Skip to content

fitness: wait for the time-up from the GM's /start, not the runner's arrival - #1009

Merged
neeerp merged 1 commit into
masterfrom
v83/fitness-timeup
Oct 25, 2026
Merged

neeerp merged 1 commit into
masterfrom
v83/fitness-timeup

Conversation

@neeerp

@neeerp neeerp commented Oct 25, 2026

Copy link
Copy Markdown
Owner

The master suite's fitness test failed twice on 125ac8b at the "out of stage 1" step.

Cause: the test's wait, not the time-up.

  • until gives the runner 15 s to be warped out, counted from the runner's arrival on stage 1.
  • The race also lasts 15 s (900 s scaled by 60), counted from the GM's /start.
  • So the only slack was the second or less between /start and the arrival. Against that come the 100 ms field tick and the warp itself, which can take longer when the suite runs other tests at the same time.

The server side is fine.

  • The runner got its countdown on stage 1, so the race was running there.
  • end_race warps every occupant who isn't a GM.
  • The failing trace has the last notice, at 14.83 s, but not the end, at 15.02 s, which is consistent with the wait running out first.

Change

  • trace::until_within(c, trace, wait, done) waits for a given time; until now calls it with 15 s.
  • The time-up step in fitness.rs waits 30 s.

Verification: fitness passed three times in a row on a fresh lane stack, and gm_ban passed alongside it.

…art, not from the runner's arrival on stage 1 (until_within)
@neeerp
neeerp merged commit 932027e into master Oct 25, 2026
1 check passed
@neeerp
neeerp deleted the v83/fitness-timeup branch October 25, 2026 20:14
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant