It might happen you want to make the reference column have a unique
value, or you want to create partial indexes. So instead of only
accepting a `true` value, also accept a hash of options.

Convert "New menu" feature spec to a view spec

Closes #54157

See merge request gitlab-org/gitlab-ce!23355

Speed up "show GPG badge" feature specs

See merge request gitlab-org/gitlab-ce!23359

Resolve "Warn in Web Editor when user navigates away"

Closes #53728

See merge request gitlab-org/gitlab-ce!23004

Merge branch '52828-inconsistency-in-fonts-used-for-branch-name-and-create-from-fields-when-creating-new-branch-from-ui' into 'master'

Resolve "Inconsistency in fonts used for branch name and create from fields when creating new branch from UI"

Closes #52828

See merge request gitlab-org/gitlab-ce!23120

Don't use rugged write-ref anymore

See merge request gitlab-org/gitlab-ce!23286

Gracefully handle references with null bytes

Closes #54466

See merge request gitlab-org/gitlab-ce!23365

[Docs] Update issue boards images for recent cards redesign

See merge request gitlab-org/gitlab-ce!23321

Reword Knative cluster application

See merge request gitlab-org/gitlab-ce!23289

Eliminate duplicated words (for docs)

See merge request gitlab-org/gitlab-ce!23183

improve browse JUnit details

See merge request gitlab-org/gitlab-ce!23257

Fix broken links in CI help page

See merge request gitlab-org/gitlab-ce!22920

standardize periods & fix typo

See merge request gitlab-org/gitlab-ce!23294

CE port of 'Move EE specific code from Gitbab::Database into ee'

See merge request gitlab-org/gitlab-ce!23361

Fix handling of filenames with hash characters in tree view

Closes #54473

See merge request gitlab-org/gitlab-ce!23368

Addressable::URI interprets the `#` in a URI as a URI fragment
and does not escape it, but Rails has special helpers that treats
these as bona-fide characters that need to be escaped.

Closes https://gitlab.com/gitlab-org/gitlab-ce/merge_requests/23368

`Rugged::Reference.valid_name?` used in
`Gitlab::GitRefValidator.validate` fails on strings containing null
bytes because it uses `StringValueCStr()`. Per
https://silverhammermba.github.io/emberb/c/:

Ruby’s String kinda corresponds to C’s char*. The simplest macro is
StringValueCStr() which returns a null-terminated char* for a
String. The problem here is that a Ruby String might contain nulls - in
which case StringValueCStr() will raise an ArgumentError!

Closes https://gitlab.com/gitlab-org/gitlab-ce/issues/54466

Fix deadlock on ChunkedIO

See merge request gitlab-org/gitlab-ce!23329

Rails 5 deprecation: Passing an argument to force an association to reload is now deprecated

See merge request gitlab-org/gitlab-ce!23337

Allow profiler to authenticate by stubbing users directly

Closes #54327

See merge request gitlab-org/gitlab-ce!23320

Backport of gitlab-ee!8470

See merge request gitlab-org/gitlab-ce!23150

Add events index on project_id and created_at

Closes #53992

See merge request gitlab-org/gitlab-ce!23354

Upgrade better_errors gem to 2.5.0

See merge request gitlab-org/gitlab-ce!23312

After converting the test that verifies this partial's behavior from a
feature spec to a view spec, we're able to remove a few CSS classes that
were only used for scoping tests found in the feature spec. All the view
spec needs is the `have_link` matcher which allows us to check for a
specific `href` attribute.

A feature spec to test this simple behavior takes about 2 minutes to run
in CI. Everything it's testing is conditionals and `href` attributes,
which can easily be done in a view spec that runs in about 8 seconds.

Fix Image Lazy Loader for some older browsers

Closes #54407

See merge request gitlab-org/gitlab-ce!23349

Override CI_COMMIT_REF_SLUG for QA branches

See merge request gitlab-org/gitlab-ce!23346

Use shelling out to git to write refs instead of rugged, hoping to
avoid creating invalid refs.

To update HEAD we switched to using `git symbolic-ref`.

It's more semantically accurate for what we're verifying.

Previously, we wasted time creating a maintainer and signing them in,
then loaded the entire commit list just to verify the GPG status of a
single commit.

Now, we rely on the project being public so no user needs to be
authenticated, and load just the page for the commit we care about.

We're trying to give the arbitrary SHAs required by each spec a
meaningful name.

This also adds an explicit `ref` definition to each spec so we're not
dealing with a mystery guest.

Previously, this group of specs would create a new user, add them as a
maintainer, sign them in, load the project's commit list, click a
specific GPG key badge, and then verify the content in the popover.

Now, we make the project public in order to skip the user setup and
login, and load a specific commit page because it loads faster than the
commit list while still showing the GPG badge.

CE port of "Move merge request approval settings"

See merge request gitlab-org/gitlab-ce!23157