## Context
- Removing search* integration tests instead of fixing them because they
will be replaced by global search very soon
- Fixed billing + add missing seeds to make them work
- Fixed integration tests not using consistently the correct "test" db
- Fixed ci not running the with-db-reset configuration due to nx
configuration being used twice for different level of the command
- Enriched .env.test
- Fixed parts where exceptions were not thrown properly and not caught
by exception handler to convert to 400 when needed
- Refactored feature flag service that had 2 different implementations
in lab and admin panel + added tests
- Fixed race condition when migrations are created at the same timestamp
and doing the same type of operation, in this case object deletion could
break because table could be deleted earlier than its relations
- Fixed many integration tests that were not up to date since the CI has
been broken for a while
---------
Co-authored-by: Charles Bochet <charlesBochet@users.noreply.github.com>
## Context
Field metadata service was reusing validators from
validate-**OBJECT**-metadata-input which were throwing ObjectMetadata
exceptions not handled in fieldMetadataGraphqlApiExceptionHandler and
were going to Sentry.
To solve the issue since this validator is associated with both fields
and objects I'm moving the util to the root utils folder of metadata
module and throwing a common metadata user input exception
## Context
Config was programmatically loaded in our datasources however the
default behavior of dotenv is to ignore vars if they are already
defined. This means we need to be careful about the order of env
injection and sometimes it's done at a higher level (for example
db:reset will depend on build). To make things easier I'm using the
override flag to properly override the PG_DATABASE_URL if different (and
to properly work with the 'test' DB instead of 'default' during
testing).
## Context
When REST API calls fail due to metadata cache version not found, we
throw an exception without trying to recover.
In Graphql implementation, the exception is thrown after recomputing the
cache. This PR implements the same logic for REST for users only user
the REST API.
## Context
ActivityTargetsInlineCell was not using the useIsFieldValueReadOnly hook
but a dedicated prop instead. To align with the rest of the
implementation I've updated that part of the code however we still want
the table/kanban views to always be in read-only mode regardless of the
hook logic so I've updated the hook to take the ContextStoreViewType
into account.
## Context
Workspace creation and more specifically sync-metadata performances are
bad at the moment. We are trying to identify bottlenecks and one of the
root causes is the migration runner that can take up to 10s when setting
up a new workspaces with all its tables.
First observation is we do a lot of things sequentially, mostly to make
the code easier to read and debug but it impacts performances. For
example, a table creation is done in two steps, we first ask typeorm to
create the table then ask typeorm to create columns (and sometimes
columns one by one), each instruction can take time because typeorm
seems to do some checks internally.
The proposition here is to try to merge migrations when possible, for
example when we create a table we want the migration to also contain the
columns it will contain so we can ask typeorm to add the columns at the
same time. We are also using batch operations when possible (addColumns
instead of addColumn, dropColumns instead of dropColumn)
Still, we could go further with foreign keys creations or/and try with
raw query directly.
## Test
New workspace creation:
See screenshot, 9865.40233296156ms is on main, the rest is after the
changes:
<img width="610" alt="Screenshot 2025-02-28 at 09 27 21"
src="https://github.com/user-attachments/assets/42e880ff-279e-4170-b705-009e4b72045c"
/>
ResetDB and Sync-metadata on an existing workspace commands still work
## Context
Removing the ability to assign yourself from the UI. The backend already
checks that. This is because a member can only have one role for the V1
of permissions
Took the opportunity to move some roles related components in dedicated
folders
## Context
Following the strategy where we want to block custom object creation
when the type is reserved by core objects. The issue happened again with
the recently introduced role table.
## Context
Adding a defaultRole to each workspace, this role will be automatically
added when a member joins a workspace via invite link or public link
(seeds work differently though).
Took the occasion to refactor a bit the frontend components, splitting
them in smaller components for more readability.
## Test
<img width="948" alt="Screenshot 2025-02-24 at 14 54 02"
src="https://github.com/user-attachments/assets/13ef1452-d3c9-4385-940c-2ced0f0b05ef"
/>
## Context
With the new permissions system, we now need to hide some items from the
settings navigation and gate some routes so they can't be accessed
directly.
