* Add shared AI bridge seam
Co-authored-by: Nick Misasi <nick13misasi@gmail.com>
* Add AI bridge test helper API
Co-authored-by: Nick Misasi <nick13misasi@gmail.com>
* Add AI bridge seam test coverage
Co-authored-by: Nick Misasi <nick13misasi@gmail.com>
* Add Playwright AI bridge recap helpers
Co-authored-by: Nick Misasi <nick13misasi@gmail.com>
* Fix recap channel persistence test
Co-authored-by: Nick Misasi <nick13misasi@gmail.com>
* Restore bridge client compatibility shim
Co-authored-by: Nick Misasi <nick13misasi@gmail.com>
* Expand recap card in Playwright spec
Co-authored-by: Nick Misasi <nick13misasi@gmail.com>
* Recaps e2e test coverage (#35543)
* Add Recaps Playwright page object
Co-authored-by: Nick Misasi <nick13misasi@gmail.com>
* Expand AI recap Playwright coverage
Co-authored-by: Nick Misasi <nick13misasi@gmail.com>
* Format recap Playwright coverage
Co-authored-by: Nick Misasi <nick13misasi@gmail.com>
* Fix recap regeneration test flows
Co-authored-by: Nick Misasi <nick13misasi@gmail.com>
---------
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
* Fix AI bridge lint and OpenAPI docs
Co-authored-by: Nick Misasi <nick13misasi@gmail.com>
* Fix recap lint shadowing
Co-authored-by: Nick Misasi <nick13misasi@gmail.com>
* Stabilize failed recap regeneration spec
Co-authored-by: Nick Misasi <nick13misasi@gmail.com>
* Fill AI bridge i18n strings
Co-authored-by: Nick Misasi <nick13misasi@gmail.com>
* Fix i18n
* Add service completion bridge path and operation tracking fields
Extend AgentsBridge with CompleteService for service-based completions,
add ClientOperation/OperationSubType tracking to BridgeCompletionRequest,
and propagate operation metadata through to the bridge client.
Made-with: Cursor
* Fill empty i18n translation strings for enterprise keys
The previous "Fix i18n" commit added 145 i18n entries with empty
translation strings, causing the i18n check to fail in CI. Fill in
all translations based on the corresponding error messages in the
enterprise and server source code.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* Fix i18n
* Fix i18n again
* Rename Complete/CompleteService to AgentCompletion/ServiceCompletion
Align the AgentsBridge interface method names with the underlying
bridge client methods they delegate to (AgentCompletion, ServiceCompletion).
Made-with: Cursor
* Refactor
* Add e2eAgentsBridge implementation
The new file was missed from the prior refactor commit.
Made-with: Cursor
* Address CodeRabbit review feedback
- Add 400 BadRequest response to AI bridge PUT endpoint OpenAPI spec
- Add missing client_operation, operation_sub_type, service_id fields to
AIBridgeTestHelperRecordedRequest schema
- Deep-clone nested JSON schema values in cloneJSONOutputFormat
- Populate ChannelID on recap summary bridge requests
- Fix msg_count assertion to mention_count for mark-as-read verification
- Make AgentCompletion/ServiceCompletion mutex usage atomic
Made-with: Cursor
* fix(playwright): align recaps page object with placeholder and channel menu
Made-with: Cursor
* fix(playwright): update recaps expectEmptyState to match RecapsList empty state
After the master merge, the recaps page now renders RecapsList's
"You're all caught up" empty state instead of the old placeholder.
Made-with: Cursor
* chore(playwright): update package-lock.json after npm install
Made-with: Cursor
* Revert "chore(playwright): update package-lock.json after npm install"
This reverts commit 95c670863a.
* style(playwright): fix prettier formatting in recaps page object
Made-with: Cursor
* fix(playwright): handle both recaps empty states correctly
The recaps page has two distinct empty states:
- Setup placeholder ("Set up your recap") when allRecaps is empty
- RecapsList caught-up state ("You're all caught up") when the
filtered tab list is empty
Split expectEmptyState into expectSetupPlaceholder and
expectCaughtUpEmptyState, used by the delete and bridge-unavailable
tests respectively.
Made-with: Cursor
---------
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Single-channel guests are excluded from billable seat counts for
licensing. Include this metric in the support packet so support
engineers can understand seat calculation discrepancies.
Made-with: Cursor
Co-authored-by: Mattermost Build <build@mattermost.com>
* updated go to version 1.25.8
* updated gotestsum version to work with go 1.25.8
go 1.25 does not work with indirect tools 0.11 dependency pulled by
gotestsum.
* Use sync.WaitGroup.Go to simplify goroutine creation
Replace the wg.Add(1) + go func() { defer wg.Done() }() pattern with
wg.Go(), which was introduced in Go 1.25.
* pushes fips image on workflow dispatch to allow fips test to run on go version update
* fix new requirements for FIPS compliance imposed on updating to go 1.25.8
* updates openssl symbol check for library shipped with FIPS new versions
go-openssl v2 shipped with FIPS versions starting from 1.25 uses mkcgo to generate
bindings causing symbol names to be different.
* removes temp workflow-dispatch condition
* keep versions out of agents md file
* dont panic in CleanupSqlSettings so we can surface migration failures
* update mattermost-govet
* nolint existing missing CONCURRENT
* Revert "nolint existing missing CONCURRENT"
This reverts commit 45b3110bf0.
* simplify
* approve up to 166
- Skip draft PRs and trigger on ready_for_review to run when
a draft is marked ready
- Remove use_sticky_comment (broken upstream, see
anthropics/claude-code-action#1052) and inline review comments
which caused claude[bot] to appear as a PR reviewer
- Write analysis to a file and post it via github-script as a
sticky comment from github-actions[bot], avoiding conflicts
with the Claude Code Review App
- Manage docs/needed label from the file directly
Made-with: Cursor
* improves time limit checks
* consistently check for presence of patch fields
* fix variable shadowing in test
* allow idempotent pinning operations with time limit expired
* new utility function for post limit time check
* fix style issue
The claude-code-action blocks bot-initiated workflows by default.
