[toast] Enable closing all toasts

[chuganzy]

This PR adds a way to close all toasts at once.

The close API now accepts an optional id:

• close(id) closes a specific toast
• close() closes all active toasts

Alternatively, I considered introducing a separate closeAll API, but decided to keep the change minimal for the following reasons:

• Adding closeAll could require changes to ToastManagerEvent used by subscribe. I was not sure whether it is a public API, and making changes to the ToastManagerEvent might be considered a breaking change (though that can be decoupled).

• Other libraries such as Sonner and Chakra expose the same behavior through a single dismiss API, where calling it without an argument dismisses all toasts.

• I have followed (at least) the PR section of the contributing guide.

[pkg-pr-new]

• base-ui-tanstack-start

• vite-css-base-ui-example

  pnpm add https://pkg.pr.new/mui/base-ui/@base-ui/react@3979
  

  pnpm add https://pkg.pr.new/mui/base-ui/@base-ui/utils@3979
  

commit: 473da83

[netlify]

✅ Deploy Preview for base-ui ready!

Name	Link
🔨 Latest commit	2163762
🔍 Latest deploy log	https://app.netlify.com/projects/base-ui/deploys/69aaa94eb904360008b78758
😎 Deploy Preview	https://deploy-preview-3979--base-ui.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.

---

To edit notification comments on pull requests, go to your Netlify project configuration.

[greptile-apps]

Greptile Overview

Greptile Summary

This PR extends the toast close API to support dismissing all active toasts by calling close() with no ID (while preserving close(id) for a specific toast). The change is wired through createToastManager → ToastProvider subscription → ToastStore.closeToast, with added tests for both the global manager and the useToastManager hook.

Two issues to address before merge:

• The Toast docs are now inconsistent: the new “close all toasts” example was added, but the useToastManager return-value table still documents close as requiring a string argument.
• ToastStore.closeToast() now calls onClose for all toasts during close-all even if some toasts are already transitionStatus: 'ending', which can cause duplicate onClose callbacks compared to the previous single-id close behavior.

Confidence Score: 4/5

• Generally safe to merge after fixing a couple behavioral/docs issues around close-all semantics.
• Core API/type wiring and tests for close-all look consistent across manager, provider, and hook. Remaining concerns are a concrete docs/type mismatch and a likely behavior regression where onClose can fire for toasts already ending during close-all.
• packages/react/src/toast/store.ts, docs/src/app/(docs)/react/components/toast/page.mdx

Important Files Changed

Filename	Overview
docs/reference/generated/toast-provider.json	Updates generated ToastProvider prop type so toastManager.close accepts an optional id (string
docs/src/app/(docs)/react/components/toast/page.mdx	Adds docs example for toastManager.close() closing all toasts, but the return-value table still lists close as requiring a string id.
packages/react/src/toast/createToastManager.test.tsx	Adds test for close() without id to dismiss all toasts via global manager; test asserts empty state via queryByTestId.
packages/react/src/toast/createToastManager.ts	Makes ToastManager.close accept an optional id and emits a close event with options.id possibly undefined.
packages/react/src/toast/provider/ToastProvider.tsx	Updates toastManager subscription handling so action 'close' always calls store.closeToast(id), enabling close-all when id is undefined.
packages/react/src/toast/store.ts	Implements store.closeToast(toastId?) with close-all semantics and updates focus/timer cleanup; potential behavior mismatch for onClose callbacks when closing a toast already ending.
packages/react/src/toast/useToastManager.test.tsx	Adds test for useToastManager().close() without id removing all toasts; covers hook API surface for close-all.
packages/react/src/toast/useToastManager.ts	Updates UseToastManagerReturnValue.close signature to accept an optional toastId to match new close-all behavior.

[atomiks]

I'm okay with this change, but it will conflict a lot with #3464, not sure whether to merge it before or after @flaviendelangle

[mui-bot]

Bundle size report

Bundle	Parsed size	Gzip size
@base-ui/react	🔺+277B(+0.06%)	🔺+79B(+0.05%)

Details of bundle changes

---

Check out the code infra dashboard for more information about this PR.