To avoid having to set permission gates in all the component pages, I'm
introducing wrapper at the route level and in the Navigation. This is
not required and is mostly for pages that are strictly mapped to a
single permission, for the rest we still need to use the different hooks
manually but it should avoid a bit of boilerplate for most of the cases.
- currentUserWorkspaceState to access settingsPermissions
- SettingsProtectedRouteWrapper in the router that can take a
settingFeature or a featureFlag as a gate logic, if the currentUser does
not have access to the settingFeature or the featureFlag is not enabled
they will be redirected to the profile page.
- SettingsNavigationItemWrapper & SettingsNavigationSectionWrapper. The
former will check the same logic as SettingsProtectedRouteWrapper and
not display the item if needed. The later will check if all
SettingsNavigationItemWrapper are not visible and hide itself if that's
the case.
- useHasSettingsPermission to get a specific permission state for the
current user
- useSettingsPermissionMap to get a map of all permissions with their
values for the current user
- useFeatureFlagsMap same but for featureFlags
## Context
Regression was introduced 3 weeks ago when we added relations v2.
Because the relation logic is recursive during the life of a request, we
were querying the featureFlags many times.
We are now always using the featureFlag map and it's now available in
the base resolver so we don't need to query it everywhere, preferably
passing it as a parameter instead.
Note: We should introduce a cache for featureFlags in the future, this
is something easy to control and invalidate when needed.
## Context
Integration test ci was not running properly. This PR should fix that
Todo: This was not running for a while, many tests are not succeeding so
I'll try to fix them in this PR now that the action is running properly.
## Context
In some CustomException exceptions, we were instantiating a code without
initializing it which was overriding the parent code and it was then
lost when retrieving it in filters.
Removing them to make sure we don't reproduce this pattern
## Context
Introducing the "Permissions" tab in the role page
Next: Need to address some css improvements, some components might be
reusable and it still does not fully match the figma (icon missing for
permission types for example). We decided to merge like this for now so
we have something functional and I will update the code in an upcoming
PR
<img width="633" alt="Screenshot 2025-02-12 at 13 54 16"
src="https://github.com/user-attachments/assets/762db5d7-e0a6-4ee1-b299-24de6645bad1"
/>
## Context
We recently introduced a createMany on the field metadata service to
improve seeding performances. This broke relation metadata creation
because it was using a method with the same name that was inherited from
TypeOrmQueryService.
## Context
This PR adds a new SettingsRoleEdit page, the existing roles page now
redirects to the role edition page when clicking on it.
For now, we can't edit anything. Next step is to allow role assignment
in the corresponding tab.
<img width="941" alt="Screenshot 2025-02-05 at 17 16 14"
src="https://github.com/user-attachments/assets/ca46de15-6237-4de6-88e1-2384a09d4a27"
/>
## Context
Not exposed in the API yet, this new method allows us to reduce the time
to create multiple fields at once, mostly during seeding. This allows us
to batch transactions and avoid recomputing the cache everytime.
With this change, we recompute the cache 7 times instead of 35 during
seeding. We could do the same for objects.
## Context
This PR introduces a new Roles settings page, accessible through the
settings menu when the isPermissionsEnabled feature flag is enabled. The
page provides a foundation for managing user roles within the workspace.
This is not fetching the roles from the BE for now and will be done in a
followup PR.
## Context
This config tells bullmq what to do with jobs once they have been
processed.
We used to retain 100 completed jobs and 500 failed jobs. We never
really needed that much retention and keeping succeeded job is not
useful either in our case. I'm updating the strategy so all completed
jobs are removed from the queue and we now keep 100 failed jobs in case
we need to debug. (Although we should have good monitoring with
ExceptionHandler driver)
## Context
avatarUrl is a TEXT field type so non nullable, which means if we try to
run a mutation with null it will either fail or be ignored. Here this is
the second option, done in sanitizeRecordInput where when a
fieldMetadata has isNullable=false and the value is null, we return
undefined. This caused the mutation to send an empty input and not
remove the avatar
## Context
<img width="1349" alt="Screenshot 2025-01-13 at 17 18 24"
src="https://github.com/user-attachments/assets/4f5da0e9-0245-41c6-bde2-4d52e0ba34ed"
/>
Feature flags are stored in DB and then cast as FeatureFlag gql type
from its corresponding enum.
