The table update/duplicate steps awaited waitForDatabaseToLoad (a
public-infinite_tables waitForResponse) registered right after the mutation —
the same too-late-listener race that flaked the schema-visualizer test under
TanStack's faster render. The toBeVisible assertion immediately after each one
already waits for the list to refetch, so the waits are redundant. Removes
both and the now-unused import. Verified locally (x2) against the TanStack build.
database.spec ':24' also switched to the auth schema mid-test and called
waitForSchemaVisualizerToLoad(...'auth') *after* the schema-selector click —
a waitForResponse registered too late for the (SSR-streamed under TanStack)
auth-infinite_tables request, so it raced and flaked. The focusTableInVisualizer
loop right below already auto-waits for the auth tables to load, so drop the
redundant wait (and its now-unused import).
- database.spec: the 3 Schema Visualizer/Tables tests waited on the
'public-infinite_tables' pg-meta XHR after page.goto. Under the TanStack
build that data arrives via the SSR stream, so the client request never
fires and waitForResponse times out (flaky by shard). Gate on
focusTableInVisualizer (auto-waits for the schema incl. the new table).
- prettier: wrap the long createFileRoute() path in the generated
vector-buckets indexes [indexName] route.
Tanstack E2E failures were caused by:
- 12 Next pages/api routes never ported to TanStack routes/api (they 404'd
under the Vite build). Notably /api/platform/deployment-mode, which made
the Connect sheet's direct-connection step render a skeleton with no
connection-method radios (connect.spec 'pre-selects direct connection
method'). Also ports vector-buckets, signing-keys, api-keys/[id],
get-s3-keys, parse-query, parse-client-code, secrets/update-status.
- realtime-inspector beforeEach registered waitForResponse(/settings/) AFTER
navigation already waited on the UI; settings is a one-shot non-refetching
request that, under TanStack's faster render, resolves before the listener
attaches, hanging every realtime test to the 120s timeout. Removed the
redundant wait (the UI assertion is the real gate).
- 'clicking broadcast message shows detail panel' center-clicked a data-grid
row whose untruncated-JSON cell overflows under the detail panel; click the
row's visible timestamp instead.
## Context
Dashboard currently doesn't have any support for managing stored
procedures. In the event that the security advisor surfaces a warning
about a stored procedure, users hence run into a dead-end as there's
currently no way to self-remediate via the dashboard
## Changes involved
We're hence adding support for managing stored procedures within
Database Functions
<img width="1082" height="546" alt="image"
src="https://github.com/user-attachments/assets/2598a5fe-e58f-4e8a-ad2f-9cb6d0eb2f53"
/>
Creating a function now shows a dropdown to select the type
<img width="500" alt="image"
src="https://github.com/user-attachments/assets/acc9249d-7b25-4416-aae8-89c630e1c62b"
/>
In which if stored procedure is selected, the following fields will be
hidden since they're irrelevant for stored procedures
- Return type
- Behaviour (Under advanced settings)
Some other minor UI changes as well:
- Field inputs are re-ordered a little, opting to group "Schema" and
"Name" into one section, followed by "Type" and "Return type"
- Opting to show "Return type" when editing a function but disabled
- Add schema filter for fetching database functions to reduce
unnecessary load on the database
## To test
- [ ] Can create, update, delete, read stored procedures via database
functions page
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
## Summary
- **New Features**
- Added PostgreSQL **procedure** support alongside functions, including
a **Type** selector in the create/edit flow.
- Updated Functions UI with a new **Type** column and procedure-aware
return/argument details.
- **Improvements**
- Refreshed create/edit headers and language help text for clearer
context.
- Improved argument parsing/display, including better handling of
procedure argument modes.
- **Bug Fixes**
- Corrected routine-type handling during function/procedure delete and
update SQL operations.
- **Tests**
- Updated unit snapshots and end-to-end UI flows/labels for the new “New
function” control.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
Visual cleanup of the Logs sidebar (unified logs):
- Restyle the "Go back to old logs" and "Try Unified Logs" banners as
full-bleed bordered strips (no card), with an icon-only switch-back
button and a tooltip
- Add a "Beta" badge next to the "Logs" sidebar title
- Rename the product from "Logs & Analytics" to "Logs"
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Added Unified Logs banner with “Try Unified Logs” and “More
information” actions, plus a “Go back to old logs” utility layout.
* Show a **“Beta”** badge for Unified Logs eligibility.
* **UX Updates**
* Updated the Logs experience to consistently display under the
**“Logs”** product layout.
* **Tests / Documentation**
* Updated end-to-end checks and examples to expect the **“Logs”**
heading text exactly.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Co-authored-by: kemal <hello@kemal.earth>
## I have read the
[CONTRIBUTING.md](https://github.com/supabase/supabase/blob/master/CONTRIBUTING.md)
file.
YES
## What kind of change does this PR introduce?
Refactored the Database Create Index side panel to align with the
recommended Studio form patterns.
## What is the current behavior?
The side panel was using a legacy layout that did not follow the common
Studio form patterns.
It's also not a real form.
## What is the new behavior?
The Create Index side panel in the Database section is now consistent
with the Studio form patterns.
It's now a real form using `react-hook-form`
## Additional context
Before:
<img width="668" height="855" alt="image"
src="https://github.com/user-attachments/assets/d15ee63c-8d03-4780-8945-1792f0493133"
/>
After:
<img width="606" height="856" alt="image"
src="https://github.com/user-attachments/assets/78e0269c-e79c-42d1-a2c2-9465c0b0c46a"
/>
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **UI Improvements**
* Reworked index-creation panel into a validated, form-driven workflow
with clearer field interactions, multi-column selector, live read-only
SQL preview, and updated submit/cancel behavior.