Adding cursor to allowed_bots so PRs opened by Cursor are analyzed.
Made-with: Cursor
* Add CreatedBy and UpdatedBy to the properties fields and values (#34485)
* Add CreatedBy and UpdatedBy to the properties fields and values
* Fix types
---------
Co-authored-by: Miguel de la Cruz <miguel@ctrlz.es>
* Adds ObjectType to the property fields table (#34908)
Co-authored-by: Miguel de la Cruz <miguel@ctrlz.es>
* Update ObjectType migration setting an empty value and marking the column as not null (#34915)
Co-authored-by: Miguel de la Cruz <miguel@ctrlz.es>
* Adds uniqueness mechanisms to the property fields (#35058)
* Adds uniqueness mechanisms to the property fields
After adding ObjectType, this commit ensures that both the PSAv1 and
PSAv2 schemas are supported, and enforces property uniqueness through
both database indexes and a logical check when creating new property
fields.
* Adds uniqueness check to property updates
Updates are covered on this commit and we refactor as well the SQL
code to use the squirrel builder and work better with the conditional
addition of the `existingID` piece of the query.
* Add translations to error messages
* Fixing retrylayer mocks
* Remove retrylayer duplication
* Address review comments
* Fix comment to avoid linter issues
* Address PR comments
* Update server/channels/db/migrations/postgres/000157_add_object_type_to_property_fields.down.sql
Co-authored-by: Alejandro García Montoro <alejandro.garciamontoro@gmail.com>
* Update server/channels/db/migrations/postgres/000157_add_object_type_to_property_fields.up.sql
Co-authored-by: Alejandro García Montoro <alejandro.garciamontoro@gmail.com>
* Update server/channels/db/migrations/postgres/000157_add_object_type_to_property_fields.up.sql
Co-authored-by: Alejandro García Montoro <alejandro.garciamontoro@gmail.com>
* Update field validation to check only for valid target types
* Update migrations to avoid concurrent index creation within a transaction
* Update migrations to make all index ops concurrent
* Update tests to use valid PSAv2 property fields
* Adds a helper for valid PSAv2 TargetTypes
---------
Co-authored-by: Miguel de la Cruz <miguel@ctrlz.es>
Co-authored-by: Alejandro García Montoro <alejandro.garciamontoro@gmail.com>
* Fix property tests (#35388)
Co-authored-by: Miguel de la Cruz <miguel@ctrlz.es>
* Adds Integrated Boards feature flag (#35378)
Co-authored-by: Miguel de la Cruz <miguel@ctrlz.es>
* Adds Integrated Boards MVP API changes (#34822)
This PR includes the necessary changes for channels and posts
endpoints and adds a set of generic endpoints to retrieve and manage
property fields and values following the new Property System approach.
Co-authored-by: Miguel de la Cruz <miguel@ctrlz.es>
Co-authored-by: Mattermost Build <build@mattermost.com>
* Property System Architecture permissions for v2 (#35113)
* Adds uniqueness mechanisms to the property fields
After adding ObjectType, this commit ensures that both the PSAv1 and
PSAv2 schemas are supported, and enforces property uniqueness through
both database indexes and a logical check when creating new property
fields.
* Adds uniqueness check to property updates
Updates are covered on this commit and we refactor as well the SQL
code to use the squirrel builder and work better with the conditional
addition of the `existingID` piece of the query.
* Add translations to error messages
* Add the permissions to the migrations, model and update the store calls
* Adds the property field and property group app layer
* Adds authorization helpers for property fields and values
* Make sure that users cannot lock themselves out of property fields
* Migrate permissions from a JSON column to three normalized columns
* Remove the audit comment
* Use target level constants in authorization
* Log authorization membership failures
* Rename admin to sysadmin
* Fix i18n sorting
---------
Co-authored-by: Miguel de la Cruz <miguel@ctrlz.es>
* Add Views store and app layer (#35361)
* Add Views store and app layer for Integrated Boards
Implements the View entity (model, SQL store, service, app) as described
in the Integrated Boards tech spec. Views are channel-scoped board
configurations with typed props (board, kanban subviews) and soft-delete.
- public/model: View, ViewBoardProps, Subview, ViewPatch types with
PreSave/PreUpdate/IsValid/Patch/Clone/Auditable
- Migration 158: Views table with jsonb Props column and indexes
- SqlViewStore: CRUD with nil-safe Props marshaling (AppendBinaryFlag)
- ViewService: CreateView seeds default kanban subview and links the
boards property field; caches boardPropertyFieldID at startup
- App layer: CreateView/GetView/GetViewsForChannel/UpdateView/DeleteView
with channel-membership permission checks and WebSocket events
(view_created, view_updated, view_deleted)
- doSetupBoardsPropertyField: registers the Boards property group and
board field in NewServer() before ViewService construction
- GetFieldByName now returns store.ErrNotFound instead of raw sql.ErrNoRows
* Move permission checks out of App layer for views
- Remove HasPermissionToChannel calls from all App view methods
- Drop userID params from GetView, GetViewsForChannel, UpdateView, DeleteView
- Fix doSetupBoardsPropertyField to include required TargetType for PSAv2 field
* Make View service generic and enforce board validation in model
- Remove board-specific auto-setup from service and server startup
- Enforce that board views require Props, at least one subview, and at least one linked property in IsValid()
- Move default subview seeding out of app layer; callers must provide valid props
- Call PreSave on subviews during PreUpdate to assign IDs to new subviews
- Update all tests to reflect the new validation requirements
* Restore migrations files to match base branch
* Distinguish ErrNotFound from other errors in view store Get
* Use CONCURRENTLY and nontransactional for index operations in views migration
* Split views index creation into separate nontransactional migrations
* Update migrations.list
* Update i18n translations for views
* Fix makeView helper to include required Props for board view validation
* Rename ctx parameter from c to rctx in OAuthProvider mock
* Remove views service layer, call store directly from app
* Return 500 for unexpected DB errors in GetView, 404 only for not-found
* Harden View model: deep-copy Props, validate linked property IDs
- Add ViewBoardProps.Clone() to deep-copy LinkedProperties and Subviews
- Use it in View.Clone() and View.Patch() to prevent shared-slice aliasing
- Iterate over LinkedProperties in View.IsValid() and reject invalid IDs
with a dedicated i18n key
- Register ViewStore in storetest AssertExpectations so mock expectations
are enforced
- Add tests covering all new behaviours
* Restore autotranslation worker_stopped i18n translation
* Fix view store test IDs and improve error handling in app layer
- Use model.NewId() for linked property IDs in testUpdateView to fix
validation failure (IsValid rejects non-UUID strings)
- Fix import grouping in app/view.go (stdlib imports in one block)
- Return 404 instead of 500 when Update/Delete store calls return
ErrNotFound (e.g. concurrent deletion TOCTOU race)
* Add View store mock to retrylayer test genStore helper
The View store was added to the store interface but the genStore()
helper in retrylayer_test.go was not updated, causing TestRetry to panic.
