Brett Walker authored 5 years ago and Mayra Cabrera committed 5 years ago

- Due to https://github.com/exAspArk/batch-loader/pull/32,
we  changed BatchLoader.for into BatchLoader::GraphQL.for
- since our results are wrapped in a BatchLoader::GraphQL,
calling `sync` during authorization is required to get real object
- `graphql` now has it's own authorization system.  Our
`authorized?` method conflicted and required renaming

A new `discussion_id` argument on the `createNote` mutation allows
people to create a note within that discussion.

The ability to lazy-load Discussions has been added, so
GraphQL.object_from_id can treat Discussions the same as AR objects and
batch load them.

https://gitlab.com/gitlab-org/gitlab-ce/issues/62826
https://gitlab.com/gitlab-org/gitlab-ee/issues/9489

- Move `calls_gitaly_check` to public
- Add instrumentation for flagging missing CallsGitaly declarations
- Wrap resolver proc in before-and-after Gitaly counts to get the net
Gitaly call count for the resolver.

This exposes all fields named `id` as GlobalIDs so they can be used
across our entire GraphQL implementation.

When the objects loaded are `ApplicationRecord`s. We'll use our
existing batchloading to find them. Otherwise, we'll fall back to the
default implementation of `GlobalID`: Calling the `.find` method on
the class.

Enabling GraphQL batch requests allows for multiple queries
to be sent in 1 request reducing the amount of requests
we send to the server.

Responses come come back in the same order as the queries were
provided.

58404 - add change log


58404 - add spec


58404 - add more spec to test depth 2


58404 - fix spec


58404 - fix rubocop


58404 - refactor the code by Bob's advice


58404 - revert changes of all_graphql_fields_for


58404 - change text only


58404 - fix rspec according to gitlab's standard


58404 - revert previous spec


58404 - fix rubocop

With this we only check abilities on the rendered edges of a GraphQL
connection instead of all the nodes in it.

It makes all Types::BaseField default to a complexity of 1.

Queries themselves now have limited complexity, scaled
to the type of user: no user, authenticated user, or an
admin user.

This is mainly the setup of mutations for GraphQL. Including
authorization and basic return type-structure.

This adds Keyset pagination to GraphQL lists. PoC for that is
pipelines on merge requests and projects.

When paginating a list, the base-64 encoded id of the ordering
field (in most cases the primary key) can be passed in the `before` or
`after` GraphQL argument.

- All definitions have been replaced by classes:
  http://graphql-ruby.org/schema/class_based_api.html
- Authorization & Presentation have been refactored to work in the
  class based system
- Loaders have been replaced by resolvers
- Times are now coersed as ISO 8601

By specifying a presenter for the object type, we can keep the logic
out of `GitlabSchema`.

The presenter gets initialized using the object being presented, and
the context (including the `current_user`).