* Footer actions: Cancel resets the form and closes the panel; Create
triggers form submission and reflects loading state.
* **Tests**
* Updated end-to-end flow to click "Select a table" (replacing "Choose a
table") during index creation.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Co-authored-by: Gildas Garcia <1122076+djhi@users.noreply.github.com>
## I have read the
[CONTRIBUTING.md](https://github.com/supabase/supabase/blob/master/CONTRIBUTING.md)
file.
YES
## What kind of change does this PR introduce?
Feature / performance improvement.
## What is the current behavior?
The Database > Tables list renders every table for the selected schema
in a single `useTablesQuery({ includeColumns: true })` fetch.
## What is the new behavior?
- Database > Tables list uses `useInfiniteTablesQuery` with a 50-row
page size, streaming pages as the user scrolls.
- An `IntersectionObserver` sentinel attached to the footer status row
triggers `fetchNextPage()` while `hasNextPage` is true; the same row
doubles as the table count / loading indicator.
- Search is debounced (300 ms) and passed as `nameFilter` to the hook,
which forwards it to pg-meta's `getTablesPaginatedSql`. The backend
returns only matching rows, so search works across the whole schema
rather than only the loaded pages.
## Additional context
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Debounced search filtering for tables
* Infinite-scroll pagination with sentinel-driven loading
* Footer shows “Loading more tables…” and dynamic table counts
* **Bug Fixes**
* Ensure table list refresh after create/duplicate by invalidating
infinite-list cache
* **Tests**
* Updated end-to-end waits and helpers to match the new paginated
table-loading API
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
## I have read the
[CONTRIBUTING.md](https://github.com/supabase/supabase/blob/master/CONTRIBUTING.md)
file.
YES/NO
## What kind of change does this PR introduce?
Bug fix, feature, docs update, ...
## What is the current behavior?
Please link any relevant issues here.
## What is the new behavior?
Feel free to include screenshots if it includes visual changes.
## Additional context
Add any other context or screenshots.
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **Accessibility**
* Improved screen-reader label for table row action menus so table
controls are clearer for assistive‑technology users.
* **Tests**
* Enhanced end-to-end test reliability: tightened selectors, added
dialog/toast visibility and API-wait synchronization, scoped lookup
fixes, removed redundant cleanup helper, and updated test setup to mark
a terms-of-service dismissal to reduce flakiness.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
## I have read the
[CONTRIBUTING.md](https://github.com/supabase/supabase/blob/master/CONTRIBUTING.md)
file.
YES
## What kind of change does this PR introduce?
Performance improvement / feature
## What is the current behavior?
The Schema Designer fetches all tables in a single request via
`useTablesQuery`. For schemas with 400+ tables this blocks first paint
on a large payload.
## What is the new behavior?
`SchemaGraph` uses `useInfiniteTablesQuery` (pageSize: 100) so the first
100 tables paint immediately. A "Load more tables" button appears above
the legend whenever more pages remain, letting users load the rest on
demand.
## Additional context
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Added a "Find table…" selector and keyboard shortcut to quickly locate
and focus tables; supports incremental loading and debounced name search
(with literal wildcard handling).
* Schema Graph shows a bottom "Load more tables" control with loading
state and preserves view after loading more.
* **Refactor**
* Table listing switched to infinite/paginated retrieval and improved
"no tables" logic; server-side name filtering supported.
* **Tests**
* E2E tests add a schema-visualizer wait helper and update flows to
support the paginated visualizer.
<!-- review_stack_entry_start -->
[](https://app.coderabbit.ai/change-stack/supabase/supabase/pull/46402?utm_source=github_walkthrough&utm_medium=github&utm_campaign=change_stack)
<!-- review_stack_entry_end -->
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
## I have read the
[CONTRIBUTING.md](https://github.com/supabase/supabase/blob/master/CONTRIBUTING.md)
file.
YES
## What kind of change does this PR introduce?
- Minor issues here, the validation for creating names is there but
users can create crons with empty names through SQL
- When they edit the name in the Cron editor, since we use names as the
where clause it treats it as a new create
- So a duplicate cron is created
- Since creating requires a name, the validation is moved to the
component rather than zod and disabled when editing mode is on!
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Cron jobs can now be created without requiring a name field.
* Improved handling to properly distinguish between creating new cron
jobs and editing existing ones.
* **Bug Fixes**
* Fixed issue where editing unnamed cron jobs would create duplicate
entries instead of updating the existing job in place.
<!-- review_stack_entry_start -->
[](https://app.coderabbit.ai/change-stack/supabase/supabase/pull/46486?utm_source=github_walkthrough&utm_medium=github&utm_campaign=change_stack)
<!-- review_stack_entry_end -->
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
## I have read the
[CONTRIBUTING.md](https://github.com/supabase/supabase/blob/master/CONTRIBUTING.md)
file.
