6fa24a70114246fe9644b6d2d743732515029e9ec93e078a9207ed1c4c98ba61964ca
-  data.tar.gz: 8ac718e348b8cc6559b87b45d117cb20ddbca819687208124e1c8e7df557e97528d0da93823fe502e9d689646457037ce9be4c7529f3bdf853ecbcbb6f7a98e7
+  metadata.gz: 222031b5d7b0a6b01cad104fda83347a9e10af957c7a5d1dfede9f46cf8a275435c3e0c9dfb02ad7788b3dfc2d4c3b9513a60ee3ccf0d3bfd68a692758a38f1e
+  data.tar.gz: f2828567b26b580ffc66568f8b14581cb160f4bdf14a61622b605b5832d5d3bc86cff211190628c5480604f0ec1c758609ec8aa490b357016e3565615c635be2

data/CHANGELOG.md

@@ -1,247 +1,54 @@
-## [1.0.5] - 2026-05-17
+## [1.1.0] - 2026-04-20
 
-### Fixed
-
-- **`example_id` is now stable across runs and across no-op edits** —
-  `Example.from` previously fed `example.example_group.name` and
-  line numbers into the identity-hash payload. Two unrelated effects
-  shifted the resulting MD5 across runs that should have been
-  identical:
-
-  - `example_group.name` is RSpec's generated class name with a
-    load-order-dependent `_2` / `_3` suffix when two spec files share
-    a `describe` name (very common — e.g. `describe User do` in
-    `user_spec.rb` and `models/user_spec.rb`). The same example got
-    a different id depending on rspec's file load order, silently
-    thrashing the cache and breaking failed / pending always-re-run
-    guarantees.
-  - For unnamed examples (`it { }` / `specify { }` / `example { }`),
-    `example.description` falls back to RSpec's line-bearing
-    `"example at <path>:<line>"` string, so a no-op blank-line edit
-    above the example flipped its id and orphaned its cache entry.
-
-  The digest now uses `example_group.description` (the user's
-  string, not the generated class name), strips the trailing `:LINE`
-  from `shared_group_inclusion_backtrace` entries, and for unnamed
-  examples substitutes a line-independent positional discriminator
-  (the example's 0-based ordinal among the unnamed examples of its
-  group). `line_number` / `rerun_file_name` / `rerun_line_number`
-  still ride along in the stored payload for the reporter's location
-  columns but no longer enter the digest. Contract: *rename = new
-  identity; restructure = same identity.*
-
-  Affected since v1.0.0 (2021). One-time cold run on upgrade: every
-  cached `example_id` changes shape, so the first 1.0.5 run misses
-  the 1.0.4 cache uniformly and treats every example as no-cache;
-  the second run is warm again. Surfaced by the 2.0.0.pre.1 field
-  test against third-party Rails apps; ports
-  [#209](https://github.com/avmnu-sng/rspec-tracer/pull/209) and
-  [#211](https://github.com/avmnu-sng/rspec-tracer/pull/211) onto
-  this 1.0.x line. Closes
-  [#196](https://github.com/avmnu-sng/rspec-tracer/issues/196) and
-  [#210](https://github.com/avmnu-sng/rspec-tracer/issues/210).
-
-- **`RSpecTracer.start` no longer crashes when the user pre-started
-  `::Coverage`** — users who want branch coverage typically call
-  `Coverage.start(lines: true, branches: true)` before loading the
-  tracer. On the previous code path, `setup_coverage` called bare
-  `::Coverage.start` unconditionally, which raised
-  `RuntimeError: coverage measurement is already setup` and crashed
-  the tracer init. `setup_coverage` now guards on `Coverage.running?`
-  (Ruby 2.7+) and rescues `RuntimeError` for older Rubies; the
-  tracer attaches to the already-running `::Coverage` instance
-  instead of raising. Ports
-  [#207](https://github.com/avmnu-sng/rspec-tracer/pull/207)'s
-  Coverage-guard half (the 2.0-only `coverage_modes` DSL is not
-  backported). Closes
-  [#195](https://github.com/avmnu-sng/rspec-tracer/issues/195).
-
-- **`remote_cache` download and upload now log on success** — the
-  success paths on `RemoteCache::Cache#download` / `#upload` returned
-  silently, so a successful `rake rspec_tracer:remote_cache:download`
-  produced zero output and users couldn't tell from CI logs whether
-  the cache restored. Two `puts` lines now announce success:
-  `rspec-tracer remote_cache: restored cache from <sha>` after a
-  successful download, and `rspec-tracer remote_cache: uploaded cache
-  to <ref>` after a successful upload. Ports the basic-INFO portion of
-  [#201](https://github.com/avmnu-sng/rspec-tracer/pull/201) (the
-  cross-branch-fallback qualifier and `prune_all!` line are
-  2.0-only — 1.x has neither a multi-tier cache nor a prune_all
-  path). Closes
-  [#188](https://github.com/avmnu-sng/rspec-tracer/issues/188).
-
-### Changed
-
-- **Gemspec now requires MFA for publishing** — adds
-  `spec.metadata['rubygems_mfa_required'] = 'true'`. Pure packaging
-  metadata; no runtime impact for gem consumers. Ports
-  [#214](https://github.com/avmnu-sng/rspec-tracer/pull/214).
-
-## [1.0.4] - 2026-05-06
-
-### Fixed
-
-- **Parallel-tests purge race when a sibling worker is still mid-flush**
-  — the elected worker trusted only `parallel_tests`'s pid-file barrier
-  (`ParallelTests.wait_for_other_processes_to_finish`), which under
-  specific scheduling/I/O timing on GHA Linux x86_64 can return while a
-  sibling's `parallel_tests_N/` dir hasn't fully flushed. The elected
-  then merged + purged, racing the in-progress sibling. Symptoms:
-  intermittent leftover `parallel_tests_N/` dir post-purge AND/OR
-  silently dropped peer caches in the merge.
-
-  Backport of upstream PR
-  [#168](https://github.com/avmnu-sng/rspec-tracer/pull/168). Adds a
-  filesystem barrier layered on top of the pid-file wait. Each worker
-  writes a `.rspec_tracer_boot` marker at `RSpecTracer.start` time and
-  a `.rspec_tracer_done` marker as the first step of its at_exit tasks;
-  the elected worker waits for every booted peer's `.done` to
-  materialize before proceeding to merge + purge. Two independent
-  signals (pid file + filesystem) must agree before the elected worker
-  declares the peer set stable. Bounded at 5 s with a graceful warn
-  for crashed peers — their dirs are purged regardless of completion
-  state, and the merge accepts whatever's on disk.
-
-## [1.0.3] - 2026-05-04
-
-### Fixed
+### Added
 
