- to owners only
- hide to all other users
- add tests

Update Pikaday colors for dark mode

See merge request gitlab-org/gitlab!65247

Report to Sentry for long Redis durations

See merge request gitlab-org/gitlab!67512

Moved Source Editor's theming into utilities

See merge request gitlab-org/gitlab!67916

Fix expired artifacts causing dependency failures on locked pipelines

See merge request gitlab-org/gitlab!67764

Merge branch '323245-provide-terraform-backend-environment-variables-in-operations-terraform-ui' into 'master'

Provide Terraform backend environment variables in Operations-->Terraform UI

See merge request gitlab-org/gitlab!67417

Finalise package details refactor

See merge request gitlab-org/gitlab!67991

Raise error when using invalid relation type in member finder

See merge request gitlab-org/gitlab!67849

Updated reporting period message on DevOps Adoption Report

See merge request gitlab-org/gitlab!68034

checks: Fix combinatorial explosion in `#commits_for()`

See merge request gitlab-org/gitlab!67827

Revert "Merge branch '337719-linear-project-ancestors' into 'master'"

See merge request gitlab-org/gitlab!68051

Change Feature Flag group::ecosystem to group::integrations

See merge request gitlab-org/gitlab!67963

The function `#commits_for()` has been introduced via 156ce433
(checks: Implement infrastructure for bulk diff checks, 2021-07-29) in
order to allow bulk-loading of commits. What this function does is given
a set of new commits and a specific object ID, it will do a graph walk
of these new commits starting from this ID in order to extract only
those commits which have been newly introduced via this object ID.

As it turns out though, the implementation has a bug which causes
combinatorial explosion: if a commit is reachable via multiple commits,
then it will be walked and returned multiple times. This can happen if
there are merge commits, where we'll now repeatedly walk all common
ancestors of both commits. In case these again contain merge commits,
the common ancestors again get walked multiple times. The result is that
commits get walked exponentially many times. For the following graph
with criss-cross merges, this causes us to return 768 commits instead of
the expected 18:

      o---o---o---o---o---o---o---o
     / \ / \ / \ / \ / \ / \ / \ / \
    o   X   X   X   X   X   X   X   o
    \  / \ / \ / \ / \ / \ / \ / \ /
      o---o---o---o---o---o---o---o

Fix the bug by removing seen commit IDs from the hash tracking commits
by their object ID. Furthermore, the pending queue is converted to a set
such that we don't re-add IDs which we have already seen before such
that it doesn't exhibit exponential growth.

Changelog: fixed

Set 36h limit for reviewer to respond when at capacity

See merge request gitlab-org/gitlab!67481

This reverts merge request !67565

Message now more clearly shows that the update is automatic at the
beginning of the month, and reflects the previous calendar month.

Fixes https://gitlab.com/gitlab-org/gitlab/-/issues/335640

Changelog: changed
EE: true
Signed-off-by: Raimund Hook <raimund.hook@exfo.com>

Add performance_indicator_type metric

See merge request gitlab-org/gitlab!68043

Move Metric life cycle docs in a seprate page

See merge request gitlab-org/gitlab!67968

Add Analytics menu

See merge request gitlab-org/gitlab!66904

`#commits_for()` has a bug where commits are returned multiple times in
case they're reachable via multiple commits, leading to combinatorial
explosion. Add a testcase which demonstrates this bug.

Fix: Code block input rule doesn’t save the code block language

See merge request gitlab-org/gitlab!67980

Add support for CI_COMMIT_REF_NAME variable in includes

See merge request gitlab-org/gitlab!67902

Remove `ref_name` parameter in `RemoteMirror`

See merge request gitlab-org/gitlab!67668

Remove ff_evaluate_group_level_compliance_pipeline feature flag

See merge request gitlab-org/gitlab!67984

Remove agent_kubeconfig_ci_variable feature flag

See merge request gitlab-org/gitlab!67734

Migrate dashboard_resize_browser_spec to feature test

See merge request gitlab-org/gitlab!67868

Add performance_indicator_type metric

See merge request gitlab-org/gitlab!68033

On-disk remotes are deprecated, and in fact we don't create any remotes
in GitLab anymore in favor of using in-memory remotes, where we just
pass the URL to Gitaly. We thus don't need to pass information about
remote names anymore given that it's unused nowadays.

Remove the `ref_name` parameter in `RemoteMirror` to reflect this. Note
that this parameter is misnamed, and should really have been called
`remote_name`. Furthermore, drop now-unneeded infrastructure which was
required to track the remote name.

Only dispatch auto-tracked "render" events once per page load

See merge request gitlab-org/gitlab!67275