YES
## What kind of change does this PR introduce?
Right now material views and views don't have any options on the context
menu, they only have a copy name. This adds copy schema, export CSV,
export SQL and delete table to that list
Added E2E tests to cover the use cases
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Delete views and materialized views via confirmation dialogs with
optional cascade
* Copy SQL definitions for views and materialized views
* Export views and materialized views as CSV and SQL from the entity
menu
* Confirmation modals now show dependency warnings and cascade toggle
consistently
* **Tests**
* End-to-end tests covering copy, export, and delete flows for views and
materialized views in the table editor
<!-- review_stack_entry_start -->
[](https://app.coderabbit.ai/change-stack/supabase/supabase/pull/46383?utm_source=github_walkthrough&utm_medium=github&utm_campaign=change_stack)
<!-- review_stack_entry_end -->
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
Two changes that unbreak the spec in Next.js mode:
- `navigateToQueuesPage` now waits for the "Create queue" header
button instead of `getByRole('grid')`. The grid only renders when
at least one queue exists; an empty queue list shows an
empty-state placeholder instead, so the previous signal could
fail at navigation time depending on prior test state. Under
workers=3 + Next.js the page also doesn't always populate the
list fast enough before the 30s timeout. The Create queue button
is in the page header and renders unconditionally.
- The "can create a (new|unlogged) queue" post-create assertion
now polls for "queue exists in UI" via either the queue-detail
URL or the row in the list. The previous shape locked in the
TanStack-mode signal (router.push races nuqs and the URL stays
on the list), which inverted in Next.js mode where router.push
reliably navigates away and the row is no longer visible. The
poll accepts whichever lands first.
Verified locally in Next mode with workers=3 (the default for
self-hosted): 7/7 pass. TanStack mode also stays green.
<img width="1285" height="654" alt="image"
src="https://github.com/user-attachments/assets/10074f07-3ac0-4f1f-b920-4255c67968d6"
/>
Updates file storage buckets to make use of new page components and
follow more standard full width pages as recommended in our design
system.
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Search input now always visible with dynamic placeholder and cancel
control.
* “Go up one level” navigation button shown in list view when not at
bucket root.
* **Style**
* Thinner icon strokes and smaller button/icon sizing across the Storage
Explorer and preview.
* Header, breadcrumbs, and page container spacing/layout refined;
explorer outer container simplified.
* Selection count UI made compact/monospaced; new layout tokens for
header and bulk-action rows.
* **Tests**
* Unit and e2e tests updated; added e2e helper for rename/create inputs.
<!-- review_stack_entry_start -->
[](https://app.coderabbit.ai/change-stack/supabase/supabase/pull/46184?utm_source=github_walkthrough&utm_medium=github&utm_campaign=change_stack)
<!-- review_stack_entry_end -->
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
## What kind of change does this PR introduce?
Chore: component migration and copy update.
## What is the current behavior?
`DownloadSnippetModal` and `RenameQueryModal` use an awkward tabs-based
layout. The download action is labelled "Download" throughout, and the
NPX toggle was incorrectly labelled "NPM".
## What is the new behavior?
- Both modals are migrated to the `Dialog` component.
- The download dialog replaces the tab layout with a select (Migration /
Seed file / SQL file) and a CLI/NPX toggle.
- Action language changed from "Download" to "Export" (context menu
item, dialog title, select label) — more accurate since the user runs a
CLI command rather than triggering a browser download.
- NPM toggle label corrected to NPX; internal `npm` property key renamed
to `npx` for consistency.
| Before | After |
| --- | --- |
| <img width="1024" height="759" alt="Add Auth Hook (General) SQL Editor
Pickles Pantry Supabase-C6F22F8B-19FF-486E-8C08-915895495875"
src="https://github.com/user-attachments/assets/98637802-5e05-4431-87f8-b2e83216082a"
/> | <img width="1024" height="759" alt="Add Auth Hook (General) SQL
Editor Pickles Pantry Supabase-5B63AD0C-7CD0-4BB2-BEC8-5DBAE94963CE"
src="https://github.com/user-attachments/assets/a8f48c2f-a7e2-42fd-b52c-89a133811ad8"
/> |
| <img width="1024" height="759" alt="Add Auth Hook (General) SQL Editor
Pickles Pantry Supabase-1F28FADA-46D1-4C0A-BE96-6CFF2317FDCF"
src="https://github.com/user-attachments/assets/afb904d0-689b-4756-b0b4-8177703934e4"
/> | <img width="1024" height="759" alt="Add Auth Hook (General) SQL
Editor Pickles Pantry Supabase-652E6E8D-1CD2-4465-BC82-46EDB974CD8C"
src="https://github.com/user-attachments/assets/f1adfb6b-fba6-495f-b571-1713c3eebb4d"
/> |
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **Refactor**
* Updated the SQL query export modal UI from a tabs-based interface to a
dropdown-based selection for choosing export formats.
* Revised menu and dialog labels to "Export query" for improved clarity.
* Enhanced code block presentation in the export modal.
<!-- review_stack_entry_start -->
[](https://app.coderabbit.ai/change-stack/supabase/supabase/pull/46242?utm_source=github_walkthrough&utm_medium=github&utm_campaign=change_stack)
<!-- review_stack_entry_end -->
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Co-authored-by: Gildas Garcia <1122076+djhi@users.noreply.github.com>
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
## Summary
- Closes
[FE-3399](https://linear.app/supabase/issue/FE-3399/make-table-editor-filter-bar-wait-for-a-value)
- Table editor's filter bar was firing a row request as soon as a
property was selected, before any operator/value was set. This regressed
after #46071 unified the table editor and unified logs onto the shared
`FilterBar`.
- `LogsFilterBar` already validates that each condition has a
propertyName, operator, and non-empty value before applying.
`FilterPopoverNew.handleApply` did not — it called `setFilters`
unconditionally on every `onApply` from the shared bar (property change,
operator change, blur, etc.). This PR mirrors the `LogsFilterBar`
validation so the table editor stops requesting rows until the condition
is complete.