Also removes the duplicate Recap mock registration.
* Refactor view deletion and websocket event handling; update SQL store methods to use query builder
* revert property field store
* Remove useless migrations
* Add cursor-based pagination to View store GetForChannel
- Add ViewQueryCursor and ViewQueryOpts types with validation
- Return (views, cursor, error) for caller-driven pagination
- PerPage clamping: <=0 defaults to 20, >200 clamps to 200
- Support IncludeDeleted filter
- Add comprehensive store tests for pagination, cursor edge cases,
PerPage clamping, and invalid input rejection
- Add app layer test for empty channelID → 400
- Update interface, retrylayer, timerlayer, and mock signatures
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* Refactor test loops in ViewStore tests for improved readability
* change pagination to limit/offset
* Add upper-bound limits on View Subviews and LinkedProperties
Defense-in-depth validation: cap Subviews at 50 and LinkedProperties
at 500 to prevent abuse below the 300KB payload limit.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
* MM-67388, MM-66528, MM-67750: Add View REST API endpoints, websocket events, and sort order (#35442)
* Add Views store and app layer for Integrated Boards
Implements the View entity (model, SQL store, service, app) as described
in the Integrated Boards tech spec. Views are channel-scoped board
configurations with typed props (board, kanban subviews) and soft-delete.
- public/model: View, ViewBoardProps, Subview, ViewPatch types with
PreSave/PreUpdate/IsValid/Patch/Clone/Auditable
- Migration 158: Views table with jsonb Props column and indexes
- SqlViewStore: CRUD with nil-safe Props marshaling (AppendBinaryFlag)
- ViewService: CreateView seeds default kanban subview and links the
boards property field; caches boardPropertyFieldID at startup
- App layer: CreateView/GetView/GetViewsForChannel/UpdateView/DeleteView
with channel-membership permission checks and WebSocket events
(view_created, view_updated, view_deleted)
- doSetupBoardsPropertyField: registers the Boards property group and
board field in NewServer() before ViewService construction
- GetFieldByName now returns store.ErrNotFound instead of raw sql.ErrNoRows
* Move permission checks out of App layer for views
- Remove HasPermissionToChannel calls from all App view methods
- Drop userID params from GetView, GetViewsForChannel, UpdateView, DeleteView
- Fix doSetupBoardsPropertyField to include required TargetType for PSAv2 field
* Make View service generic and enforce board validation in model
- Remove board-specific auto-setup from service and server startup
- Enforce that board views require Props, at least one subview, and at least one linked property in IsValid()
- Move default subview seeding out of app layer; callers must provide valid props
- Call PreSave on subviews during PreUpdate to assign IDs to new subviews
- Update all tests to reflect the new validation requirements
* Restore migrations files to match base branch
* Distinguish ErrNotFound from other errors in view store Get
* Use CONCURRENTLY and nontransactional for index operations in views migration
* Split views index creation into separate nontransactional migrations
* Update migrations.list
* Update i18n translations for views
* Fix makeView helper to include required Props for board view validation
* Rename ctx parameter from c to rctx in OAuthProvider mock
* Remove views service layer, call store directly from app
* Return 500 for unexpected DB errors in GetView, 404 only for not-found
* Harden View model: deep-copy Props, validate linked property IDs
- Add ViewBoardProps.Clone() to deep-copy LinkedProperties and Subviews
- Use it in View.Clone() and View.Patch() to prevent shared-slice aliasing
- Iterate over LinkedProperties in View.IsValid() and reject invalid IDs
with a dedicated i18n key
- Register ViewStore in storetest AssertExpectations so mock expectations
are enforced
- Add tests covering all new behaviours
* Restore autotranslation worker_stopped i18n translation
* Fix view store test IDs and improve error handling in app layer
- Use model.NewId() for linked property IDs in testUpdateView to fix
validation failure (IsValid rejects non-UUID strings)
- Fix import grouping in app/view.go (stdlib imports in one block)
- Return 404 instead of 500 when Update/Delete store calls return
ErrNotFound (e.g. concurrent deletion TOCTOU race)
* Add View store mock to retrylayer test genStore helper
The View store was added to the store interface but the genStore()
helper in retrylayer_test.go was not updated, causing TestRetry to panic.
Also removes the duplicate Recap mock registration.