This means if a value from the DB does not match that enum type, the gql
server will reject the call when returning the object in the resolver.
(see screenshot above)
To solve that, we want to do 2 things:
- The ORM should still return the feature flag even if it's not valid,
this is actually in the DB so we don't want to "hide" that, however we
now have a warning message.
- The service is not changed for the same reason, the limitation comes
from gql behaviour so this is not the goal of the service nor the ORM to
act on it (except the warning message)
- The resolver should be updated, here we want to filter-out non-valid
feature flags so it does not break the API.
Because featureFlags used to be auto-generated by nestjsquery and we
want to change its behavior, I had to manually create a resolveField for
featureFlags and remove the auto-generated one. That means we lose some
features such as filter/sort coming from nestjs-query pagination (which
is something we will want to implement once we will remove nestjs-query
but that's a whole other subject)
Script is using here-string <<< which is not supported by sh, using bash
instead
Also removing "sh" from the command to actually use the bash from the
shebang
## Context
Following this https://github.com/twentyhq/twenty/issues/4199
This has not been fully implemented, after 5months of dead code I'm
removing the feature for the time being until we re-prioritise the
feature (unlikely during these next 6 months) to keep the codebase a bit
cleaner (no need to maintain dead features)
Feel free to reopen / revert this PR once feature is ready
## Test
locally after importing emails
## Context
2 issues here:
- We use a metadata repository find method without providing a
workspaceId: In practice this is not an issue in this specific part but
let's avoid that pattern
- await this.workspaceService.deleteWorkspace(workspaceId); deletes
almost everything, emitting an event on workspaceMember could
potentially bring issues and not much values imho
Restoring https://github.com/twentyhq/twenty/pull/9185
Also fixing sync-metadata with test values in jsonb
## Test
sync-metadata on existing workspaces should replace colorSchema in both
metadata and workspaceMember tables
Fixes https://github.com/twentyhq/twenty/issues/8863
## Description
This PR fixes an issue with cursor-based pagination when dealing with
composite fields (like `fullName`). Previously, the pagination direction
was incorrectly determined for composite fields because the code wasn't
properly handling nested object structures in the `orderBy` parameter.
Refactored the code accordingly.
We have conflicts which prevents us from running demo seed command after
a database-reset or/and if dev seed has already been populated. This is
because we are trying to create Tim user twice (but due to the
insert()...onConflict() in the seed creation, it does not create the
second one), for seed-dev and seed-demo .
This PR fixes that by using the same user id, allowing us to use the
same Tim apple for both dev/demo seeds
Test
<img width="802" alt="Screenshot 2024-12-16 at 15 36 56"
src="https://github.com/user-attachments/assets/72244978-130f-4561-8709-43376453b247"
/>
<img width="780" alt="Screenshot 2024-12-16 at 15 37 17"
src="https://github.com/user-attachments/assets/01ce221b-34f6-4e48-ae30-d84bdf9d3cc2"
/>
Fixes https://github.com/twentyhq/twenty/issues/8810
Fixes https://github.com/twentyhq/twenty/issues/5268
Fixes https://github.com/twentyhq/twenty/issues/8971
- Fixing Task/Note creation not sending position during creation
- Adding a command to backfill position being null, using existing
backfill command.
- Removed unused backfill job.
- Updated workspace entities to set position non-nullable and set a
default value to make it non-required on the API
- Updated position factory to set a default position for all objects
having a POSITION field instead of only company/people
- Moved the try/catch in each resolver factory calling
GraphqlQueryRunnerException handler, makes more sense to call it in the
actual graphql-query-runner and removing some duplicate codes
- Adding validations for input in QueryRunnerArgs factories
- Allow sync-metadata to override and sync defaultValues for certain
field types (that can't be updated by users)
- Removing health-check from sync-metadata command during force mode to
improve performances
removeBillingFKWithCore migration is in the wrong folder and is not
executed as intended.
