Array item removal binds a surviving sibling's sub-Field to a stale, out-of-range index for one render (stable/non-index keys)
維護者通常 1 天內回覆
還沒有人認領這個 Issue。
評估
- 難度
- 4/5
- 預估耗時
- 3-5 天
- 新手友好度
- 52/100
- Issue 類型
- 缺陷
- 描述清晰度
- 基本清楚
- 活躍度
- 冷清
- 技術堆疊
- react, typescript
- 領域
- frontend
研究方向
從 @tanstack/form-core 中的陣列變更與欄位路徑推導開始,使用此處描述的 React AppField 重現、穩定的 key 以及 form.getFieldValue API。重現移除 options 的第一個項目,接著追蹤為什麼保留下來的 sub-Fields 仍保留 options[1];完成條件是它們使用位移後的路徑,且不會出現短暫的 undefined 值。
由索引模型根據 Issue 內容生成。
描述
[Opus-written, can add human replies if something's not clear or helpful!]
First off, thanks for all the work on TanStack Form. It's been a joy to build with, and this is a pretty narrow edge case rather than anything blocking us. Wanted to write it up carefully in case it helps.
Summary
When you remove a non-last item from an array field, a surviving sibling's sub-Fields re-render for one frame with their name still pointing at the old, now out-of-range index. During that single frame both field.state.value and form.getFieldValue(field.name) come back undefined, even though the field's value type is non-optional (for example a plain string). On the very next render everything reconciles to the correct index and is fine.
This shows up when the array item containers use stable (non-index) React keys. We need stable keys for drag-to-reorder and view transitions, so switching to index keys isn't really an option for us.
I can confirm it still reproduces on @tanstack/[email protected], which is a good number of releases after the earliest related report.
How this relates to existing issues
I dug through the tracker first, and two closed issues are in the neighborhood:
- #1363 ("Non-index keys in React causing inputs to be uncontrolled when removing array fields") looks like the same root cause. With non-index keys, removing an item leaves a sibling's field
name/path pointing at a stale index, the value reads asundefinedfor a render, and React flips the input to uncontrolled. It was last reproduced on 1.2.3 and closed without a fix landing, so I figured a fresh writeup on a current version might be useful. - #1518 is close but subtly different. There,
form.getFieldValue(field.name)still returns the correct value whilestate.valueis transientlyundefined. In our case that workaround doesn't hold, because the field'snameitself is stale, so the store lookup comes back empty too. The lag is in the field's path, not just its cached value.
Happy to be pointed at whichever of these you'd rather consolidate into if you feel it's a duplicate.
Versions
@tanstack/react-form1.28.3,@tanstack/form-core1.28.x- React 19, Next.js 16 (App Router)
Reproduction
- An array field
options, where each item is{ name: string, values: {...}[] }. - The item containers are keyed by a stable key (
key={opt.name}), because rows are drag-reorderable (dnd-kit) and animated with<ViewTransition>, both of which want stable identity. - Each row renders its sub-fields via
form.AppField name={options[${idx}].name}and similar, whereidxis the.map()index passed down as a prop.
Steps:
- Start with two options, say
Tamaño(index 0) andColor(index 1). - Remove the first one (
Tamaño, index 0). The array becomes length 1, andColorshould shift to index 0. - The surviving
Colorrow's sub-fields re-render once.
What I see
On that re-render, some temporary tracing logged (formatted for readability):
OptionHeader.name transient { fieldName: "options[1].name", stateValue: null, storeValue: null }
OptionBody.values transient { fieldName: "options[1].values", stateValue: null, storeValue: null }
fieldNameis stilloptions[1], the old index, even though the array now only hasoptions[0].field.state.valueisundefined.form.getFieldValue(field.name)is alsoundefined, becauseoptions[1]no longer exists.
The next render reconciles to options[0] and it's all correct.
What I'd expect
A sub-Field of an array item shouldn't render bound to an index that no longer exists after a sibling is removed, and a field whose value type is non-optional shouldn't transiently expose undefined.
What we're doing for now
We early-return when the value is missing, since the field is about to reconcile or unmount anyway:
<form.AppField name={`options[${optionIdx}].name`}>
{(nameField) => {
// Transient undefined guard, TanStack Form bug
// oxlint-disable-next-line typescript/no-unnecessary-condition
if (nameField.state.value === undefined) return null
// ...
}}
</form.AppField>
That avoids the crash and the final render is correct. The tradeoff is that the row renders nothing for one frame, and the typed value is briefly a small lie.
So
Would it be feasible for the array re-index to update a surviving item's sub-Field name/value in step with the array mutation (or to skip rendering removed or shifted paths), so that stable-keyed array items stay type-correct without forcing index keys? I know that's easier said than done given how derivation works, so no pressure. I'm glad to put together a minimal StackBlitz on 1.28.x if that would help you dig in.
- 主要語言
- TypeScript
- 星號
- 6.8k
- 分支
- 717
- 平均合併
- 5 天 22 小時
- 30 天內合併 PR
- 14
環境準備
- 沒有 Dockerfile 或 Docker Compose 檔案
- 有 Pull Request 範本
- 閱讀貢獻指南
從這裡開始
- 先讀完整個 Issue,再讀專案的貢獻指南。
- 在 Issue 下留言說明你要接手 —— 這能避免兩個人做同樣的事。
- Fork 儲存庫,在一個分支上完成修改。
- 送出 Pull Request,並在描述裡引用這個 Issue 編號。
TanStack/form 的其他 Issue
-
難度 1/5 1 小時以內 新手友好度 88/100
維護者通常 1 天內回覆
-
Submit listener exceptions leave isSubmitting stuck true可能已有人在做 @henricop 於 15 天前認領。 未關閉
難度 2/5 1-3 小時 新手友好度 72/100
維護者通常 1 天內回覆
-
handleSubmit calls onSubmit again while a submit is still pending (since 1.33.1)可能已有人在做 @twotnguyen 於 14 天前認領。 未關閉
難度 2/5 1-3 小時 新手友好度 78/100
維護者通常 1 天內回覆
-
`isDefaultValue` is wrong when a field-level `defaultValue` is `null`可能已有人在做 @fenet-as 於 40 天前認領。 未關閉
難度 2/5 1-3 小時 新手友好度 76/100
維護者通常 1 天內回覆
-
Async validator doesn't re-trigger `isValidating` after first failed validation可能已有人在做 @drewsephski 於 44 天前認領。 未關閉
難度 2/5 1-2 天 新手友好度 72/100
維護者通常 1 天內回覆
相似的 Issue
-
clawsweeper:needs-maintainer-review clawsweeper:needs-product-decision clawsweeper:no-new-fix-pr clawsweeper:source-repro impact:ux-friction issue-rating: 🦞 diamond lobster P3
難度 2/5 1-3 小時 新手友好度 76/100
openclaw/openclaw#203385 · 1 則留言 · 1 個 reaction ·
維護者通常 1 天內回覆
-
難度 2/5 1-3 小時 新手友好度 72/100
nwverse/persona#12800 · 1 則留言 ·
維護者通常 1 天內回覆
-
難度 2/5 1-3 小時 新手友好度 80/100
維護者通常 1 天內回覆
-
[MCP]: page snapshots in .playwright-mcp/ are not git-ignored (.playwright-cli/ is, since #42318)未關閉
難度 2/5 1-3 小時 新手友好度 62/100
microsoft/playwright#44289 ·
維護者通常 1 天內回覆
-
good first issue
難度 2/5 半天 新手友好度 65/100
維護者通常 1 天內回覆