* Refactor view deletion and websocket event handling; update SQL store methods to use query builder
* revert property field store
* Add View API endpoints with OpenAPI spec, client methods, and i18n
Implement REST API for channel views (board-type) behind the
IntegratedBoards feature flag. Adds CRUD endpoints under
/api/v4/channels/{channel_id}/views with permission checks
matching the channel bookmark pattern.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* Remove useless migrations
* Add cursor-based pagination to View store GetForChannel
- Add ViewQueryCursor and ViewQueryOpts types with validation
- Return (views, cursor, error) for caller-driven pagination
- PerPage clamping: <=0 defaults to 20, >200 clamps to 200
- Support IncludeDeleted filter
- Add comprehensive store tests for pagination, cursor edge cases,
PerPage clamping, and invalid input rejection
- Add app layer test for empty channelID → 400
- Update interface, retrylayer, timerlayer, and mock signatures
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* Add cursor-based pagination to View API for channel views
* Enhance cursor handling in getViewsForChannel and update tests for pagination
* Refactor test loops in ViewStore tests for improved readability
* Refactor loop in TestGetViewsForChannel for improved readability
* change pagination to limit/offset
* switch to limit/offset pagination
* Add upper-bound limits on View Subviews and LinkedProperties
Defense-in-depth validation: cap Subviews at 50 and LinkedProperties
at 500 to prevent abuse below the 300KB payload limit.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* Add view sort order API endpoint
Add POST /api/v4/channels/{channel_id}/views/{view_id}/sort_order
endpoint following the channel bookmarks reorder pattern. Includes
store, app, and API layers with full test coverage at each layer.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* Add connectionId to view WebSocket events and sort_order API spec
Thread connectionId from request header through all view handlers
(create, update, delete, sort_order) to WebSocket events, matching
the channel bookmarks pattern. Add sort_order endpoint to OpenAPI
spec. Update minimum server version to 11.6.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* Remove duplicate View/ViewPatch definitions from definitions.yaml
The merge from integrated-boards-mvp introduced duplicate View and
ViewPatch schema definitions that were already defined earlier in
the file with more detail (including ViewBoardProps ref and enums).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* Update minimum server version to 11.6 in views API spec
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* Add missing translations for view sort order error messages
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* Merge integrated-boards-mvp into ibmvp_api-views; remove spec files
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* Fix flaky TestViewStore timestamp test on CI
Add sleep before UpdateSortOrder to ensure timestamps differ,
preventing same-millisecond comparisons on fast CI machines.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* remove duplicate views.yaml imclude
* Use c.boolString() for include_deleted query param in GetViewsForChannel
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* Fix views.yaml sort order schema: use integer type and require body
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* Refactor view sort order tests to use named IDs instead of array indices
Extract idA/idB/idC from views slice and add BEFORE/AFTER comments
to make stateful subtest ordering easier to follow.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* Return 404 instead of 403 for view operations on deleted channels
Deleted channels should appear non-existent to callers rather than
revealing their existence via a 403. Detailed error text explains
the context for debugging.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* add missing channel deleteat checks
* Use c.Params.Page instead of manual page query param parsing in getViewsForChannel
c.Params already validates and defaults page/per_page, so the manual
parsing was redundant.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* Add support for total count in views retrieval
* Add tests for handling deleted views in GetViewsForChannel and GetView
* Short-circuit negative newIndex in UpdateSortOrder before opening transaction
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* Add per-channel limit on views to bound UpdateSortOrder cost
Without a cap, unbounded view creation makes sort-order updates
increasingly expensive (CASE WHEN per view, row locks). Adds
MaxViewsPerChannel=50 constant and enforces it in the app layer
before saving. Includes API and app layer tests.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* Remove include_deleted support from views API
Soft-deleted views are structural metadata with low risk, but no other
similar endpoint (e.g. channel bookmarks) exposes deleted records without
an admin gate. Rather than adding an admin-only permission check for
consistency, remove the feature entirely since there is no current use case.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* Update view permissions to require `create_post` instead of channel management permissions
* Remove obsolete view management error messages for direct and group messages
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
* feat(migrations): add user tracking and object type to property fields
- Introduced user tracking columns (CreatedBy, UpdatedBy) to PropertyFields and PropertyValues.
- Added ObjectType column to PropertyFields with associated unique indexes for legacy and typed properties.
- Created new migration scripts for adding and dropping these features, including necessary indexes for data integrity.
- Established views for managing property fields with new attributes.
This update enhances the schema to support better tracking and categorization of property fields.
* Add Property System Architecture v2 API endpoints (#35583)
* Adds uniqueness mechanisms to the property fields
After adding ObjectType, this commit ensures that both the PSAv1 and
PSAv2 schemas are supported, and enforces property uniqueness through
both database indexes and a logical check when creating new property
fields.
* Adds uniqueness check to property updates
Updates are covered on this commit and we refactor as well the SQL
code to use the squirrel builder and work better with the conditional
addition of the `existingID` piece of the query.
* Add translations to error messages
* Add the permissions to the migrations, model and update the store calls
* Adds the property field and property group app layer
* Adds authorization helpers for property fields and values
* Make sure that users cannot lock themselves out of property fields
* Migrate permissions from a JSON column to three normalized columns
* Remove the audit comment
* Use target level constants in authorization
* Log authorization membership failures
* Rename admin to sysadmin
* Adds the Property System Architecture v2 API endpoints
* Adds permission checks to the create field endpoint
* Add target access checks to value endpoints
* Add default branches for object_type and target_type and extra guards for cursor client4 methods
* Fix vet API mismatch
* Fix error checks
* Fix linter
* Add merge semantics for property patch logic and API endpoint
* Fix i18n
* Fix duplicated patch elements and early return on bad cursor
* Update docs to use enums
* Fix i18n sorting
* Update app layer to return model.AppError
* Adds a limit to the number of property values that can be patched in the same request
* Require target_type filter when searching property fields
* Add objectType validation as part of field.IsValid()
* Fix linter
* Fix test with bad objecttpye
* Fix test grouping
---------
Co-authored-by: Miguel de la Cruz <miguel@ctrlz.es>
* MM-67968: Flatten view model — remove icon, subviews, typed board props (#35726)
* feat(views): flatten view model by removing icon, subview, and board props
Simplifies the View data model as part of MM-67968: removes Icon, Subview,
and ViewBoardProps types; renames ViewTypeBoard to ViewTypeKanban; replaces
typed Props with StringInterface (map[string]any); adds migration 000167
to drop the Icon column from the Views table.
Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
* feat(api): update views OpenAPI spec to reflect flattened model
Removes ViewBoardProps, Subview, and icon from the View and ViewPatch
schemas. Changes type enum from board to kanban. Replaces typed props
with a free-form StringInterface object. Aligns with MM-67968.
Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
* refactor(views): simplify store by dropping dbView and marshalViewProps
StringInterface already implements driver.Valuer and sql.Scanner, so the
manual JSON marshal/unmarshal and the dbView intermediate struct were
redundant. model.View now scans directly from the database. Also removes
the dead ViewMaxLinkedProperties constant and wraps the Commit() error in
UpdateSortOrder.
Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
* fix(api): allow arbitrary JSON in view props OpenAPI schema
The props field was restricted to string values via
additionalProperties: { type: string }, conflicting with the Go model's
StringInterface (map[string]any). Changed to additionalProperties: true
in View, ViewPatch, and inline POST schemas.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
* Adds basic implementation of the generic redux store for PSAv2 (#35512)
* Adds basic implementation of the generic redux store for PSAv2
* Add created_by and updated_by to the test fixtures
* Make target_id, target_type and object_type mandatory
* Wrap getPropertyFieldsByIds and getPropertyValuesForTargetByFieldIds with createSelector
* Address PR comments
---------
Co-authored-by: Miguel de la Cruz <miguel@ctrlz.es>
* Adds websocket messages for the PSAv2 API events (#35696)
* Adds uniqueness mechanisms to the property fields
After adding ObjectType, this commit ensures that both the PSAv1 and
PSAv2 schemas are supported, and enforces property uniqueness through
both database indexes and a logical check when creating new property
fields.
* Adds uniqueness check to property updates
Updates are covered on this commit and we refactor as well the SQL
code to use the squirrel builder and work better with the conditional
addition of the `existingID` piece of the query.
* Add translations to error messages
* Add the permissions to the migrations, model and update the store calls
* Adds the property field and property group app layer
* Adds authorization helpers for property fields and values
* Make sure that users cannot lock themselves out of property fields
* Migrate permissions from a JSON column to three normalized columns
* Remove the audit comment
* Use target level constants in authorization
* Log authorization membership failures
* Rename admin to sysadmin
* Adds the Property System Architecture v2 API endpoints
* Adds permission checks to the create field endpoint
* Add target access checks to value endpoints
* Add default branches for object_type and target_type and extra guards for cursor client4 methods
* Fix vet API mismatch
* Fix error checks
* Fix linter
* Add merge semantics for property patch logic and API endpoint
* Fix i18n
* Fix duplicated patch elements and early return on bad cursor
* Update docs to use enums
* Fix i18n sorting
* Update app layer to return model.AppError
* Adds a limit to the number of property values that can be patched in the same request
* Adds websocket messages for the PSAv2 API events
* Add IsPSAv2 helper to the property field for clarity
* Add guard against nil returns on field deletion
* Add docs to the websocket endpoints
---------
Co-authored-by: Miguel de la Cruz <miguel@ctrlz.es>
* migrations: consolidate views migrations and reorder after master
- Merged 000165 (create Views) with 000167 (drop Icon) since Icon was never needed
- Renumbered branch migrations 159-166 → 160-167 so master's 000159 (deduplicate_policy_names) runs first
- Regenerated migrations.list
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* Add API endpoint to retrieve posts for a specific view (#35604)
Automatic Merge
* Apply fixes after merge
* Return a more specific error from getting multiple fields
* Prevent getting broadcast params on field deletion if not needed
* Remove duplicated migration code
* Update property conflict code to always use master
* Adds nil guard when iterating on property fields
* Check that permission level is valid before getting rejected by the database
* Validate correctness on TargetID for PSAv2 fields
* Avoid PSAv1 using permissions or protected
* Fix test data after validation change
* Fix flaky search test
* Adds more posts for filter use cases to properly test exclusions
---------
Co-authored-by: Miguel de la Cruz <miguel@ctrlz.es>
Co-authored-by: Alejandro García Montoro <alejandro.garciamontoro@gmail.com>
Co-authored-by: Julien Tant <julien@craftyx.fr>
Co-authored-by: Mattermost Build <build@mattermost.com>
Co-authored-by: Julien Tant <785518+JulienTant@users.noreply.github.com>
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
* Add H/M/S sub-day units to validateRelativePattern
* Fix datetime MinDate/MaxDate to use validateDateTimeFormat
* Add H/M/S sub-day resolution to resolveRelativeDateToMoment
* Add minDateTime/maxDateTime props to DateTimeInput
* Wire min_date/max_date resolution in AppsFormDateTimeField
* Align client relative pattern bounds with server validation
* Fix allowPastDates when minDateTime is in the past
* ci: add test sharding plumbing to server CI
Add infrastructure for upcoming test sharding without changing behavior:
- Add shard-index and shard-total inputs to server-test-template.yml
(defaults preserve existing single-runner behavior)
- Add timing cache restore step (activates only when shard-total > 1)
- Add merge-postgres-test-results job to server-ci.yml that:
- Merges JUnit XML reports from shard artifacts
- Saves timing data cache for future shard balancing
- Handles both single-artifact and multi-shard scenarios
- Add .gitignore entries for timing cache and shard work files
Co-authored-by: Claude <claude@anthropic.com>
* ci: shard server Postgres tests into 4 parallel runners
Extract sharding logic into standalone, tested scripts and enable
4-shard parallel test execution for server Postgres CI:
Scripts:
- server/scripts/shard-split.js: Node.js bin-packing solver that
assigns test packages to shards using timing data from previous runs.
Two-tier strategy: light packages (<2min) whole, heavy packages
(api4, app) split at individual test level.
- server/scripts/run-shard-tests.sh: Multi-run wrapper that calls
gotestsum directly for each package group with -run regex filters.
