prisma-risk / tsoracle / 26371115568
95%

DEFAULT BRANCH: main
Ran
Jobs 1
Files 75
Run time 1min
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
coverage: 94.953% (+0.04%) from 94.911%
26371115568

push

github

web-flow 
test(server): run timing-fragile fence/churn tests under turmoil DST (#305)

The fence retry / leadership-churn tests in leadership_churn.rs waited up to
10-15s of real time for the fence to retry transient persist failures with
backoff, and one raced the retry loop against the leadership stream. Those
real-time waits are the same class of flake that bit the coverage job: timing
that only resolves under specific scheduling.

Add a shared deterministic-simulation harness in tsoracle-server::test_support
(behind a new opt-in `dst` feature pulling turmoil): the Accepted/Connected +
connector transport glue, plus into_sim_parts (exposes state_rx and spawns the
leader-watch task under simulated time), serve, and client. turmoil is an
optional dependency, so a default build of the published crate is unaffected.

Convert the three timing-fragile tests to leadership_churn_dst.rs running under
turmoil: tokio::time::sleep backoff advances simulated time, so the same
coverage runs deterministically and instantly (~0.03s vs 10-15s) from a fixed
seed. Remove the flaky real-time originals; keep the four extend_window_maps_*
error-mapping tests and serve_with_shutdown (MockClock-based, not timing-fragile).

Refactor dst_window_race.rs onto the shared harness, dropping its inline glue,
and slim tsoracle-tests' dev-deps accordingly (the glue now lives in the server
crate; tsoracle-tests' `dst` feature enables tsoracle-server/dst).

11269 of 11868 relevant lines covered (94.95%)

467373.13 hits per line

Jobs
ID Job ID Ran Files Coverage
1 26371115568.1 75
94.95
 GitHub Action Run
Source Files on build 26371115568