Overview - Ghost

Ghost

7 days30 days

Latest CI Pipeline Executions

Fuzzy

Succeeded
main
f7d7df8f Added @tryghost/test-data, moved factories over (#29243) no ref This is a strategic move to centralize our data *builders* so that they can be used by both `e2e` (where they are currently located) and future `admin` acceptance tests. There's many tests in /e2e/ that are not true e2e tests and this is the first step towards being able to move those into `admin`. The idea is that the builders are housed in `@tryghost/test-data` and `e2e` can layer on the persistence, which isn't needed by acceptance tests.
by Steve Larson
Succeeded
29244
d5311029 Added a browser-mode acceptance test tier to Admin no ref - runs full-app tests in real Chromium via Vitest Browser Mode, with the Ghost Admin API served by MSW resource handlers fed by @tryghost/test-data - tests mount the real app shell through the shared AdminAppRoot (extracted from main.tsx), so the harness renders the production provider stack by construction - runs in CI through the existing playwright-tag Nx matrix via the new test:acceptance target
by Steve Larson
Failed
main
f7d7df8f Added @tryghost/test-data, moved factories over (#29243) no ref This is a strategic move to centralize our data *builders* so that they can be used by both `e2e` (where they are currently located) and future `admin` acceptance tests. There's many tests in /e2e/ that are not true e2e tests and this is the first step towards being able to move those into `admin`. The idea is that the builders are housed in `@tryghost/test-data` and `e2e` can layer on the persistence, which isn't needed by acceptance tests.
by Steve Larson
Succeeded
main
f7d7df8f Added @tryghost/test-data, moved factories over (#29243) no ref This is a strategic move to centralize our data *builders* so that they can be used by both `e2e` (where they are currently located) and future `admin` acceptance tests. There's many tests in /e2e/ that are not true e2e tests and this is the first step towards being able to move those into `admin`. The idea is that the builders are housed in `@tryghost/test-data` and `e2e` can layer on the persistence, which isn't needed by acceptance tests.
by Steve Larson
Succeeded
29243
72455b82 Added @tryghost/test-data, a shared test-data package for Ghost API shapes no ref - gives entity builders, API response envelopes, and canned response defaults a single home, extracted from e2e's data-factory builders so e2e (real persistence) and future in-app acceptance tests build from the same shapes - e2e's persistence-aware factories now delegate entity building to the package; its id/uuid/slug utils move there too so they aren't maintained twice
by Steve Larson
Succeeded
main
fea5fadd 🎨 Improved analytics chart bucketing for Year to date and All time ranges (#29237) fixes https://linear.app/ghost/issue/PLA-236/ fixes https://linear.app/ghost/issue/PLA-237/ fixes https://linear.app/ghost/issue/PLA-238/ Consolidated analytics date range behavior resulting in some fixes + improvements in the form of consistency across views and some slight behavior changes in terms of summary periods. See PR for more info.
by Steve Larson
Succeeded
main
fea5fadd 🎨 Improved analytics chart bucketing for Year to date and All time ranges (#29237) fixes https://linear.app/ghost/issue/PLA-236/ fixes https://linear.app/ghost/issue/PLA-237/ fixes https://linear.app/ghost/issue/PLA-238/ Consolidated analytics date range behavior resulting in some fixes + improvements in the form of consistency across views and some slight behavior changes in terms of summary periods. See PR for more info.
by Steve Larson
Succeeded
main
fea5fadd 🎨 Improved analytics chart bucketing for Year to date and All time ranges (#29237) fixes https://linear.app/ghost/issue/PLA-236/ fixes https://linear.app/ghost/issue/PLA-237/ fixes https://linear.app/ghost/issue/PLA-238/ Consolidated analytics date range behavior resulting in some fixes + improvements in the form of consistency across views and some slight behavior changes in terms of summary periods. See PR for more info.
by Steve Larson
Succeeded
main
fea5fadd 🎨 Improved analytics chart bucketing for Year to date and All time ranges (#29237) fixes https://linear.app/ghost/issue/PLA-236/ fixes https://linear.app/ghost/issue/PLA-237/ fixes https://linear.app/ghost/issue/PLA-238/ Consolidated analytics date range behavior resulting in some fixes + improvements in the form of consistency across views and some slight behavior changes in terms of summary periods. See PR for more info.
by Steve Larson
Succeeded