## Test plan
- [ ] Open the table editor on any table
- [ ] Click the filter bar, pick a column → no network request fires, no
toast, grid unchanged
- [ ] Pick an operator → still no request
- [ ] Type a value and hit Enter → row request fires, grid filters
- [ ] Remove the filter → grid refetches with no filter
- [ ] New e2e test: `pnpm --prefix e2e/studio run e2e --
features/filter-bar.spec.ts --grep "does not trigger a row request"`
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **Bug Fixes**
* Filter validation now prevents incomplete filter conditions from being
applied, ensuring only fully-specified filters affect table data.
* **Tests**
* Added end-to-end test coverage for filter bar operations to verify
expected behavior during filter setup.
<!-- review_stack_entry_start -->
[](https://app.coderabbit.ai/change-stack/supabase/supabase/pull/46113?utm_source=github_walkthrough&utm_medium=github&utm_campaign=change_stack)
<!-- review_stack_entry_end -->
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
Backports a batch of e2e test stabilization fixes — each commit is
scoped to a single failure class and only touches `e2e/studio/` files.
**Changed:**
- **`_global.setup` — playwright-locks cleanup was dead code**: the
lock-cleanup block was at the bottom of `Global Setup`, but every branch
above it returns early — so it never ran. Tests that use
`withFileOnceSetup` (cron-jobs) would see a stale `setup.done.json`
marker from the previous run and silently skip their setup, leaving e.g.
`pg_cron` uninstalled and all 11 cron-jobs specs failing. Moved the
cleanup to before any early return.
- **filter-bar — Home key**: macOS Chromium doesn't honor a standalone
`Home` keypress inside text inputs (macOS routes "go to line start" via
`Cmd+ArrowLeft` / `Fn+ArrowLeft`). Tests that expected the cursor to
jump to position 0 silently kept the previous selection. Replaced with
`el.setSelectionRange(0, 0)` so the assertion runs against a known
cursor position on every OS.
- **filter-bar — date filters**: tests inserted rows with `CURRENT_DATE`
/ `NOW()` (postgres session TZ = UTC) and asserted with JS-local dates
from `getDateValue()`. Near midnight the two diverged and the filter
returned 0 rows. Switched the inserts to explicit `getDateValue()`
strings so insert and assert use the same calendar day.
- **queue-table-operations — `networkidle`**: Studio holds long-poll /
SSE connections (PostHog, realtime), so `page.reload({ waitUntil:
'networkidle' })` never resolves and timed out. Replaced with a targeted
`waitForTableToLoad` API waiter.
- **sql-editor — RLS smoke test**: a hard-coded table name
(`pw_rls_smoke_test`) collided across 3 parallel workers running against
the same db. Suffixed with `test.info().parallelIndex`.
- **table-editor — FK spec timeout**: `waitForApiResponseWithTimeout`
for `query?key=table-update` returns `null` on timeout (silent), then
the panel-close assertion fails. Bumped 15s → 30s to absorb
parallel-load latency.
- **table-editor / storage-helpers — URL encoding & redirect race**:
post-action URL assertions were over-specific (`%20` vs `+` encoding)
and the bucket-delete redirect could race other history updates. Relaxed
the regex to accept both encodings; asserting the row removal directly
is a more stable signal than the redirect URL.
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **Tests**
* Improved end-to-end determinism with explicit dates/timestamps and
stable cursor positioning
* Prevented parallel-test collisions by using unique identifiers for
resources
* Made page reloads and API waits more robust for long-lived connections
and increased timeouts
* Strengthened assertions to rely on stable UI signals instead of
transient navigation/network state
* Ensured test setup reliably cleans up temporary locks before any setup
steps run
<!-- review_stack_entry_start -->
[](https://app.coderabbit.ai/change-stack/supabase/supabase/pull/46039?utm_source=github_walkthrough&utm_medium=github&utm_campaign=change_stack)
<!-- review_stack_entry_end -->
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Co-authored-by: Alaister Young <10985857+alaister@users.noreply.github.com>
The "create and remove foreign key with column selection" spec wrapped
`table-update` API waits at 15s. Under parallel load (3 self-hosted
workers hitting the same db) the request occasionally crosses that
budget; `waitForApiResponseWithTimeout` silently returns `null`, the
panel-close assertion then waits for a transition that never happens
because the mutation's onSuccess hasn't fired yet, and the test fails
at line 1187 with the panel still visible.
Bump both `query?key=table-update` waiters from 15s to 30s. The test
budget is 120s overall, so there's plenty of room.
The "warns on CREATE TABLE without RLS" spec used a fixed
`pw_rls_smoke_test` table name. With 3 self-hosted workers in
parallel, two workers can race: worker A's `dropTable` lands between
worker B's "Run and enable RLS" click and worker B's verifying
`pg_class` query, leaving `rows[0]?.relrowsecurity` as undefined and
asserting `expect(undefined).toBe(true)`.
Suffix the table with `test.info().parallelIndex`, matching the
pattern used elsewhere in the suite for parallel-safe per-worker
tables.
The queue-table-operations `beforeEach` called
`page.reload({ waitUntil: 'networkidle' })`. Studio keeps long-lived
polling and SSE connections open (PostHog, Sentry, realtime), so the
network never goes idle and the reload kept running until the 120 s
test timeout fired — sporadically failing whichever spec happened to
trigger it. The skill at .claude/skills/studio-e2e-tests already calls
this out explicitly.
Replace with a `waitForTableToLoad` waiter set up before the reload —
the entity-types query is the precise signal that the editor finished
loading after the reload, and it doesn't depend on whether the host's
background pollers happen to settle.
