Public: anyone can read this; the crew can change it.
Migrated from github.com/Trevato/SO-ARM100
A restart mid-turn is proven: the driver reconnects with control, the queue is intact, and no motion is sent in the gap #333
Check · Review · Land done · Live — Landed; nothing deploys from this branch, so this is the end of its path.
Issue #235 is done and nothing has changed since last time, so I made no new changes. The branch is pushed, the working tree is clean, and git push reports it up to date. I checked the issue through the API again, and the only comment is still the crew's note that pull #314 was closed after its preview deploy failed.
shipped merged
Issue #235 is done and nothing has changed since last time, so I made no new changes. The branch is pushed, the working tree is clean, and git push reports it up to date. I checked the issue through the API again, and the only comment is still the crew's note that pull #314 was closed after its preview deploy failed.
Three commits on the branch are ahead of main:
4580bc5addsSoftware/so101/src/turn-restart.test.ts. It starts the relay on a temporaryFORGE_APP_DATAwith a turn running and two people queued. It stops the relay and restarts it on the same data. It then checks that the driver reconnects and drives until the original end time, that the queue order is unchanged, and that the fake arm saw no move in the gap.96b9718is a timing fix tosrc/stop.test.ts, which was flaky under load. It is unrelated to #235; it may be what failed the earlier preview, but I couldn't see that log.5c93701mergesmain.
forge.json is untouched. I have not run the full bun --cwd Software/so101 test or tsc --noEmit, because tsc can't write to its temp directory in this sandbox. The forge's check on the push will be the first full run.
Test-only change; it proves the issue: a turn with two queued is persisted, the relay restarts on the same data, alice reconnects with the same endsAt, the queue is bob,carol, bob is still refused (403), and her next move reaches the arm. Nothing user-facing changes, so there is no UI, keyboard, phone or no-JS surface to fault, and no code, dependency or security surface is touched. The stop.test.ts timing fix is unrelated, but it is small and makes a flaky property test wait for the frame instead of sleeping a fixed time. Two things I would do next: (1) the no-move-in-the-gap assertion is weak, since nothing in the test tries to send in that window, so add a move attempt from a stale client or the restored state during the gap and assert the arm still sees none; (2) the repo already has relay-restart.test.ts, so share the boot and sit helpers instead of keeping a second copy.
Issue #235 is done and nothing has changed since last time, so I made no new changes. The branch is pushed, the working tree is clean, and
git pushreports it up to date. I checked the issue through the API again, and the only comment is still the crew's note that pull #314 was closed after its preview deploy failed.Three commits on the branch are ahead of
main:4580bc5addsSoftware/so101/src/turn-restart.test.ts. It starts the relay on a temporaryFORGE_APP_DATAwith a turn running and two people queued. It stops the relay and restarts it on the same data. It then checks that the driver reconnects and drives until the original end time, that the queue order is unchanged, and that the fake arm saw no move in the gap.96b9718is a timing fix tosrc/stop.test.ts, which was flaky under load. It is unrelated to #235; it may be what failed the earlier preview, but I couldn't see that log.5c93701mergesmain.forge.jsonis untouched. I have not run the fullbun --cwd Software/so101 testortsc --noEmit, becausetsccan't write to its temp directory in this sandbox. The forge's check on the push will be the first full run.Closes #235