Moving to billing folder to fix that and to be only run in billing mode
## Context
Fix add-subdomain-to-workspace command not included in global module
also fixing the command regex logic that was not generating subdomain
properly
## Context
Fix wrong unique index on user email as we don't want an index on
email/deletedAt but a partial on a where condition on deletedAt. This
should enforce email unicity excluding the ones that have a deletedAt
## Test
Run
```sql
SELECT email, COUNT(*) as duplicate_count
FROM core."user"
WHERE "deletedAt" IS NULL
GROUP BY email
HAVING COUNT(*) > 1
ORDER BY duplicate_count DESC;
```
to check duplicates before running the migration
## Context
Fix seed demo command since we added a subdomain non-nullable column we
needed to update the command to fill that info for the demo workspace
## Context
The recent addition of object renaming introduced issues with enum
names. Enum names should follow the pattern
`${schemaName}.${tableName}_${columnName}_enum`. To address this, and to
allow users to customize the API name (which is included in the enum
name, columnName), this PR implements behavior similar to object
renaming by introducing a `isLabelSyncedWithName` boolean.
<img width="624" alt="Screenshot 2024-12-02 at 11 58 49"
src="https://github.com/user-attachments/assets/690fb71c-83f0-4922-80c0-946c92dacc30">
<img width="596" alt="Screenshot 2024-12-02 at 11 58 39"
src="https://github.com/user-attachments/assets/af9a0037-7cf5-40c3-9ed5-d51b340c8087">
## Context
Enum are named after this pattern
`${schema}_${tableName}_${columnName}_enum` however now that we allowed
table name update, we need to make sure their enums are renamed as well.
Fixes https://github.com/twentyhq/twenty/issues/8760
## Context
Index names are based on table names and column names, which means
renaming an object (or a field) should also trigger a recompute of index
names. As of today it raises a bug where you can't create an object with
a name that was previously used.
I also took the occasion to refactor a bit the part where we create and
update (after renaming) relations. Basically the only relations we want
to affect are standard relations so I've aligned the logic with
sync-metadata which uses standardId as a source of truth to simplify the
code.
Note: We don't create index for custom relations
Next step should be to do that and update that code to update the index
name as well. Also note that we need to update the sync-metadata logic
for that as well
## Context
Some mutations are not working properly, workspaceMember soft deletion
for example. workspaceMember being a camelCase table name, it's probably
not propagated properly to pgql (which needs double quote for the table
name to keep it as camelCase)
I didn't have time to dig too much but if the `where` is before
`softDelete`, the query is `WHERE workspaceMember.id = $1` while if it's
after, the query becomes `WHERE id = $1`.
Probably due to the fact that once you call delete/softDelete/update,
the standard builder (SelectQueryBuilder) becomes a
DeleteQueryBuilder/etc... and filters are not handled the same way.
## Context
We recently added a command to ensure uniqueness on the viewId column in
the viewField table. This created some issues for some old workspaces
that had viewFields with an empty viewId.
This command should get rid of those and set the column as non-nullable.
Also updating the onDelete action accordingly and set one missing for
FavoriteFolder
## Context
Now that each operation has its own resolver, we need to make sure they
all map to query arg getters. CreateOne was not properly mapped to the
position getter which made record creation fail because "position:
first" was not properly converted to a float.
Also fixing queries with custom object where we were wrongly using the
table name instead of entity name
Fixes https://github.com/twentyhq/twenty/issues/8300
## Context
API events were created too late and were already formatted as Gql
responses (including nesting with edges/node/type + formatting that
should not exist in an event payload). This PR moves the emit logic to
the resolver where we actually do the DB query
Note: Also added RESTORED events
## Context
- Fixing folder structure where 0.33 was inside 0.32.
- Updating the command so it runs enforce uniqueness on all tables
- Updating workspaces entities so the sync metadata adds the index
## Context
The object metadata map is becoming quite large and its structure could
be simplified.
We are removing byNameSingular/byNamePlural keys, the former can be
retrieved through a new helper and the latter is not used in the code
base currently.
## Context
We recently added an infinite ttl to metadata version however other keys
such as the object metadata map are also linked to the version so
keeping the version in cache without the rest breaks the app. I'm
editing all engine related keys with infinite ttl