The Date Column Filters and Timestamp Column Filters specs inserted
rows using `CURRENT_DATE` / `NOW()`, which evaluate against postgres's
session timezone (≈ UTC in the supabase containers). The test then
filters using `getDateValue(N)`, which formats `new Date()` in the
host's local timezone — and the studio's "Today / Yesterday / Last 7
days / Last 30 days" dropdown options are built the same way. When
local TZ ≠ UTC, "today" in JS lands on a different calendar day than
postgres's CURRENT_DATE, the filter target doesn't match any inserted
row, and the test fails with "expected 2 rows, received 1".
CI runners default to UTC so this never surfaces there. On macOS in
non-UTC zones the date tests fail intermittently around the moments
when local and UTC dates diverge.
Insert explicit date strings derived from `getDateValue` instead of
postgres's `CURRENT_DATE`. The inserts and the filter value now share
a single source of truth (JS local), regardless of host TZ. For the
timestamp test, pin to mid-day so the row falls on the right calendar
day from postgres's perspective.
Confirmed: all 40 specs in `filter-bar.spec.ts` pass after the change,
including the 5 that were failing on Mac (Today / Yesterday / Last 7
days / date less than / timestamp), plus the ArrowDown/Enter dropdown
test that was getting cascaded by `maxFailures=3`.
The two specs under "Filter Bar › Keyboard Navigation - Value Input"
pressed `Home` to put the cursor at position 0 before exercising the
ArrowLeft-migrates-condition / typing-in-middle flows. That works on
the Linux CI runners but is a no-op on macOS Chromium — macOS routes
"go to start of line" through Cmd+ArrowLeft / Fn+ArrowLeft, and the
standalone Home key doesn't move the caret inside text inputs.
Locally the Home press did nothing, the cursor stayed at end of the
input, and:
- ArrowLeft hit `selectionStart === value.length` instead of `0`, so
the value-input ArrowLeft branch in useKeyboardNavigation never
fired and focus didn't migrate to the previous condition.
- The ArrowRight loop then matched the cursor-at-end branch in
handleArrowRight, which moved activeInput to the freeform input,
sending all subsequent typed characters there instead of into the
value input — so `_Middle_` never landed anywhere asserted.
Drive the cursor directly with `setSelectionRange` on each input. It's
portable across OSes and exercises exactly the keyboard handlers the
tests are actually about.
Studio code (e.g. `buildTableEditorUrl`) builds navigation targets with
`new URL().toString()`, which emits fully qualified
`http://localhost:8082/project/.../editor/123?…` strings. Next's router
tolerated this by treating same-origin absolute URLs as relative paths.
TanStack Router doesn't — it appends them to the current path,
producing garbage like
`/project/default/database/schemas/http:/localhost:8082/project/default/editor/123`.
Detect same-origin absolute URLs in the next/router compat shim and
strip the origin before handing the path to TanStack. Cross-origin
URLs pass through so TanStack can still hand them to the browser as
external navigations.
Unblocks `database.spec.ts:78 "Schema Visualizer › table actions"`
which clicks "View in Table Editor" → `buildTableEditorUrl()`. Also
trims the `view definition preserves security_invoker` table-editor
spec to focus on what it actually asserts (the link href encodes the
view option) rather than the SQL editor's hydration from
`?content=`, which is a separate flow.
Two unrelated test assertions were over-specific against behaviour that
varies between the legacy Next pages router and the TanStack-router
compat shim:
- `deleteBucket` in `utils/storage-helpers.ts` waited for the page to
redirect to `/storage/files` after delete. The studio's
`DeleteBucketModal` does fire `router.push('/storage/files')`, but on
TanStack Router that push can race other history updates and the URL
doesn't reliably change. Asserting the row is gone from the list is a
stronger signal anyway.
- `view definition preserves security_invoker` in `table-editor.spec.ts`
expected the "Open in SQL Editor" href to contain
`security_invoker%20%3D%20true` (percent-encoded spaces). The compat
Link shim round-trips the search params through `URLSearchParams`,
which emits `+` for spaces, so the actual href has
`security_invoker+%3D+true`. Both decode to the same SQL on the
receiving end; accept either form in the regex.
The two "can create a (new|unlogged) queue" specs were waiting on
`waitForURL(/.*\/integrations\/queues\/queues\/{queueName}/)` to confirm
the post-create navigation. With the next/router shim on TanStack
Router, the `router.push` after creation races nuqs's `?new=true`
history update and the URL doesn't always change client-side — the
mutation succeeds, the toast appears, the row appears in the list, but
the browser stays on /queues/queues/?new=true.
Switch the assertions to "dialog hides + the new row is visible in the
list", which is a tighter signal of "create succeeded" and is stable
regardless of how the post-create navigation resolves. The row
matcher anchors on word boundaries so `pw_queue_create` doesn't also
match `pw_queue_create_unlogged`.
All 6 specs in `queue-integration.spec.ts` pass after this change.
The cleanup of /tmp/playwright-locks was at the bottom of the global
setup test, but every branch above it returns early — including the
no-auth path that the self-hosted run always takes. So the cleanup was
dead code, and any prior run's `setup.done.json` marker for a file
using `withFileOnceSetup` would persist into the next run.
Because `supabase db reset` between runs wipes database state but the
lock files survive on disk, the cron-jobs spec (which relies on
`withFileOnceSetup` to create the pg_cron extension once per file) would
silently skip its setup on every second run and onwards, leaving
pg_cron uninstalled. All 11 cron-jobs specs then failed.
Move the cleanup to before any early return so it runs unconditionally.