-- **Carry-forward filter contract** — newly-added filters now apply
-  uniformly to both fresh attribution AND prior-snapshot carry-forward.
-  Previously, `Cache#load_all_files_cache` and `load_dependency_cache`
-  read previous-run state without re-applying the current filter list.
-  A user adding a new filter mid-development saw the filter take
-  effect only for fresh attributions on cold runs; previously-cached
-  paths matching the new filter persisted in `all_files` and
-  `dependency` until the next cold run. Filter additions now take
-  effect on the very next warm run. (from upstream
-  [PR #161](https://github.com/avmnu-sng/rspec-tracer/pull/161))
-
-### Note on exclusions
-
-The companion default-filter expansion shipped in upstream PR #161
-(adds `rspec_tracer_cache/`, `rspec_tracer_coverage/`,
-`rspec_tracer_report/`, `rspec_tracer.lock` to the default
-`add_filter` / `add_coverage_filter` lists) is intentionally NOT
-backported to this 1.0.x line, for the same reason the broader
-1.1.0 default-filter expansion was excluded from 1.0.1: changing
-the default filter set shifts the files present in `all_files.json`
-for users who track tracer-self paths, which would invalidate their
-existing caches on upgrade. Users on 1.0.x who want this default
-hygiene can either upgrade to 1.2.x / 2.0.x OR add the four paths
-to their own `.rspec-tracer` config explicitly. The carry-forward
-filter check shipped here MAKES that user-side `add_filter` take
-effect on the next warm run.
-
-## [1.0.2] - 2026-05-01
+- Ruby 3.1, 3.2, 3.3, 3.4, 4.0 are now CI-gated.
+- Rails 7.0, 7.1, 7.2 are CI-gated via a reference sample app.
+- Regression specs for four crash bugs (B1–B4) under
+  `spec/lib/rspec_tracer/`.
 
