Flaky: SheetHot.test.jsx "saves a quantity and a removal with the sheet" drops the removal under load #1221

New issue

Closed

opened 2026-10-17 21:27:51 +00:00 by claude-bot · 1 comment

claude-bot commented

2026-10-17 21:27:51 +00:00

Collaborator

Seen on 17 Oct 2026 in the v4.6.0-rc2 release PR's run (13481, Frontend shard 1/2). The same commit's code passed on main's run (13474).

src/pages/sheet/SheetHot.test.jsx:634, "a row's own fields, edited › saves a quantity and a removal with the sheet, on its version". The test took 3,750 ms. The saved body had update: [{… qty: 40}] and no remove: ["inv-rope"].

That suggests the test asserts before the removal has joined the pending save, a race between the quantity edit's debounce and the remove, which a slow runner exposes. It could also be a real ordering bug in how the sheet batches a removal made just after an edit. That would matter, since a removal silently not saved would be data a player thinks they changed. Worth checking which before calling it a flake. Nothing in 4.6.0 touched the sheet.

Seen on 17 Oct 2026 in the v4.6.0-rc2 release PR's run (13481, Frontend shard 1/2). The same commit's code passed on main's run (13474). `src/pages/sheet/SheetHot.test.jsx:634`, "a row's own fields, edited › saves a quantity and a removal with the sheet, on its version". The test took 3,750 ms. The saved body had `update: [{… qty: 40}]` and **no** `remove: ["inv-rope"]`. That suggests the test asserts before the removal has joined the pending save, a race between the quantity edit's debounce and the remove, which a slow runner exposes. It could also be a real ordering bug in how the sheet batches a removal made just after an edit. That would matter, since a removal silently not saved would be data a player thinks they changed. Worth checking which before calling it a flake. Nothing in 4.6.0 touched the sheet.

claude-bot added this to the v4.7.0 - Browse the Rules milestone

2026-10-27 17:38:55 +00:00

claude-bot commented

2026-10-30 19:23:04 +00:00

Author

Collaborator

Picked up for v4.7.0 (2026-10-30), on branch fix/1221-sheet-removal-save. First: decide whether the test asserts too early, or whether a removal made just after an edit is really lost. The second would be data loss for a player, fixed in the component with a deterministic regression test.

Picked up for v4.7.0 (2026-10-30), on branch `fix/1221-sheet-removal-save`. First: decide whether the test asserts too early, or whether a removal made just after an edit is really lost. The second would be data loss for a player, fixed in the component with a deterministic regression test.

claude-bot referenced this issue from a pull request that will close it,

2026-10-30 19:31:59 +00:00

test(webapp): a quantity and a removal may go in two saves (#1221) #1290