Confirmed by re-running the affected specs: all 11 cron-jobs tests pass.
`ensureWebhooksEnabled` was timing out on `Database Webhooks allow you
to send real-time data` — copy that was removed when PR #44277 migrated
the webhooks page to the new integrations UI. The new page only renders
"Database Webhooks can be used to trigger serverless functions..." inside
an Admonition that conditionally appears (and disappears once the
supabase_functions extension is enabled), so it can't be a stable wait.
Wait on the page-header heading instead — it's rendered regardless of
the extension state.
All 5 specs in `database-webhooks.spec.ts` pass after this change.
## What kind of change does this PR introduce?
Studio UI cleanup.
## What is the current behaviour?
The SQL editor query warning modal repeats itself with an admonition, a
nested warning panel, and extra confirmation copy. Single-warning cases
are harder to scan than they need to be.
## What is the new behaviour?
The warning now uses `AlertDialog` and resolves each detected issue into
concise title/body copy. Single-warning cases show one short
description, while multi-warning cases show a compact list. The existing
RLS actions are preserved.
## Testing instructions
Manual SQL editor checks:
Open the Studio SQL editor, paste each snippet, click **Run**, verify
the warning copy, then click **Cancel**. These snippets are only
intended to trigger the warning UI, so do not confirm the dangerous
ones.
Destructive operation warning:
```sql
drop table if exists public.codex_warning_destructive;
```
UPDATE without WHERE warning:
```sql
update public.codex_warning_table set id = id;
```
Prevent database connections warning:
```sql
alter database postgres connection limit 0;
```
Missing RLS warning:
```sql
create schema if not exists codex_warning;
create table codex_warning.missing_rls (id bigint);
```
Multiple issues warning:
```sql
drop table if exists public.codex_warning_destructive;
update public.codex_warning_table set id = id;
create schema if not exists codex_warning;
create table codex_warning.missing_rls_multi (id bigint);
```
## Additional context
| Before | After |
| --- | --- |
| <img width="1024" height="759" alt="Codex Warning Table Maintenance
SQL Editor Hammer Toolshed Supabas-BDBD32C7-FCE8-4623-ACF2-D2554233EBB4"
src="https://github.com/user-attachments/assets/aaed16c2-9910-424a-8a3c-f9815139b1bf"
/> | <img width="1024" height="759" alt="Codex Warning Table Maintenance
SQL Editor Hammer Toolshed Supabas-C8964C98-1CF9-4992-89D6-86C081C884E8"
src="https://github.com/user-attachments/assets/d291c559-1e64-4c63-b918-b20b58d9a2a5"
/> |
| <img width="1024" height="759" alt="Codex Warning Table Maintenance
SQL Editor Hammer Toolshed Supabas-C341A032-B5B0-49A2-8EA2-E3E6EEC54E4F"
src="https://github.com/user-attachments/assets/667d9d1a-e34b-4411-9f91-4972ee8d1a23"
/> | <img width="1024" height="759" alt="Codex Warning Table Maintenance
SQL Editor Hammer Toolshed Supabas-FC66ADE9-6AF1-44D2-A6B6-F7B2FC935C0E"
src="https://github.com/user-attachments/assets/1348377e-6606-47c0-aa95-128d7f86ed56"
/> |
| <img width="1024" height="759" alt="Codex Warning Table Maintenance
SQL Editor Hammer Toolshed Supabas-43AAD9FA-7FAC-4DCE-A713-00E8FC76B343"
src="https://github.com/user-attachments/assets/bdadedc2-f17d-4011-ae67-5248097b3e92"
/> | <img width="1024" height="759" alt="Codex Warning Table Maintenance
SQL Editor Hammer Toolshed Supabas-279EC81D-31C7-49C7-B4A8-EEEF1738740A"
src="https://github.com/user-attachments/assets/0c178fff-ff49-4522-870d-7a3401c6af30"
/> |
| <img width="1024" height="759" alt="Codex Warning Table Maintenance
SQL Editor Hammer Toolshed Supabas-D405AED5-613F-4C78-909E-F718C67CF17E"
src="https://github.com/user-attachments/assets/a4399935-3596-471b-854a-c689e2e0df07"
/> | <img width="1024" height="759" alt="Codex Warning Table Maintenance
SQL Editor Hammer Toolshed Supabas-725FBCE6-5606-4BC3-B13F-6210DBADF6F2"
src="https://github.com/user-attachments/assets/0dcd08fc-cfb4-4d67-b167-eb6eaa768764"
/> |
| <img width="1024" height="759" alt="Codex Warning Table Maintenance
SQL Editor Hammer Toolshed Supabas-EA6563AF-DE7E-4AB8-9164-AC66164CA581"
src="https://github.com/user-attachments/assets/ef56fe0f-0243-4ff3-a2b2-ee8b3fe2330a"
/> | <img width="1024" height="759" alt="Codex Warning Table Maintenance
SQL Editor Hammer Toolshed Supabas-EC85EFB1-715E-4841-BBCA-51F88B539595"
src="https://github.com/user-attachments/assets/5be05527-920d-4f23-92eb-c7cd0bbff13e"
/> |
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **Bug Fixes**
* Prevented unintended modal dismissal and duplicate handling when
confirming.
* **Refactor**
* Redesigned SQL editor warning modal with structured, consolidated
warnings, adaptive title/confirmation copy, and centralized handling of
missing-RLS table names.
* Added conditional "Run and enable RLS" confirmation when available.
* **Tests**
* Updated end-to-end tests to match the new modal headings and body
text.