 ### Fixed
 
-- **Parallel-tests merge silently dropped peer caches and left worker
-  directories behind** when the spawned-worker count exceeded
-  `ENV['PARALLEL_TEST_GROUPS']`. The merge + purge call-sites in
-  `lib/rspec_tracer.rb` (`merge_parallel_tests_reports`,
-  `merge_parallel_tests_coverage_reports`,
-  `purge_parallel_tests_reports`) iterated `1..ENV['PARALLEL_TEST_GROUPS']`
-  to construct per-worker directory names. But parallel_tests sets
-  `PARALLEL_TEST_GROUPS = num_processes.to_s` for each child, where
-  `num_processes` is the user-requested process count
-  (`Parallel.processor_count` by default) — not the actual worker
-  count. When `num_processes < spawned_worker_count` (e.g. when the
-  spec-count partition produces more non-empty groups than
-  `num_processes`, or shared-runner CPU detection drifts mid-run),
-  peer caches with `TEST_ENV_NUMBER` above the env bound were silently
-  dropped from the merge (warm-run skip decisions get made against
-  an under-sampled merged manifest) and left behind by the purge
-  (visible as straggler `parallel_tests_<N>/` directories under
-  `rspec_tracer_cache/`). The same gem behaviour was documented on
-  v1.1.1's `last_process?` fix
-  ([PR #101](https://github.com/avmnu-sng/rspec-tracer/pull/101)) but
-  the iteration call-sites kept the buggy bound. Each method now globs
-  the actual `parallel_tests_*` subdirectories under its base path,
-  making the merge + purge robust to whatever count parallel_tests
-  spawned. (from v1.2.1) No cache format change.
-
-## [1.0.1] - 2026-04-24
-
-Long-tail maintenance release. Backports high-impact crash and correctness
-fixes from 1.1.x / 1.2.x onto the v1.0.0 foundation so users on Ruby 2.5 -
-3.0 who can't upgrade can still pick them up via `gem 'rspec-tracer', '~> 1.0'`.
+- **B1** — `Cache#cached_examples_coverage` returns `{}` (not `nil`)
+  when `last_run.json` is present but `examples_coverage.json` is
+  missing; previously this leaked nil into the missed-coverage merge.
+- **B2** — `Runner#generate_missed_coverage` tolerates a nil cached
+  coverage map and nil line-strength entries; both are treated as
+  empty / zero.
+- **B3** — `Runner#register_{file,example_file}_dependency` skips
+  (logs debug, returns false) when `SourceFile.from_path` /
+  `.from_name` cannot resolve the file (e.g. gem-generated examples
+  or files deleted between runs).
+- **B4** — `CoverageReporter#merge_coverage` treats nil existing
+  line coverage as 0 when summing skipped-test contributions.
+- Custom filter and coverage-filter blocks now reach
+  `RSpecTracer::Filter.register` — the DSL wrappers were dropping
+  the block, causing `add_filter { |sf| … }` to raise
+  `ArgumentError`.
+- `load_global_config.rb` wraps `Dir.home` / `Etc.getpwuid.dir` /
+  `File.expand_path("~user")` in `rescue ArgumentError` so gem
+  load never crashes in minimal containers where HOME is unset and
+  the passwd database has no matching entry.
 
