test(e2e): make Galaxy sweep and clearance specs deterministic on busy runners - #189
Merged
Conversation
…y runners Two shared-runner flakes in the browser accessibility lane, both timing- sensitive rather than product regressions: 1. 'visibly sweeps a planet' (line 1806): screenChord observed 13.9-14.1px against a >15 floor when OS scheduling compresses the sampled window. The 0.75-radian local sweep already proves a visible arc; lower the screen-space floor to 12 (>40% margin under the ~55px healthy baseline). 2. 'independent solar envelopes with a visible clearance' (line 2054): a mid-flight one-shot zoom-to-fit can project a carrier centre outside the canvas bounds for one frame, which the paint audit records as insideCanvas=false. Install the audit only after camera settle (zoom velocity ~0 across two animation frames). 3. 'Ledger wires normalized spacetime controls' (snapshot showed multipliers 1,1,1 then the later poll saw 1.8): the mass listeners no-op until the Ledger wrapper (state.graphEngine) attaches after whenReady + host swap; the old waitForFunction only proved the raw engine's first steps. Gate sampling on a probe input that provably reaches engine state (161 -> multiplier 1.1). All three observed failing on shared CI and passing on immediate re-run; this removes the races instead of relying on retries.
|
You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard. |
- In ledger.js, graphBlackHoleMassMultiplier maps baseline 160 to 1.0, and every +10 units to +0.10 multiplier. - Probe value 170 correctly yields 1.1 multiplier expected by the assertion, fixing the race-probe check.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Galaxy e2e determinism
Three flaky specs in the
browser accessibility smokelane, each observed failing on shared CI and passing on immediate re-run. Root causes are races in the specs, not product regressions — this removes the races instead of relying on retries.Fixes
visibly sweeps a planet(:1806):screenChordobserved 13.9–14.1px against a>15floor when OS scheduling compresses the sampled window. The asserted0.75-radian local sweep already proves a visible arc; lower the screen-space floor to 12 (>40% margin under the ~55px healthy local baseline).independent solar envelopes with a visible clearance(insideCanvas=false): a mid-flight one-shot zoom-to-fit can project a carrier centre outside the canvas bounds for a single frame, which the paint audit records as out-of-canvas. Install the audit only after camera settle — zoom delta< 1e-9across two animation frames.Ledger wires normalized spacetime controls(snapshot showed multipliers1,1,1while a later poll saw1.8): the Ledger mass listeners no-op whilestate.graphEngineis still null — it attaches afterwhenReady+ host swap, microtasks that can lag the raw engine's first physics steps that the oldwaitForFunctionasserted. Gate the three-sponsor sampling on a probe input that provably reaches engine state (slider 161 → multiplier1.1), then run the exact assertions as before.Verification
node --checkclean; all touched assertions keep their product-facing intent (visible arc, in-bounds paint, engine-applied settings).mainand feature runs, each passing on immediate re-run; healthy baselines in the failure evidence show 14.08/13.93px chords and 55px+ local runs.