<!-- review_stack_entry_start -->
[](https://app.coderabbit.ai/change-stack/supabase/supabase/pull/45912)
<!-- review_stack_entry_end -->
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
## Problem
We run the e2e tests across two shards to reduce their duration.
However, one takes longer than the other.
## Solution
Use a currently hidden Playwright feature to balance their load
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **Chores**
* Upgraded Playwright test dependency to a newer version for refreshed
test tooling and fixes.
* Adjusted end-to-end test shard weighting to rebalance distribution
across parallel runs.
* **Tests**
* Stabilized several end-to-end editor interactions to make E2E flows
more reliable.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
Mark provenance of SQL via the branded types SafeSqlFragment and
UntrustedSqlFragment. Only SafeSqlFragment should be executed;
UntrustedSqlFragments require some kind of implicit user approval (show
on screen + user has to click something) before they are promoted to
SafeSqlFragment.
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Editor and RLS tester show loading states for inferred/generated SQL
and include a dedicated user SQL editor for safer edits.
* **Refactor**
* Platform-wide SQL handling tightened: snippets and AI-generated SQL
are treated as untrusted/display-only until promoted, improving safety
and consistency.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
## Summary
Adds the first batch of keyboard shortcuts for the Database → Schema
Visualizer page, following the registry pattern established for the SQL
editor and table editor.
Fixes [FE-3115](https://linear.app/supabase/issue/FE-3115).
## Shortcuts
| Shortcut | Action |
| --- | --- |
| `Mod+Shift+C` | Copy schema as SQL |
| `Mod+Shift+M` | Copy schema as Markdown |
| `D` then `P` | Download schema as PNG |
| `D` then `S` | Download schema as SVG |
| `O` then `A` | Open the auto-layout confirmation dialog |
| `O` then `S` | Open the schema selector |
All six entries appear in the Cmd+K command menu under "Shortcuts" and
in the global shortcuts sheet (`Mod+/`) under a new "Schema Visualizer"
group while the page is mounted. None are surfaced in Account →
Preferences yet (`showInSettings: false`), matching how the SQL/table
editor batches shipped.
The schema selector and auto-layout button are wrapped in the unified
`Shortcut` component so the keybind is shown on hover (Linear-style).
The dropdown items for copy/download don't get hover hints since
tooltips on dropdown items don't make sense — they're discoverable via
Cmd+K instead.
## Toasts
Each user-visible action now confirms via a sonner toast:
- `Successfully copied as SQL` — fires on Copy as SQL (button or
`Mod+Shift+C`).
- `Successfully copied as Markdown` — fires on Copy as Markdown
(dropdown or `Mod+Shift+M`).
- `Successfully downloaded as PNG` / `Successfully downloaded as SVG` —
already present in `useExportSchemaToImage`; fires on click or `D → P` /
`D → S`.
- `Failed to download current view: …` — error toast on download failure
(also pre-existing).
## Notes
- `Mod+Shift+C` and `Mod+Shift+M` collide with the SQL editor's
`results.copy-csv` / `results.copy-markdown` shortcuts. They coexist
cleanly because `useShortcut` only fires while the hook is mounted, and
the two pages live on different routes. Both labels appear in the global
shortcuts sheet honestly scoped per surface.
- `SchemaSelector` was promoted to a `forwardRef` component that spreads
extra props onto its outer `<div>`. This was needed for `<TooltipTrigger
asChild>` to attach event handlers and the ref properly — previously
they were silently dropped and the hover tooltip didn't render.
- `SchemaSelector` and the auto-layout `AlertDialog` accept controlled
`open` props now so the shortcuts can drive them and the tooltip can be
suppressed while the popover/dialog is open (`Shortcut` gained a
`tooltipOpen` passthrough for this).
- Auto-layout still pops the existing confirmation dialog rather than
running directly — destructive enough to keep the guardrail.
## Test plan
- [x] On the Schema Visualizer page, each of the six shortcuts fires the
corresponding action.
- [x] Hover the schema selector and the Auto layout button — tooltip
shows the action label and keybind badge.
- [x] Open the schema selector popover (click or `O → S`) — hover
tooltip is suppressed while open.
- [x] Open the auto-layout confirm dialog (click or `O → A`) — hover
tooltip is suppressed while open.
- [x] Cmd+K shows all six entries under "Shortcuts" while on the page;
navigating away unregisters them.
- [x] `Mod+/` shortcuts sheet has a "Schema Visualizer" group listing
all six.
- [x] Copy as SQL / Markdown each fire a confirmation toast; PNG / SVG
downloads each fire a confirmation toast.
- [x] On the SQL editor results page, `Mod+Shift+M` / `Mod+Shift+C`
still copy results (no regression from the duplicate keybinds).
- [x] The download dropdown items still work via click; PNG/SVG
downloads succeed.
- [x] All other consumers of `SchemaSelector` (~15 callsites) render
unchanged after the `forwardRef` promotion.