- server/scripts/shard-split.test.js: 8 test cases covering round-robin
fallback, timing-based balancing, heavy package splitting, JUnit XML
fallback, and enterprise package separation.
Workflow changes:
- server-test-template.yml: Add shard splitting step that discovers test
packages and runs the solver. Modified Run Tests step to use wrapper
script when sharding is active.
- server-ci.yml: Add 4-shard matrix to test-postgres-normal. Update
merge job artifact patterns for shard-specific names.
Performance: 7.2 min with timing cache vs 62.5 min baseline = 88%
wall-time improvement. First run without cache uses JUnit XML fallback
or round-robin, then populates the cache for subsequent runs.
Co-authored-by: Claude <claude@anthropic.com>
* fix: raise heavy package threshold to 5 min to preserve test isolation
sqlstore integrity tests scan the entire database and fail when other
packages' test data is present. At 182s, sqlstore was just over the
120s threshold and getting split at test level. Raising to 300s keeps
only api4 (~38 min) and app (~15 min) as heavy — where the real
sharding gains are — while sqlstore, elasticsearch, etc. stay whole
and maintain their test isolation guarantees.
Co-authored-by: Claude <claude@anthropic.com>
* ci: only save test timing cache on default branch
PR branches always restore from master's timing cache via restore-keys
prefix matching. Timing data is stable day-to-day so this eliminates
cache misses on first PR runs and reduces cache storage.
Co-authored-by: Claude <claude@anthropic.com>
* ci: skip FIPS tests on PRs (enterprise CI handles compile check)
Per review feedback: the enterprise CI already runs a FIPS compile
check on every PR. Running the full FIPS test suite on PRs is redundant
since it uses the identical test suite as non-FIPS — the only
FIPS-specific failure mode is a build failure from non-approved crypto
imports, which the enterprise compile check catches.
Full FIPS tests continue to run on every push to master.
Co-authored-by: Claude <claude@anthropic.com>
* fix: address review feedback on run-shard-tests.sh
- Remove set -e so all test runs execute even if earlier ones fail;
track failures and exit with error at the end (wiggin77)
- Remove unused top-level COVERAGE_FLAG variable (wiggin77)
- Fix RUN_IDX increment position so report, json, and coverage files
share the same index (wiggin77)
- Update workflow comment: heavy threshold is 5 min, not 2 min (wiggin77)
Co-authored-by: Claude <claude@anthropic.com>
* style: use node: prefix for built-in fs module in shard-split.js
Co-authored-by: Claude <claude@anthropic.com>
* fix: avoid interpolating file paths into generated shell script
Read shard package lists from files at runtime instead of interpolating
them into the generated script via printf. This prevents theoretical
shell metacharacter injection from directory names, as flagged by
DryRun Security.
Co-authored-by: Claude <claude@anthropic.com>
* fix(ci): rename merged artifact to match server-ci-report glob
The merged artifact was named postgres-server-test-logs-merged which
does not match the *-test-logs pattern in server-ci-report.yml,
causing Postgres test results to be missing from PR/commit reports.
Also pins junit-report-merger to exact version 7.0.0 for supply chain
safety.
Co-authored-by: Claude <claude@anthropic.com>
* fix(ci): pass RACE_MODE env into Docker container
RACE_MODE was set on the host runner but never included in the docker
run --env list. The light-package path worked because the heredoc
expanded on the host, but run-shard-tests.sh reads RACE_MODE at
runtime inside the container where it was unset. This caused heavy
packages (api4, app) to silently lose -race detection.
Co-authored-by: Claude <claude@anthropic.com>
* fix(ci): discover new tests in heavy packages not in timing cache
Tests not present in the timing cache (newly added or renamed) would
not appear in any shard -run regex, causing them to silently skip.
After building items from the cache, run go test -list to discover
current test names and assign any cache-missing tests to shards via
the normal bin-packing algorithm with a small default duration.
Co-authored-by: Claude <claude@anthropic.com>
* fix(ci): add missing line continuation backslash in docker run
The previous --env FIPS_ENABLED line was missing a trailing backslash
after adding --env RACE_MODE, causing docker run to see a truncated
command and fail with "requires at least 1 argument".
Co-authored-by: Claude <claude@anthropic.com>
* fix(ci): add setup-go step for shard test discovery
go test -list in shard-split.js runs on the host runner via execSync,
but Go is only available inside the Docker container. Without this
step, every invocation fails silently and new-test discovery is a
no-op. Adding actions/setup-go before the shard split step ensures
the Go toolchain is available on the host.
Co-authored-by: Claude <claude@anthropic.com>
---------
Co-authored-by: Claude <claude@anthropic.com>
* Refactor property system with app layer routing and access control separation
Establish the app layer as the primary entry point for property operations
with intelligent routing based on group type. This architecture separates
access-controlled operations (CPA groups) from standard operations,
improving performance and code clarity.
Architecture Changes:
- App layer now routes operations based on group type:
- CPA groups -> PropertyAccessService (enforces access control)
- Non-CPA groups -> PropertyService (direct, no access control)
- PropertyAccessService simplified to handle only CPA operations
- Eliminated redundant group type checks throughout the codebase
* Move access control routing into PropertyService
This change makes the PropertyService the main entrypoint for property
related operations, and adds a routing mechanism to decide if extra
behaviors or checks should run for each operation, in this case, the
property access service logic.
To add specific payloads that pluggable checks and operations may
need, we use the request context. When the request comes from the API,
the endpoints are in charge of adding the caller ID to the payload,
and in the case of the plugin API, on receiving a request, the server
automatically tags the context with the plugin ID so the property
service can react accordingly.
Finally, the new design enforces all these checks migrating the actual
property logic to internal, non-exposed methods, so any caller from
the App layer needs to go through the service checks that decide if
pluggable logic is needed, avoiding any possibility of a bypass.
