lsm / HyperNeo / 30283927832
83%

DEFAULT BRANCH: dev
Ran
Jobs 0
Files 0
Run time –
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst
30283927832

push

github

web-flow 
test: add external-event delivery end-to-end reliability suite (#2248)

* test: add external-event delivery end-to-end reliability suite

Wire the full delivery chain in one suite — real GitHubEventExtension
(webhook handler + poller with a routed fake fetch) → ExternalEventService/
Store → internal bus → SpaceRuntimeService/SpaceRuntime → captured worker
injection — covering the paths the reliability batch fixed in isolation:

- blocked run wakes exactly once from a PR review webhook; duplicate
  deliveries short-circuit at the store
- check_run failure webhooks reach the PR worker; success conclusions drop
- polling fallback delivers PR comments with no webhook configured
- /pulls pages with stale rows clear the backlog so check_run polling is
  not starved; genuine backlogs defer check_run polling without dropping it
- inactive-worker events persist and flush exactly once on activation
- idle subscribers receive webhook events in defer mode
- /pulls metadata bumps on the same head sha do not re-trigger workers
- PR events linked to a run stay published until a subscription registers;
  unlinked events are terminally ignored

Registered in the 5-space-runtime-b daemon test shard.

* test: address review nits in external-event e2e suite

- Rephrase the duplicate-delivery comment: the review dedupeKey is keyed
  on the review's own id, with X-GitHub-Delivery only a fallback. The
  duplicate webhook now also uses a different delivery id so the test
  proves dedupe holds across delivery headers.
- Drop the `?? 1` fallback in the processedPages.pulls assertion so a
  missing cursor field fails instead of passing silently.

* test: fail loudly when the startup tick does not go idle in e2e setup

The bounded polling loop could expire with the fire-and-forget startup
tick still running, and the follow-up executeTick() would then no-op on
tickInFlight — silently reintroducing the seeding race the wait exists to
prevent. Throw after a 10s deadline instead so ... (continued)
Source Files on build 30283927832
Detailed source file information is not available for this build.