Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

feat: add feature to get inherited member for project/group #1187

Open
wants to merge 5 commits into
base: master
from

Conversation

@Shkurupii
Copy link

@Shkurupii Shkurupii commented Sep 12, 2020

Hi!
This PR about feature #1133. Here I propose to introduce new custom managers ProjectMemberAllManager (/projects/:id/members/all/:user_id) and GroupMemberAllManager (/groups/:id/members/all/:user_id). Subsequently, in order to fetch a list or instance of members including inherited ones, you will need to:

project = gl.projects.get(project_id)
project.members_all.list()
project.members_all.get(member_id)
group = gl.groups.get(group_id)
group.members_all.list()
group.members_all.get(member_id)

I had thought about group.members.all(id=group_id) also. But then changed my mind.
If the approach is right?

Copy link
Member

@nejch nejch left a comment

Nice, I really like this. It helps remove the confusion around having all() methods from /all endpoints, all=True for pagination, and "all": True in query params where it's an attribute GitLab itself accepts. With the CLI this has caused a lot of confusion.

This solution is also very similar to @chrisoldwood's proposal in #593.

I just have some comments on keeping backwards compatibility and naming, if you can take a look, because this can probably later be reused in some other places (maybe for runners as well) :)

@@ -312,7 +312,7 @@

group1.members.delete(user1.id)
assert len(group1.members.list()) == 2
assert len(group1.members.all())
assert len(group1.members_all.list())

This comment has been minimized.

@nejch

nejch Sep 13, 2020
Member

Suggested change
assert len(group1.members_all.list())
assert len(group1.members.all()) # Deprecated
assert len(group1.members_all.list())

As you can see in the test you've had to change in 50f4b9c, this would break the existing behavior for anyone who has relied on .all() so far. I think it might be better to preserve the old behavior with a DeprecationWarning and remove it later with a major release.

Maybe that can be done (temporarily) with a ListAllMixin that contains the method, to avoid copy/pasting. Just need to make sure to register_custom_action for both Project and Group members.

This comment has been minimized.

@Shkurupii

Shkurupii Sep 14, 2020
Author

I rolled back the test case group1.members.all() and added MemberAllMixin with register_custom_action and DeprecationWarning , ListAllMixin sounds very general, we won't use ListAllMixin somewhere I guess.

@@ -4595,6 +4558,7 @@ class Project(SaveMixin, ObjectDeleteMixin, RESTObject):
("issues", "ProjectIssueManager"),
("labels", "ProjectLabelManager"),
("members", "ProjectMemberManager"),
("members_all", "ProjectMemberAllManager"),

This comment has been minimized.

@nejch

nejch Sep 13, 2020
Member

For me the dilemma here is: members_all/ProjectMemberAllManager or all_members/ProjectAllMemberManager 😁 WDYT @max-wittig? all_members (or allmembers) sounds more natural to me. On the other hand members_all does follow the URL segments.

Just asking because this pattern can be a precedent for improving other /all endpoints (e.g. #593) so it might have further reach.

This comment has been minimized.

@Shkurupii

Shkurupii Sep 14, 2020
Author

I spent a few minutes scanning the GitLab API and looks GitLab doesn't tend to make its addresses natural. Anyway, I am not a python Pro, final conclusion is up to you.

gitlab/v4/objects/__init__.py Show resolved Hide resolved
@Shkurupii
Copy link
Author

@Shkurupii Shkurupii commented Sep 14, 2020

Agree, I thought about backward compatibility right after made a PR. But decided to wait for a review, because in any way, there were more questions than answers.

In the git log, I found RELEASE_NOTES.rst, where such things like mine need to be reflected. Who will manage that, I?

@max-wittig
Copy link
Member

@max-wittig max-wittig commented Sep 14, 2020

But why is this needed? We already have a few places, where we use all()

@nejch
Copy link
Member

@nejch nejch commented Sep 14, 2020

But why is this needed? We already have a few places, where we use all()

For me the all() method is problematic, because it means you can't implement the GET /resource/all/<resource-id> endpoints cleanly (at least from what I've seen - e.g. this endpoint https://docs.gitlab.com/ee/api/members.html#get-a-member-of-a-group-or-project-including-inherited-members), as this can currently only return lists. Maybe by adding the ID param to it but members.all(id=123) also seems inconsistent with the list()/get() from other managers?

It's weird that GitLab added these endpoints separately, but I guess they might add them to other resources. The other place where I've found it used is with runners and there is also a similar issue open for that. Visually it's also a bit confusing, you need to do gitlab project-member all --all/project.members.all(all=True) to combine pagination and this list 😁 and then gitlab somtimes also accepts an all param 🤦

@Shkurupii
Copy link
Author

@Shkurupii Shkurupii commented Sep 16, 2020

Hi! I think we're going to have to agree to disagree.
@max-wittig, how you would see the implementation of /projects/:id/members/all/:user_id?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Linked issues

Successfully merging this pull request may close these issues.

None yet

3 participants
You can’t perform that action at this time.