[chuganzy]

Given that this PR is significantly smaller than the other one, I think it may make sense for me to wait 😅
I can take care of resolving the conflicts on my end as needed 🙌

[flaviendelangle]

As you want @atomiks, I resolved all your feedback on #3464 👍

[greptile-apps]

_{8 files reviewed, 2 comments}

_{Edit Code Review Agent Settings | Greptile}

[greptile-apps]

onClose fired too often

closeToast now invokes onClose for all toasts before checking whether each toast is already ending (transitionStatus === 'ending') (store.ts:233-240 vs mapping logic at 247-249). This means calling close() (close-all) can trigger onClose for toasts that were already in the process of closing (e.g. auto-dismiss just started), which differs from the previous behavior where onClose was only called for the explicitly targeted toast ID and only when that toast existed. Consider aligning onClose invocation with the same predicate used to mark a toast as ending (skip ones already ending) to avoid duplicate onClose calls.

Prompt To Fix With AI

This is a comment left during a code review.
Path: packages/react/src/toast/store.ts
Line: 229:232

Comment:
**onClose fired too often**

`closeToast` now invokes `onClose` for *all* toasts before checking whether each toast is already ending (`transitionStatus === 'ending'`) (store.ts:233-240 vs mapping logic at 247-249). This means calling `close()` (close-all) can trigger `onClose` for toasts that were already in the process of closing (e.g. auto-dismiss just started), which differs from the previous behavior where `onClose` was only called for the explicitly targeted toast ID and only when that toast existed. Consider aligning `onClose` invocation with the same predicate used to mark a toast as ending (skip ones already ending) to avoid duplicate `onClose` calls.

How can I resolve this? If you propose a fix, please make it concise.

[chuganzy]

It is true. This can also happen with single-close as well.
Since this is an existing issue, I will create a separate PR to address it.

[greptile-apps]

Additional Comments (1)

docs/src/app/(docs)/react/components/toast/page.mdx
Docs type mismatch

The useToastManager return-value table still documents close as (toastId: string) => void (page.mdx:283-286), but the implementation now allows close() with no argument to close all toasts. This makes the docs internally inconsistent and will mislead TypeScript users reading the docs; the table should reflect the new optional parameter (e.g. close: (toastId?: string) => void or similar).

Prompt To Fix With AI

This is a comment left during a code review.
Path: docs/src/app/(docs)/react/components/toast/page.mdx
Line: 282:285

Comment:
**Docs type mismatch**

The `useToastManager` return-value table still documents `close` as `(toastId: string) => void` (page.mdx:283-286), but the implementation now allows `close()` with no argument to close all toasts. This makes the docs internally inconsistent and will mislead TypeScript users reading the docs; the table should reflect the new optional parameter (e.g. `close: (toastId?: string) => void` or similar).

How can I resolve this? If you propose a fix, please make it concise.

[chuganzy]

@atomiks

Quick ping on this PR 🙂
CI is failing, but it looks like it's during the initial repo clone rather than the tests, so it might not be related to this PR. Could you take a look when you have a chance? Thanks!

[chuganzy]

Uh-oh.. Prettier rule has changed. I will re-format and push the change soon!

[atomiks]

Codex Review

Overview

This patch adds support for dismissing every toast by calling close() without an ID, and wires that behavior through the manager, provider, store, docs, and tests. The implementation looks consistent overall, and I did not find a new correctness issue introduced by the API change.

Findings

No blocking issues found in this patch.

1. 🟡 [Non-blocking] Close-all exposes an existing duplicate onClose edge case

Impact: onClose is already not idempotent if the same toast is closed twice. This patch makes that pre-existing behavior easier to hit because close() with no ID also iterates over toasts that are already ending, so consumers may now encounter the duplicate callback through the new public API.

Evidence: packages/react/src/toast/store.ts:245-248 calls onClose for every toast in the close-all branch before packages/react/src/toast/store.ts:265-269 leaves already-ending toasts in place. I verified that master already double-calls onClose for closeToast(id) -> closeToast(id), and this branch extends the same behavior to closeToast(id) -> closeToast().

Recommendation: Consider handling this in a follow-up by only invoking onClose when a toast first transitions to 'ending', and add a regression test that covers both repeated single-close and single-close-then-close-all flows.

Confidence: 4/5

Reviewed the full branch diff against current master and ran the focused createToastManager, useToastManager, and Chromium toast suites. The API, docs, and test wiring look aligned; the remaining concern is limited to the pre-existing callback edge case above.

Notes

• Review scope covered the full branch diff against current master, not just the latest commit.
• The duplicate onClose behavior appears to predate this PR; this patch widens its reach rather than introducing it from scratch.

[chuganzy]

@atomiks As always, thanks for jumping in and taking care of this so quickly. Much appreciated!