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Keyboard shortcuts for schema visualizer: copy as SQL/Markdown,
download PNG/SVG, auto-layout, and focus selector
* Success toasts when copying content to clipboard
* **Improvements**
* Schema selector and auto-layout dialog can be opened/closed via
keyboard and programmatically
* Shortcut tooltips can be suppressed when related overlays/dialogs are
open
* Schema Visualizer added to the shortcuts reference sheet
* **Tests**
* E2E tests dismiss transient toasts to avoid UI interference
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
## Context
Resolves FE-3126
Just cleaning up the table editor header with a bit of refactors
(pre-req to investigating collapsing filter bar and table editor header
actions into a single row)
## Non-visual changes involved
- Break down components within `GridHeaderActions` into smaller ones
- `IndexAdvisorPopover`
- `SecurityDefinerViewPopover`
- `RealtimeToggle`
- Deprecate use of `useUrlState` in `GridHeaderActions` to use
`useQueryState` instead
- Improve types for `TwoOptionToggle`
## Visual changes involved
- Collapse realtime button toggle into a button icon, with no text (just
tooltip)
- Adjust layout of buttons a little
### Before
<img width="796" height="118" alt="image"
src="https://github.com/user-attachments/assets/436bca94-4d91-471a-a184-487c6f78dc04"
/>
### After
<img width="731" height="132" alt="image"
src="https://github.com/user-attachments/assets/5fd30982-a1fc-4f92-a590-146d1e69d52a"
/>
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Index Advisor popover with recommendations.
* Realtime toggle to manage realtime table publication.
* Security Definer view popover with optional autofix.
* Insert menu for adding rows/columns and CSV import.
* **Bug Fixes**
* Adjusted filter bar input sizing for improved readability.
* **Refactor**
* Header layout updated and insert/import actions moved into dedicated
components.
* **Tests**
* Updated end-to-end selectors for the Insert row menu item.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
## TL;DR
- extends https://github.com/supabase/supabase/pull/45076
- closes https://github.com/supabase/supabase/issues/45113
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Sequence values now properly sync after bulk row inserts, ensuring
correct auto-increment behavior for subsequent inserts.
* **Tests**
* Added end-to-end coverage for CSV import, including a "Paste text"
import path that verifies custom-owned sequences are synchronized before
the next insert.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Co-authored-by: Gildas Garcia <1122076+djhi@users.noreply.github.com>
Co-authored-by: Ali Waseem <waseema393@gmail.com>
## Problem
When using LLMs, it's useful to describe your tables in markdown format.
## Solution
- Add an _Copy as SQL_ and _Copy as Markdown_ in the schema visualiser
table menu
<img width="320" height="235" alt="image"
src="https://github.com/user-attachments/assets/b465d6aa-a011-4308-86de-78725328630b"
/>
- Refactor the _Copy as SQL_ and _Download current view_ buttons in a
single button/dropdown combo and add _Copy as markdown_:
<img width="333" height="143" alt="image"
src="https://github.com/user-attachments/assets/a823988b-abff-4840-b5a5-53a5830065b4"
/>
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* "Copy as Markdown" for schemas and individual tables.
* "Copy as SQL" for individual tables.
* Per-column descriptions included in schema/table exports.
* **Style**
* Export actions consolidated into a compact, grouped dropdown with
adjacent copy action for streamlined header controls.
* **Tests**
* Unit tests for markdown export helpers.
* E2E tests updated to use the new export UI and adjusted dialog timing.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Co-authored-by: Copilot Autofix powered by AI <62310815+github-advanced-security[bot]@users.noreply.github.com>
## TL;DR
fixes csv imports with explicit id values leaving auto generated primary
keys out of sync
by resolving the primary key sequence correctly after import
## ex:
| Before | After |
| --- | --- |
| <img width="378" height="201" alt="Before: stale sequence after CSV
import"
src="https://github.com/user-attachments/assets/34c827f5-c69b-4c05-aa67-9fb6fd65a040"
/> | <img width="435" height="196" alt="After: sequence synced after CSV
import"
src="https://github.com/user-attachments/assets/bb84f286-1b3e-44a5-965b-5f2faa3f7622"
/> |
## ref:
- closes https://github.com/supabase/supabase/issues/45073
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
## Release Notes
* **Bug Fixes**
* Fixed identity sequence synchronization during CSV imports. After
spreadsheet data is inserted, the system now properly updates sequence
values, ensuring subsequent rows receive correct identifiers.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
## I have read the
[CONTRIBUTING.md](https://github.com/supabase/supabase/blob/master/CONTRIBUTING.md)
file.
YES
## What kind of change does this PR introduce?
Replaces "stored procedures" with "functions" for everything related to
the Data API.
## Additional context
It's not accurate to call database functions "stored procedures". It may
have been that way before Postgres 11, but now it causes confusion
because PostgREST allows functions and not stored procedures.
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **Documentation**
* Standardized terminology across docs, SDK guides, CLI/config specs,
examples, UI, and config comments to use "database functions" instead of
"stored procedures".
* Updated API docs, CLI/config descriptions, Studio UI labels, help
text, empty-state and navigation copy, RPC documentation, and example
text for consistency.
* Adjusted explanatory text and error/help messages to reflect the
revised terminology.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
## TL;DR
fixes enum create/update failures when names contain quotes
(also added a smol e2e)
## Ex:
<table>
<tr>
<td><strong>Before</strong></td>
<td><strong>After</strong></td>
</tr>
<tr>
<td>
<img width="424" height="178" alt="Before"
src="https://github.com/user-attachments/assets/d1815f4e-3879-4f8d-8d24-40d2c1f5563d"
/>
</td>
<td>
<img width="233" height="75" alt="After fix"
src="https://github.com/user-attachments/assets/f3f9b53c-b234-4e18-9b2d-db97ca4713d5"
/>
</td>
</tr>
</table>
## ref:
- closes https://github.com/supabase/supabase/issues/45022
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **Bug Fixes**
* Fixed enumerated type description handling to preserve special
characters (quotes and apostrophes) without unintended escaping.
* **Tests**
* Extended enumerated types test coverage to include creation, updates,
and deletion of types with special characters in names and descriptions.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
Alaister Young
