Donald Cook authored 2 years ago and Florie Guibert committed 2 years ago

Removed mocks, add parent to work item

Cleaned up buttons and styling

UX review cleanup

Added tests

Fixed some merge conflicts

Mikołaj Wawrzyniak authored 2 years ago and Doug Stull committed 2 years ago

measure_service_ping_metric_collection feature was defualt enabled in
15.0, no issues were reported. It is time to remove feature flag and
release feature.

Changelog: other

Vladimir Shushlin authored 2 years ago and Andrew Fontaine committed 2 years ago

When users click on "Create release" on the tag page,
we want to direct them to the "new release" form with
tag_name preselected.

This MR:
* passes tag_name from URL parameter to the form
* explicitly sets "existing_release" if form is an edit form
    (previously we used `tag_name` for this)
* hides the "Create from" from field if we set `tag_name` using URL parameter

This change adds a new search filter so users can filter their runners
by their upgrade status.

This addition is available to GitLab Ultimate users.

Changelog: added
EE: true

Bitbucket import bypasses issue model callbacks, so we
need to explicitely pass the work item type as part of the
creation attributes

Jan Provaznik authored 2 years ago and Peter Leitzen committed 2 years ago

Because "create from task" feature is available both for Issue and
Incident issue types, we allow adding parent links also for incidents.

The toggle 'Reject packages with the same name and version'
to affect duplicate package handling to the package registry
was inverted and did not match the description of the option.

#367659

Changelog: fixed

Jan Provaznik authored 2 years ago and Alexandru Croitor committed 2 years ago

Updates task creation to create rather parent/child relationship if
parent/child is supported by the original work item (IOW if it's an
Issue).

Simon Knox authored 2 years ago and Natalia Tepluhina committed 2 years ago

Copy assignees component and update a little

Most problems come due to a breaking change in Jest 27
and new fake timers implementation. Additional extra tick
happens between Jest event hook. For instance, if our test
validates the component loading state and we create a component
in "beforeEach" hook the loading state will go away
due to an extra tick in between "beforeEach" and "it" blocks.

Another problem is changed Jest matchers for string and numbers.
Jest 27 validates the type by default when we use matchers like
"toContain", "toBe". The solution is to call "toString" function
on expected values as VTU and Vue.js in general renders everything
as strings.

This method is not used anywhere so we can remove it

Serena Fang authored 2 years ago and Alexandru Croitor committed 2 years ago

Changelog: changed

When listing parent/children for a work item, we check only
generic work_item feature flag. This is little bit confusing (as for
creating/updating hiearchy widget we still use work_item_hiearchy FF),
but the reason is that parent/child relationship can be created by
"create from task" functionality and we still need to be able to list
parent for the tasks created this way.

Added nil check

Changelog: fixed

These don't follow the pattern of our services where they inherit from
BaseService or other related classes.

Mireya Andres authored 2 years ago and Scott Hampton committed 2 years ago

Adds an easy way to copy the path of all tests with failures so users
can rerun them locally. The button can be found in the test results of
the MR widget.

Also adds a copy-to-clipboard button for individual tests in the test
case modal.

Changelog: added

James Fargher authored 2 years ago and Albert committed 2 years ago

SeedHelper uses git to directly insert repositories into git. Soon we
will be using the praefect database in tests and when this happens we
will not be able to directly manipulate repositories in gitaly without
praefects knowledge.

The repository used by the project factory has slighty different files
than the repository used by SeedHelper. It does not affect the function
of the tests.

Coung Ngo authored 2 years ago and Simon Knox committed 2 years ago

Also move the ellipsis button from next to the title to above the
title

Behind feature flag `work_items`, default disabled

#366210

Doug Stull authored 2 years ago and Peter Hegman committed 2 years ago

- show it in an alert

Changelog: changed

This allows us to pass the rewritten params to the create service so
that system notes will be created when `skip_system_notes: false`

This does not really fit under AttributesRewriter because we're not
rewriting issuable attributes but we're copying resource events which
are rendered as system notes

Also fixes the "cloned from" system note having an incorrect timestamp

Changelog: fixed

Review feedback and migrate more errors

Natalia Tepluhina authored 2 years ago and Simon Knox committed 2 years ago

- replaced mock assignees with real query
- implemented logic for multiple assignees

In specs, all feature flags are enabled by default, which led to to
GITALY_DISABLE_REQUEST_LIMITS previously ignored. Which means it was
hard to temporarily skip gitaly max call errors. So, we now process
GITALY_DISABLE_REQUEST_LIMITS first.

This asserts that tags are not returned when searching for related
branches.

Stan Hu authored 2 years ago and Mayra Cabrera committed 2 years ago

Previously if restricted e-mail domains
(#366269) or password
limit restrictions were set
(#22399), newly-signed in
LDAP users would see an obscure failure message:

```
Could not authenticate you from Ldapmain because "Undefined method
`provider' for nil:nilclass".
```

This was failing in `Gitlab::Auth::Ldap::Access.open` because
`user.ldap_identity` was `nil`. The application settings would cause
validations of newly-signed in LDAP users to fail, preventing the user
from persisting to the database.

The OmniAuth callback controller attempted to detect this by calling
`auth_user.valid_sign_in?`, but this was failing because
`Gitlab::Auth::LdapUser#valid_sign_in?` did not properly check that
the user entry was persisted to the database before it contacted the
LDAP server.

To avoid this obscure error, we just need to swap the order of the
check: if the user entry is valid and persisted, then we can safely
contact the LDAP server. If the user entry is not valid, an "Access
Denied for your LDAP account" message will be shown to the user, and a
message in the application log will list the reason why it failed.

Changelog: fixed

Harsimar Sandhu authored 2 years ago and James Fargher committed 2 years ago

EE: true
Changelog: added

Changelog: changed

- Campfire
- Datadog
- Irker
- Packagist
- Pushover
- Shimo
- Youtrack
- Zentao

Epic: gitlab-org&7652