Extracts group query into a CTE so that it can be reused by the UNION
querying project-level and group-level todos.

Changelog: performance

Replace `$gray-darkest` SCSS variable with `$gray-200`

This reverts merge request !152152

This was already no-oped in
https://gitlab.com/gitlab-org/gitlab/-/merge_requests/131717

Changelog: removed

We now import components imported from MR settings.

When a custom role is assigned, display the
custom role's title in access granted notifications.

Changelog: changed

Fixes a bug where a user could not leave a comment on a whitespace only change,
with the `Show whitespace changes` checkbox unchecked and after clicking
the `Show changes` button.

https://gitlab.com/gitlab-org/gitlab/-/issues/442351

Removed the feature flag and updated the docs accordingly

Changelog: removed

See https://docs.gitlab.com/ee/development/database/avoiding_downtime_in_migrations.html#remove-the-trigger-and-old-integer-columns-release-n--2

Changelog: removed

Replace the Zoekt feature flags with zoekt_settings.
Add a new admin setting section Exact Code Search where zoekt
related settings can be managed

Changelog: added
MR: https://gitlab.com/gitlab-org/gitlab/-/merge_requests/148378

Changelog: changed

To better align with our usage of GraphQL, mlModelDestroy needs to be
changed to mlModelDelete and the return type needs to be the model. The
first step is to add new mutation mlModelDelete so that we can then
change the frontend, otherwise the frontend can be deployed beforehand
and will try to access an mlModelDelete that doesn't exist yes.

Deleting mlModelDestroy and changing the frontend to call mlModelDelete
are followups.

The data is used for defining API arguments
https://gitlab.com/gitlab-org/gitlab/-/issues/348208#note_1908043184.

Review feedback

This is part 1 of 2 towards
implementing a new application
setting that allows admins to disallow
users from creating private profiles.

The follow up MR will implement the
setting on the user profile UI.

- Create new board options dropdown
- Add show labels toggle to board options dropdown
- Add epic swimlanes toggle to board options dropdown

Changelog: changed

MR: gitlab.com/gitlab-org/gitlab/-/merge_requests/152663

Changelog: fixed

Vue.set is no longer supported in Vue 3
has to be replaced with common mutation

This feature flag has been enabled by default for more than a year

Changelog: other

User clicks a project result in the command palette

The new worker has a deduplication strategy of until_executing instead
of until_executed

This commit changes the logic behind the alert message
for invalid GPG keys associated with a user.

Changelog: changed

When you rename a wiki page now, a redirection table is created
to map the old path to the new path, so all existing links will work.

Changelog: added

Tailwind equivalent is `gl-gap-y-*`

- iteratively building up the duo pro trial status widget.
- modeled off of the ultimate trial status widget.
- see https://gitlab.com/gitlab-org/gitlab/-/issues/457265

Changelog: added
EE: true

CI/CD settings were moved to a vue table in
https://gitlab.com/gitlab-org/gitlab/-/merge_requests/41864. We
introduced the graphql endpoint to update a trigger token in
https://gitlab.com/gitlab-org/gitlab/-/merge_requests/126537.
This MR streamlines the trigger token edit flow to use a pop up
modal and call the GraphQL endpoint instead of taking users to a
separate page.

Changelog: changed

Changelog: changed

Changelog: other

Introduces new component usage aggregator and cursor classes
to be used later in a service scheduled by cron job. The objective is
to run a regular worker that aggregates the CI component usage data
in the background, avoiding long running queries.

The Usages::Aggregator class iterates through the target scope in
batches. For each target ID, it aggregates the component usage
count in batches for the given window. It maintains a Redis cursor
so that the counter can resume from it left off on each run and
avoid unnecessary reprocessing. We collect the count in Rails
because the SQL query `COUNT(DISTINCT(*))` is not performant when
the dataset is large. This class should be run with an exclusive
lease for each unique cursor Redis key.

Changelog: fixed