This reverts commit 530f5158.

See !4892.
Signed-off-by: NRémy Coutable <remy@rymai.me>

preventing settings from being saved

This reverts commit 9ca633eb, reversing
changes made to fb229bbf.

- Cherry-picked from `mvestergaard:branch-protection-dev-merge`
- https://gitlab.com/gitlab-org/gitlab-ce/merge_requests/4220

fixes a few issues to do with import_url not being saved correctly for imports. This should prevent the import_data to be created when it should not and output an error properly validating before creating it.

Some context:
http://doc.gitlab.com/ce/api/repository_files.html#get-file-from-repository
http://doc.gitlab.com/ce/api/repositories.html#list-repository-tree

Slack:
https://gitlab.slack.com/archives/questions/p1467890450002077

1. The `open_branches` method is used to provide a list of branches
   while creating a protected branch.

2. It makes sense to include branches which are matched by one or more
   wildcard protected branches, since the user might want to make exact
   protected branches from these as well.

3. This also provides a large performance improvement. On my machine, in
   a project with 5000 branches and 2000 protected branches, the
   `ProtectedBranches#index` page went from a 40 seconds load time to 4
   seconds (10x speedup).

`avatar.present?` goes through CarrierWave, and checks that the file
exists on disk and checks its filesize. Because we're hitting the disk,
this adds extra overhead to something where the worst-case scenario is
rendering a broken image.

Instead, we now just check that the _database attribute_ is present,
which is good enough for our purposes.

See https://gitlab.com/gitlab-org/gitlab-ce/issues/19273

1. The main implementation is in the `ProtectedBranch` model. The
   wildcard is converted to a Regex and compared. This has been tested
   thoroughly.

    - While `Project#protected_branch?` is the main entry point,
      `project#open_branches` and
      `project#developers_can_push_to_protected_branch?`
      have also been modified to work with wildcard protected branches.

    - The regex is memoized (within the `ProtectedBranch` instance)

2. Improve the performance of `Project#protected_branch?`

    -  This method is called from `Project#open_branches` once _per branch_
       in the project, to check if that branch is protected or not.

    -  Before, `#protected_branch?` was making a database call every
       time it was invoked (in the above case, that amounts to once
       per branch), which is expensive.

    -  This commit caches the list of protected branches in memory, which
       reduces the number of database calls down to 1.

    -  A downside to this approach is that `#protected_branch?` _could_
       return a stale value (due to the caching), but this is
       an acceptable tradeoff.

3. Remove the (now) unused `Project#protected_branch_names` method.

    - This was previously used to check for protected branch status.

And create new Project#requesters, Group#requesters scopes.
Signed-off-by: NRémy Coutable <remy@rymai.me>

Requested in #14669

Added note to changelog

This fix avoids exposing the information from the wiki repository of other project.

This ensures that Project.find_with_namespace returns a row matching
literally as the first value, instead of returning a random value.

The ordering here is _only_ applied to Project.find_with_namespace and
_not_ Project.where_paths_in as currently there's no code that requires
Project.where_paths_in to return rows in a certain order. Since this
method also returns all rows that match there's no real harm in not
setting a specific order either. Another reason is that generating all
the "WHEN" arms for multiple values in Project.where_paths_in becomes
really messy.

On MySQL we have to use the "BINARY" operator to turn a "WHERE" into a
case-sensitive WHERE as otherwise MySQL may still end up returning rows
in an unpredictable order.

Fixes gitlab-org/gitlab-ce#18603

This reverts commit 13e37a3e.

This ensures that code such as this don't run needless SQL queries:

    project.gitlab_issue_tracker_service.project

This also means that if the root `project` eager loads any associations
the Service object will be able to re-use those.

adapted current services stuff to use new project import, plus fixes a few issues, updated routes, etc...

+ Move 'Edit Project/Group' out of membership-related partial
+ Show the access request buttons only to logged-in users
+ Put the request access buttons out of in a more visible button
+ Improve the copy in the #remove_member_message helper
Signed-off-by: NRémy Coutable <remy@rymai.me>

Signed-off-by: NRémy Coutable <remy@rymai.me>

Using update_column to store the boolean flag to avoid
any side effects with the current state of the project
instance

This method can be used to find multiple projects for multiple paths.
For example, take this snippet:

    Project.where_paths_in(%w{gitlab-org/gitlab-ce gitlab-org/gitlab-ee})

This will return an ActiveRecord::Relation containing the GitLab CE and
GitLab EE projects.

This method takes care of matching rows both case-sensitively and
case-insensitively where needed.

Project.find_with_namespace in turn has been modified to use
Project.where_paths_in without nuking any scoping (instead it uses
reorder(nil)). This means that any default scopes (e.g. those used for
"pending_delete" stay intact).

The method Project.where_paths_in was added so the various Markdown
filters can use a single query to grab all the projects referenced in a
set of documents, something Project.find_with_namespace didn't allow.

This reverts commit 3e991230.

# Conflicts:
#	app/models/project.rb