compile: retry the artifact rename through a transient Windows hold

[colinhacks]

The compile harness on the win32 CI legs intermittently fails the publish step with atomically replacing ./bin-<name>.exe with staged output … The process cannot access the file because it is being used by another process. (os error 32). It hit #718's win32-x64 run and #583's win32-arm64 run within a few hours of each other today. An antivirus scan or the search indexer holds the freshly written executable for a moment after its last handle closes; nothing of ours still has it open.

The publish step now retries a rename that reports ERROR_SHARING_VIOLATION or ERROR_LOCK_VIOLATION with a short backoff (25 ms doubling to 400 ms, about 1.6 s in total) and adds the attempt count to the error when the hold outlasts it. A non-transient failure is still reported at once. graceful-fs retries a win32 rename on the same class of error for the same reason.

Tests: the retry path (two violations, then the rename lands), the exhausted path (the destination is untouched and the error names the attempt count), and the existing non-transient test now asserts a single attempt.

Gates: CI's clippy invocation and cargo test -p nub-cli --features compile --bin nub atomic_publish on a builder.

[pullfrog]

Pullfrog stalled — likely cause: provider usage limit reached

The agent stopped emitting events for 125s and was killed by the activity-timeout watchdog. 101 events were processed before the failure.

Recent agent stderr

timestamp=2026-09-06T00:36:44.739Z level=ERROR run=2a9b2a23 message="stream error" providerID=openai modelID=gpt-5.6-sol session.id=ses_f8bdaa29dffe9u53ikZwvgxXxx small=false agent=build mode=primary error.error="AI_APICallError: The usage limit has been reached"
timestamp=2026-09-06T00:36:47.562Z level=ERROR run=2a9b2a23 message="stream error" providerID=openai modelID=gpt-5.6-sol session.id=ses_f8bdaa29dffe9u53ikZwvgxXxx small=false agent=build mode=primary error.error="AI_APICallError: The usage limit has been reached"
timestamp=2026-09-06T00:36:52.023Z level=ERROR run=2a9b2a23 message="stream error" providerID=openai modelID=gpt-5.6-sol session.id=ses_f8bdaa29dffe9u53ikZwvgxXxx small=false agent=build mode=primary error.error="AI_APICallError: The usage limit has been reached"
timestamp=2026-09-06T00:37:00.543Z level=ERROR run=2a9b2a23 message="stream error" providerID=openai modelID=gpt-5.6-sol session.id=ses_f8bdaa29dffe9u53ikZwvgxXxx small=false agent=build mode=primary error.error="AI_APICallError: The usage limit has been reached"
timestamp=2026-09-06T00:37:16.918Z level=ERROR run=2a9b2a23 message="stream error" providerID=openai modelID=gpt-5.6-sol session.id=ses_f8bdaa29dffe9u53ikZwvgxXxx small=false agent=build mode=primary error.error="AI_APICallError: The usage limit has been reached"
timestamp=2026-09-06T00:37:49.402Z level=ERROR run=2a9b2a23 message="stream error" providerID=openai modelID=gpt-5.6-sol session.id=ses_f8bdaa29dffe9u53ikZwvgxXxx small=false agent=build mode=primary error.error="AI_APICallError: The usage limit has been reached"
timestamp=2026-09-06T00:38:54.030Z level=ERROR run=2a9b2a23 message="stream error" providerID=openai modelID=gpt-5.6-sol session.id=ses_f8bdaa29dffe9u53ikZwvgxXxx small=false agent=build mode=primary error.error="AI_APICallError: The usage limit has been reached"

  ^{｜ Rerun failed job ➔ ｜ View workflow run ｜ via Pullfrog ｜ Using GPT Sol ｜ 𝕏}

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
nub	Ready	Preview	Sep 6, 2026 12:36am UTC

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[colinhacks]

Superseded by #898, which retries the same publish rename with the self-probe root cause and a Windows-gated predicate.