* Fix i18n
* Fix bad error string
* Added nil guards to property methods
* Add check for multiple group IDs on value operations
* Add nil guard to the plugin checker
* Fix build error
* Update value tests
* Fix linter
* Adds early return when content flaggin a thread with no replies
* Fix mocks
* Clean the state of plugin property tests before each run
* Do not wrap appErr on API response and fix i18n
* Fix create property field test
* Remove the need to cache cpaGroupID as part of the property service
* Split the property.go file into multiple
* Not found group doesn't bypass access control check
* Unexport SetPluginCheckerForTests
* Rename plugin context getter to be more PSA specific
---------
Co-authored-by: Miguel de la Cruz <miguel@ctrlz.es>
* MM-68071 Upgrade core-js and @babel/preset-env to latest
* Update targeted browsers in @babel/preset-env
These match what the AI plugin currently targets. We could probably go more
recent, but this should reduce the size of the core-js polyfill a fair bit.
The previous versions targeted were released around 2018 or so, and the new
ones are from late 2022-mid 2023.
* Ensure Babel uses new version of core-js
---------
Co-authored-by: Mattermost Build <build@mattermost.com>
* do not attempt to decompress 7zip files in docextractor
* use SevenZip.Match to detect a 7z file instead of our own method
* merge conflicts
---------
Co-authored-by: Mattermost Build <build@mattermost.com>
* Add shared channel integration test tool (MM-67944)
Add tools/sharedchannel-test, a standalone Go tool that validates shared
channel synchronization between two real Mattermost Enterprise instances.
The tool builds and manages two server processes with separate databases,
establishes a remote cluster connection, and runs integration tests for
membership, post, and reaction sync.
Test coverage:
- Membership: add, remove, re-add, bulk remove
- Posts: create, edit, delete
- Reactions: add, remove
Uses mlog with dual console targets (stdout for info, stderr for errors)
and exits non-zero on failure for CI integration.
* Fix plugin API CreateChannel read-after-write failure with read replicas
LocalCacheChannelStore did not override Save, so newly created channels
were never added to channelByIdCache. When a plugin called CreateChannel
followed immediately by AddChannelMember, the GetChannel lookup missed
the cache and fell through to GetReplica(), which on deployments with
DataSourceReplicas could read from a replica that hadn't replicated the
channel yet, returning "channel not found".
Add a Save override to LocalCacheChannelStore that populates
channelByIdCache after a successful save, following the same pattern
used by LocalCacheTermsOfServiceStore.Save.
* MM-63056 Migrate thread menu to new component and update label
* Keep focus visible for menu buttons when menu is open and then closed
The `[aria-expanded="true"]` keeps the button highlighted while the menu
is open.
The `:focus-visible` ensures the focus stays visible when the menu is
closed. This is normally applied by the A11yController, but that gets
removed when MUI programatically moves the focus back when the menu is
closed.
* Prevent thread from being selected when thread menu is focused
* Prevent enter/space keydown events from bubbling up from menu button
* Address feedback
* Revert "Address feedback"
This reverts commit d9df624de2.
* Revert "Prevent enter/space keydown events from bubbling up from menu button"
This reverts commit aefe3792d3.
* Fix duplicated IDs between thread list menu and thread header menu
* MM-68016, MM-68017, MM-68018 Add plugin pre-hooks for channel member add, team member add, and channel archive
Enable plugins to intercept and reject (or modify) three operations before
they are persisted: adding a channel member, adding a team member, and
archiving a channel. These are the three high-priority hooks from epic
MM-68003.
* refactor(pdf_preview): migrate PDFPreview to function component
* test(pdf_preview): migrate tests to react testing library
* test(pdf_preview): store duplicate code in a constant
* chore: remove unused code
* refactor(pdf_preview): use object instead of array
Converted the caught error into a string to match onDocumentLoadError's parameter type.
* refactor(pdf_preview): wrap functions in useCallback
Added to dependencies to the useEffect attaching the scroll event listener.
* refactor(pdf_preview): remove redundant useEffect
* refactor(pdf_preview): cleanup debounced function in useEffect
* Pass all dependencies to effects that render pages
* Render pages on first load without scrolling
* Ensure loading a page doesn't overwrite results for other pages
---------
Co-authored-by: Mattermost Build <build@mattermost.com>
Co-authored-by: Harrison Healey <harrisonmhealey@gmail.com>
* MM-67158 - fix overlap in post actions menu
* use useState callback ref, keep all 9 icons in wide mode, fix a11y test
* fix unit test
* adjust e2e to new layout
* code review fixes and layout mode constants
---------
Co-authored-by: Mattermost Build <build@mattermost.com>
* Fix datepicker calendar overflow in AppsForm modal
Use fixed positioning strategy for the floating datepicker calendar
so it renders relative to the viewport instead of the modal body.
This removes the need to toggle modal-overflow on the dialog,
preserving the modal's scroll behavior while allowing the calendar
to float above the modal boundary.
* Remove extra CSS added for .modal-overflow in AppsForm
---------
Co-authored-by: Mattermost Build <build@mattermost.com>
Co-authored-by: Harrison Healey <harrisonmhealey@gmail.com>
* Fix docs-impact-review hitting max turns by adding filesystem tools
The claude-code-action workflow instructs Claude to search ./docs/source/
for existing documentation, but only grants gh and MCP tools. Claude
repeatedly attempts filesystem access, gets denied, and exhausts the
30-turn limit. Add find, grep, cat, and ls to --allowedTools so the
docs checkout can actually be searched.
* Harden docs-impact-review against prompt-injection exfiltration
- Drop find from allowedTools (unsafe due to -exec)
- Scope ls, cat, grep to ./docs/source* paths only
- Add persist-credentials: false to both checkout steps to
prevent token extraction from .git/config
* Use built-in Read/Glob/Grep tools instead of scoped Bash wrappers
Built-in tools cannot execute commands, chain via shell operators,
or abuse flags like -exec, making them inherently safer than any
Bash glob pattern for filesystem access.