-### Fixed
+### Changed
 
-- `Cache#cached_examples_coverage` returns `{}` (not `nil`) when `last_run.json`
-  is present but `examples_coverage.json` is missing, preventing a
-  NoMethodError on the next consumer. (B1, from v1.1.0)
-- `Runner#generate_missed_coverage` tolerates a nil cached coverage map and
-  nil per-line strength entries. (B2, from v1.1.0)
-- `Runner#register_file_dependency` / `#register_example_file_dependency`
-  skip gracefully when `SourceFile.from_path` / `.from_name` resolves to
-  nil (e.g. gem-generated examples whose path is absent at runtime),
-  instead of crashing. (B3, from v1.1.0)
-- `CoverageReporter#merge_coverage` treats a nil existing line-coverage
-  entry as 0 when summing skipped-test contributions. (B4, from v1.1.0)
-- `parallel_tests`-mode merge-worker election no longer deadlocks under
-  slow CI: the elected worker is now picked via
-  `::ParallelTests.first_process?` (immutable at spawn) instead of the
-  lock-file max TEST_ENV_NUMBER (racy on slow runners). (from v1.1.1)
-- Pin `encoding: 'UTF-8'` on every legacy `File.read` / `File.write` JSON
-  I/O site so shells with `LANG=` unset no longer crash the tracer on
-  multibyte spec descriptions. Cache (11), report_writer (12), and nine
-  other lib/ call sites covered. (from v1.1.2)
-- `SourceFile.from_path` computes the file digest via `File.binread`, hashing
-  raw bytes regardless of Encoding.default_external. (from v1.1.2)
-- `SourceFile.file_path` returns an absolute-external path unchanged when
-  the referenced file exists on disk (e.g. shared examples from vendored
-  gems at `/opt/bundle/gems/...`), preventing silent drop of dependency
-  registration. The guard is narrow — `start_with?('/') &&
-  !start_with?(root) && File.file?(path)` — so stripped-root cache forms
-  like `/spec/foo.rb` continue through the existing expand_path branch and
-  cache `file_name` keys stay byte-identical to v1.0.0. (C1, from v1.2.0)
-- `RemoteCache::Aws#upload_dir` error message corrected from
-  "Failed to download files from" to "Failed to upload files from". (C3,
-  from v1.2.0)
-- `RemoteCache::Validator::ValidationError` declared as a proper
-  `StandardError` subclass inside `Validator`; previously the
-  `TEST_SUITE_ID ^ TEST_SUITES` XOR-guard raised a `NameError:
-  uninitialized constant` instead of the intended validation error.
-  (from v1.2.0)
-- `enviornment` → `environment` typo in the same XOR-guard raise message.
-  (from v1.2.0)
-- `RemoteCache::Validator`'s single-suite `@cached_files_regex` anchored
-  with a trailing `$` so files with extensions beyond `.json` (e.g.
-  `.json.backup`) no longer match as cache files. (from v1.2.0)
-- `RemoteCache::Repo#initialize` guards `ENV['GIT_BRANCH']` for nil
-  before calling `.chomp`; previously a `NoMethodError: undefined
-  method 'chomp' for nil:NilClass` crashed the init path and masked
-  the intended `RepoError` message when `GIT_BRANCH` was not set in
-  the environment. (from v1.1.0 PR #51)
-- `RemoteCache::Repo#download_branch_refs` uses `FileUtils.rm_f`
-  (not `File.rm_f`) to clean up a partial `branch_refs.json` on a
-  failed AWS download. `File.rm_f` is undefined — `rm_f` is a
-  FileUtils method — so the failing-download branch would crash with
-  `NoMethodError` instead of cleaning up and logging. (from v1.1.0
-  PR #65)
-
-### Note on exclusions
-
-The following items from 1.1.x / 1.2.0 are intentionally NOT in this
-release to preserve the 1.0.0 cache-format contract and Ruby 2.5+ floor:
-
-- Default-filter expansion (1.1.0 — adds `/lib/rspec_tracer/`,
-  `/usr/local/lib/ruby/`, etc. to the default filters). Changing the
-  default filter set shifts the files present in `all_files.json` for
-  most users, which would invalidate existing caches on upgrade.
-- `USE_TEST_SUITE_ID_CACHE` opt-in ENV flag (1.2.0). This is a new
-  feature, not a bug fix; users who want it can upgrade to 1.2.x.
-- The 1.1+ configuration DSL refactor (anonymous block forwarding,
-  alias_method wrapping, ENV.fetch normalizations). These are
-  Ruby-3.1-exclusive in places and orthogonal to the crash-fix scope.
-
-### Ruby support
-
-Gemspec `required_ruby_version` unchanged at `>= 2.5.0`. CI gates
-Ruby 2.5 - 4.0 inclusive on `ubuntu-latest`.
+- **Behavior change (default filters)** — the default dependency
+  and coverage filter lists now exclude Ruby installation /
+  toolchain paths: `/lib/rspec_tracer/`, `/lib/rspec_tracer.rb`,
+  `/usr/local/lib/ruby/`, `/usr/local/bundle/`,
+  `/opt/hostedtoolcache/`, `/.rbenv/versions/`,
+  `/.asdf/installs/ruby/`, `/.rvm/`. Previously only
+  `/vendor/bundle/` was filtered. A test that previously recorded
+  a dependency on a gem file or Ruby stdlib file (because of a
+  custom install path like rbenv or asdf) will no longer record
+  that dependency — those paths are handled by `Gemfile.lock` /
+  the Ruby version file, not by coverage tracking. If you relied
+  on the old narrow default, add your own `add_filter` /
+  `add_coverage_filter` to clear the extras.
+
+### Removed
+
+- Support for Ruby ≤ 3.0 and Rails ≤ 6.x (EOL).
 
 ## [1.0.0] - 2021-10-21