@decocms/parity 0.35.26 → 0.35.28
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/CHANGELOG.md +22 -0
- package/dist/cli.js +750 -264
- package/package.json +1 -1
package/CHANGELOG.md
CHANGED
|
@@ -41,6 +41,15 @@ and this project adheres to [Semantic Versioning](https://semver.org/).
|
|
|
41
41
|
* **Determinism test**: the same recording replayed twice must give identical checks, issues
|
|
42
42
|
(ids, severity, details, evidence), verdict and score, order included; the corpus suite repeats
|
|
43
43
|
that for every case.
|
|
44
|
+
* **`parity run --golden <dir>`** uses the prod half of a `--record` recording as the reference
|
|
45
|
+
and captures only cand (`--prod` optional). Prod is not contacted at all: pages, flows, visual
|
|
46
|
+
probes and the resolved config come from the recording, cand is captured at exactly the
|
|
47
|
+
recording's page keys, and prod-origin `fetch` calls made by checks are served only from its
|
|
48
|
+
fetch log. A prod request the recording never saw is reported as `golden.staleRequests` in
|
|
49
|
+
`report.json` and exits 2. Validation (other prod origin, missing viewports, uncovered check
|
|
50
|
+
selection, empty golden, `--pages` outside the golden) exits 2 before any browser work. Recordings
|
|
51
|
+
now also carry the detected `platform` (optional; older recordings still load). See
|
|
52
|
+
`docs/cli.md`.
|
|
44
53
|
|
|
45
54
|
### Fixed
|
|
46
55
|
|
|
@@ -101,6 +110,19 @@ and this project adheres to [Semantic Versioning](https://semver.org/).
|
|
|
101
110
|
|
|
102
111
|
### Changed
|
|
103
112
|
|
|
113
|
+
* **A missing tag manager is one finding per page, not four checks' worth.** `analytics-beacon-parity`
|
|
114
|
+
reports `analytics-beacon:tag-manager-absent:<page>` when prod loaded the GTM container and cand did
|
|
115
|
+
not, naming the container and the tags that go with it (replaces the per-tag
|
|
116
|
+
`analytics-beacon:missing:*` ids on that page; with the container present on cand they are
|
|
117
|
+
unchanged). `third-party-scripts-parity` lowers analytics/marketing vendor origins explained by it
|
|
118
|
+
to `low` + `inconclusive` (other missing origins keep `high`), and `html-structural-diff`
|
|
119
|
+
discounts those vendors' `<script>` tags from the script count and lists them once in
|
|
120
|
+
`html-structural:tag-manager-scripts`. Shared helper: `checks/lib/tag-manager.ts`.
|
|
121
|
+
* **A local dev server counts as a staging host** (`localhost`, `127.0.0.0/8`, `[::1]`,
|
|
122
|
+
`*.localhost`, prod not local), so the existing staging leniency applies there without a noindex
|
|
123
|
+
signal; `--cand-is-production` turns it off. On a dev server the consolidated tag-manager finding
|
|
124
|
+
is `low` + `inconclusive`.
|
|
125
|
+
|
|
104
126
|
* **`parity run`'s visual module runs on the `parity heatmap` engine — no LLM.**
|
|
105
127
|
The old pass sent screenshot pairs to a vision model and filed what it said.
|
|
106
128
|
On real runs it invented findings — a stat card on neither side, prod and cand
|