* Fix nil pointer dereference in UpdateUser after store update
Add nil check on userUpdate result from userService.UpdateUser to prevent
panic when the store returns nil unexpectedly. This fixes a nil pointer
dereference that occurs when accessing userUpdate.New after the store
update call.
Sentry: MATTERMOST-SERVER-VF (14 events)
Co-authored-by: Claude <claude@anthropic.com>
* Add unit test for nil userUpdate guard in UpdateUser
Test verifies that when the store returns (nil, nil) from Update,
the app layer returns an appropriate error instead of panicking
with a nil pointer dereference.
Co-authored-by: Claude <claude@anthropic.com>
* fix: gofmt user_test.go
Co-authored-by: Claude <claude@anthropic.com>
* fix: split nil checks per review feedback, add parallel test execution
Separate userUpdate==nil from userUpdate.New==nil with distinct error
detail strings for easier debugging. Add mainHelper.Parallel(t) to test
for consistency with other mock-based tests.
Addresses review feedback from @JulienTant and @coderabbitai.
Co-authored-by: Claude <claude@anthropic.com>
---------
Co-authored-by: Claude <claude@anthropic.com>
* MM-66742 - add BoR e2e tests
* polish bor test code
* fix linter issues
* Align BoR e2e tests with Playwright page object patterns
* fix linter
* Fix sender timer assertion and sidebar nav after reload
* Refactor shared channel membership sync to use ChannelMemberHistory (MM-67616)
Replace the trigger-time membership sync mechanism with a cursor-based
approach using ChannelMemberHistory, aligning membership sync with the
established pattern used by posts and reactions.
Previously, membership changes were built into SyncMsg at trigger time
and sent via a separate TopicChannelMembership code path. This meant
removals were lost if a remote was offline, since ChannelMembers
hard-deletes rows.
Now, membership changes are fetched from ChannelMemberHistory at sync
time using the LastMembersSyncAt cursor, detecting both joins and leaves
reliably. The data flows through the normal syncForRemote pipeline
alongside posts, reactions, and other sync data.
Key changes:
- Add GetMembershipChanges store method for ChannelMemberHistory
- Add fetchMembershipsForSync and sendMembershipSyncData to sync pipeline
- Replace HandleMembershipChange with NotifyMembershipChanged (trigger-only)
- Remove conflict detection (idempotent add/remove resolves naturally)
- Remove per-user membership tracking (GetUserChanges, UpdateUserLastMembershipSyncAt)
- Add MembershipErrors to SyncResponse
- Keep TopicChannelMembership receiver for one release cycle (backward compat)
* ci: cache prepackaged plugins in mmctl tests
Cache the prepackaged_plugins directory between mmctl test runs using
actions/cache. The cache key is derived from a SHA-256 hash of the
PLUGIN_PACKAGES lines in server/Makefile, so the cache automatically
invalidates whenever any plugin version is bumped — no manual key
updates needed.
Uses actions/cache output (cache-hit) to skip the download step
entirely on cache hits.
Saves ~1-2 minutes per mmctl test run by avoiding repeated plugin
downloads.
Co-authored-by: Claude <claude@anthropic.com>
* fix: align plugin cache key with actual downloaded packages
Address CodeRabbit review: cache key was derived from all 15 plugins
in the Makefile, but the download only fetched jira plugin. Now both
the hash and the download use the same MMCTL_PLUGIN_PACKAGES value,
defined once, preventing stale cache or unnecessary invalidation.
Co-authored-by: Claude <claude@anthropic.com>
* Revert "fix: align plugin cache key with actual downloaded packages"
This reverts commit 73303864bf.
* Revert "ci: cache prepackaged plugins in mmctl tests"
This reverts commit acd1ec757a.
* prepackage the existing test plugin to speed up tests
* fix: use public key for plugin signature verification in test
---------
Co-authored-by: Claude <claude@anthropic.com>
Co-authored-by: Jesse Hallam <jesse@mattermost.com>
* feat: add CI check for test analysis
* update per comments
* address comments
* on silence mode, with general prompt and webhook message for analysis and suggestions
* update test analysis and override only via comment not label
* address reviews
* MM-67953 Changed Browse Channels modal to prioritize channels with DisplayName matching search query
* MM-67953 Changed channel shortlink autocomplete to prioritize non-member channels with DisplayName matching search query
* Shared orderByDisplayNameMatch between both methods
* Add E2E tests
* Update existing E2E test to use new fixture
* Run Prettier on E2E tests
* And fix linting issue
* And fix the fix...
* Run server code linter
* automate setup-go-work
It's all to easy to forget to `make setup-go-work`, only to run into mysterious build failures. Let's default to doing this automatically, unless `SKIP_SETUP_GO_WORK` is true (or the legacy `IGNORE_GO_WORK_IF_EXISTS`, which was oddly named, since we can't actually ignore it.)
* Make setup-go-work recipe fail-fast with set -e
Add path exclusions for markdown files, NOTICE.txt, and CHANGELOG.md
to the pull_request trigger. This prevents the full 60-minute CI suite
from running when a PR only changes documentation.
Exclusions added:
- server/**/*.md
- server/NOTICE.txt
- server/CHANGELOG.md
Co-authored-by: Claude <claude@anthropic.com>
* MM-66944 Change PlatformService.IsLeader to always be true when license doesn't support clustering
* Address feedback
* Add thorough null-checking for license
Maria A NunezandMattermost Build
Andre Vasconcelos
sabrilandMattermost Build
Carlos Garcia
Jesse Hallam
Scott Bishel
Eva Sarafianou
Miguel de la Cruz
Amy BlaisandMattermost Build
Pablo VélezandMattermost Build
Adam Schildkraut
Rajat DabadeandMattermost Build
Pavel ZemanandClaude
Harrison HealeyandMattermost Build
Devin BinnieandMattermost Build
Christopher PoileandMattermost Build
Ruturaj RathodandMattermost Build
Doug Lauder
Seongbin HongandMattermost Build
Daniel Schalla