main
fea5fadd 🎨 Improved analytics chart bucketing for Year to date and All time ranges (#29237) fixes https://linear.app/ghost/issue/PLA-236/ fixes https://linear.app/ghost/issue/PLA-237/ fixes https://linear.app/ghost/issue/PLA-238/ Consolidated analytics date range behavior resulting in some fixes + improvements in the form of consistency across views and some slight behavior changes in terms of summary periods. See PR for more info.
by Steve Larson
Succeeded
main
16692bdf Fixed analytics healthcheck flooding dev logs with a request every second (#29236) no ref Reduced the health check frequency to not flood to gateway logs.
by Steve Larson
Succeeded
29237
d98b5a5b Changed analytics chart bucketing thresholds and first-bucket labels ref https://linear.app/ghost/issue/PLA-236/ytd-range-renders-unaggregated-daily-points-choppy-analytics-charts ref https://linear.app/ghost/issue/PLA-238/verify-all-time-bucketing-in-post-analytics-web-currently-summarizes Review feedback on the bucketing unification: - The weekly band ran to 356 days, so an ~11-month "All time" chart rendered ~48 weekly points — dense and jagged next to the ~12 monthly points that "Last 12 months" gets. The weekly band now caps at 270 days (~9 months); Year to date keeps its comfortable ~28 weekly points in mid-summer and flips to monthly from October. Only span-resolved ranges (Year to date, All time, post age) can land between the fixed dropdown options of 91 and 372 days, so no dropdown option changes bands. - Buckets are labeled with their calendar period start, which dated the partial first bucket before the chart's own range — a Year to date chart opened at "Week of 28 Dec" of the previous year. The first bucket's label is now clamped to the range start ("Week of 1 Jan"), or to the first data point for All time. Later buckets keep calendar period starts, and the zero-filling in the paid subscriptions chart clamps its period keys the same way, so cross-chart bucket alignment is unchanged.
by Steve Larson
Succeeded
main
b97f3ea4 Changed Nx SHA fallback behavior outside canonical CI (#29235) no ref - keep missing-successful-run errors strict in canonical Ghost CI - allow other repository instances to use `nx-set-shas`' default `HEAD~1` bootstrap fallback
by Steve Larson
Succeeded
29236
76f0a8be Fixed analytics healthcheck flooding dev logs with a request every second no ref The analytics and tinybird-local healthchecks used interval: 1s with no start_period, and that interval applies for the container's entire lifetime, not just startup. Since ghost/traffic-analytics logs every incoming request at INFO in GCP structured format, pnpm dev:analytics flooded the terminal with one log line per second indefinitely. Switched to a short start_interval during a start_period window for fast startup detection, then a quiet 30s steady-state interval.
by Steve Larson
Succeeded
29235
13847172 Changed Nx SHA fallback behavior outside canonical CI no ref Only canonical CI needs a hard failure when no successful main run exists; other repository instances need the action's HEAD~1 bootstrap fallback.
by Steve Larson
Succeeded
main
ccbeb3e4 Fixed CI failures in downstream repositories (#29234) no ref - skip the traffic analytics infra dispatch outside `TryGhost/Ghost`, where the required repository token is unavailable - size Koenig Playwright's CI worker pool from the runner CPU count instead of forcing four workers
by Steve Larson
Succeeded
main
ccbeb3e4 Fixed CI failures in downstream repositories (#29234) no ref - skip the traffic analytics infra dispatch outside `TryGhost/Ghost`, where the required repository token is unavailable - size Koenig Playwright's CI worker pool from the runner CPU count instead of forcing four workers
by Steve Larson
Succeeded
29234
5408b051 Changed Koenig Playwright workers to match runner capacity no ref CI runner sizes vary, so deriving the worker pool from available CPUs avoids oversubscription while preserving throughput.
by Steve Larson
Succeeded
29234
74966716 Changed Koenig Playwright workers to match runner capacity ref https://github.com/TryGhost/Ghost-Anvil/actions/runs/29097979933/job/86379792409 Private GitHub runners have half the CPUs of public runners, so a fixed four-worker pool overloads downstream CI.
by Steve Larson
Succeeded
main
e44f46ae Fixed flaky welcome email bookmark test (#29232) no ref Attempting to patch a flaky test - wait for the bookmark URL autocomplete option to render and select it directly rather than immediately using a keyboard input which is likely racing ahead of the suggestion.
by Steve Larson