Fix MergeRequest#source_sha when there is no diff
Closes https://gitlab.com/gitlab-org/gitlab-ce/issues/14170
### Overview
This MR fixes an unhandled Exception when visiting the page of an open Merge Request without diff.
### Description
`MergeRequest#source_sha` is expected to return the sha of the source branch last commit. But when an open Merge Request has no diff (e.g. all commits have already been merged to the target branch), `merge_request.source_sha` incorrectly returns `nil`.
This was without consequences before – but since !2217 was merged (a few days ago), it makes `Gitlab::Git::Commit.between` raise an "Unexpected nil argument" exception. This can be reproduced when visiting the http://localhost:3000/gitlab-org/gitlab-test/merge_requests/2 page on a fresh local Gitlab setup.
This MR fixes the crash, by making sure that `source_sha` returns a
correct result even when there is no diff available. I also added tests.
@DouweM I believe you wrote most of this code in the first place ; does this looks correct to you, or is there a better way to resolve this issue maybe?
See merge request !3135
When calling Repository#build_cache we _don't_ want to build the
diverging commit count cache as doing so can be _very_ slow for
repositories with lots of branches. Because these caches are built
whenever needed (= when actually viewing the list of branches in the web
UI) we can safely remove this code from
Repository#build_cache.
The avatar logic has been moved from Project to Repository as this makes
caching easier. The logic itself in turn has been changed so that the
logo file names are cached in Redis. This cache is flushed upon pushing
a commit but _only_ if:
1. The commit was pushed to the default branch
2. The commit actually changes any of the logo files
If no branch or commit is given the cache is flushed anyway, this
ensures that calling Repository#expire_cache without any arguments still
flushes the avatar cache (e.g. this is used when removing a project).
Fixesgitlab-org/gitlab-ce#14363
Adjusted behavior so canceled builds tagged as allowed to fail do not fail build
## What does this MR do?
This diff changes the 'Allowed to fail' flag to also ignore canceled builds, whereas before, canceled builds could fail the suite even if they were marked as 'Allowed to fail'
dupe of !3258 to force a build
## Are there points in the code the reviewer needs to double check?
no
## Why was this MR needed?
Unexpected behavior as a user
## What are the relevant issue numbers?
## Screenshots (if relevant)
See merge request !3271
External Users
The user has the rights of a public user execpt it can never create a project, group, or team. Also it cant view internal projects.
Partially fix for both #4009 and #13938 (except the LDAP sync or a pattern to detect external users)
See merge request !3171
`MergeRequest#source_sha` is expected to return the sha of the source
branch last commit.
But when a open Merge Request has no diff (e.g. all commits have already
been merged to the target branch), `merge_request.source_sha`
incorrectly returns `nil`.
This was un-noticed before – but now that !2217 has been merged,
it makes `Gitlab::Git::Commit.between` raise an
"Unexpected nil argument" exception.
This fixes the crash, by making sure that `source_sha` returns a
correct result even when there is no diff available.
Support Wiki with UTF-8 page name
## What does this MR do?
Support Wiki with UTF-8 page name.
See https://github.com/gollum/gollum/pull/929
## Why was this MR needed?
Relax constraints for wiki slug in aac6598482.
It allows to create a wiki with UTF-8 name, but creating a wiki with UTF-8 name causes 500 error.
Creating a wiki with UTF-8 name once, then creating or updating wiki with ascii name also cause 500 error i.e. no one can create and update any wiki pages in the project.
## Workaround
1. Go to `https://DOMAIN/repo/wikis/git_access` -> this page display the link to git clone
2. Clone the wiki repo
3. Find the page with UTF-8 name
4. Rename or Delete these files
5. Commit and push
## What are the relevant issue numbers?
- #13979
- #13891
- #13698
- #13603
- #13317
- #12906
- #12825
- #10945
## Todo
- [x] Waiting for 'gollum-rugged_adapter' that support rugged v0.24.0 release
'gollum-rugged_adapter' gem doesn't allow to install rugged v0.24.0 (it's still beta version),
but 'gitlab_git' gem depends on rugged v0.24.0b13.
So it can't install both 'gollum-rugged_adapter' and 'gitlab_git' now.
See merge request !2999
1. Make the "subscribed" text in Issuable sidebar reflect the labels
subscription status
2. Current user mut be logged-in to toggle issue/MR/label subscription
1. Allow subscribing (the current user) to a label
- Refactor the `Subscription` coffeescript class
- The main change is that it accepts a container, and conducts all
DOM queries within its scope. We need this because the labels
page has multiple instances of `Subscription` on the same page.
2. Creating an issue or MR with labels notifies users subscribed to those labels
- Label `has_many` subscribers through subscriptions.
3. Adding a label to an issue or MR notifies users subscribed to those labels
- This only applies to subscribers of the label that has just been
added, not all labels for the issue.
Ignore eager loading in Project.search UNION
This fixes issues such as filtering groups by names on pages such as
https://gitlab.com/dashboard/groups.
See merge request !3229
Revert the avatar cropping feature
Reverts !2951 since the added 'mini_magick' dependency should be replaced by client-side croppping
See merge request !3223
The queries that are UNION'd together don't need any eager loading
(since we really only use the resulting SQL instead of having
ActiveRecord actually run the queries). By dropping any eager loaded
associations queries such as the following work instead of producing a
SQL error:
Project.all.includes(:namespace).search('foo')
Related to #2235.
This is done by:
1. Delaying the notification sending after the SSH key is commited in DB
2. Gracefully exit the mailer method if the record cannot be found
This solves the problem with caching the nil value with instance variable.
Without this the every time we ask for external_issue_tracker we built AR and potentially do SQL query
Cleanup Ci::Commit, Ci::Build and CommitStatus views
This MR tries to do first sweep of cleanups to Ci::Commit and
Ci::Build objects removing all view-related functions and fixing the
API from other side.
Fixes: https://gitlab.com/gitlab-org/gitlab-ce/issues/6046
See merge request !2760
Bring from EE: Share Project with Group
- [x] Models and migrations
- [x] Logic, UI
- [x] Tests
- [x] Documentation
- [x] Share with group lock
- [x] Api feature
- [x] Api docs
- [x] Api tests
Signed-off-by: Dmitriy Zaporozhets <dmitriy.zaporozhets@gmail.com>
For #12831
cc @DouweM @rspeicher @vsizov
See merge request !3186
- Remove unused Gitlab::Application.routes.url_helpers from Ci::Build
- Remove too much logic from a view, use Ci::Commit.matrix_builds
- Use ci_status_with_icon
- Don't describe symbols
- Remove all view related methods from Ci::Build and CommitStatus
- Remove unused Ci::Commit and Ci::Build methods
- Use polymorphism to render different types of CommitStatus
Rémy Coutable
Douglas Barbosa Alexandre
Douwe Maan
Yorick Peterse
Kamil Trzciński
Geoffrey Lalonde
Zeger-Jan van de Weg
Hiroyuki Sato
Pierre de La Morinerie
Zeger-Jan van de WegandZeger-Jan van de Weg
Timothy AndrewandRémy Coutable
Robert Speicher
Robert Speicher
Dmitriy Zaporozhets