title: ForgeFed Behavior
Abstract
This document provides instructions for using ActivityPub activities and
properties to represent forge events, and describes the side-effects these
activities should have.
Introduction
The ForgeFed behavior specification is a set of instructions for
representing version control system and project management related transactions
using ActivityPub activity objects, and it describes the side effects and
expected results of sending and receiving these activities. The vocabulary for
these activities includes standard ActivityPub terms, new terms defined by
ForgeFed, and terms borrowed from other external vocabularies.
The ForgeFed vocabulary specification defines a dedicated vocabulary of
forge-related terms, and the behavior specification uses these terms, along
with terms that already exist in ActivityPub or elsewhere and can be reused for
forge federation.
The ForgeFed modeling specification defines rules for representing forge
related objects as ActivityPub JSON-LD objects, and these objects are used in
the behavior specification, included in activities or mentioned in
activities or modified due to activity side-effects.
Conformance
The key words MAY, MUST, MUST NOT, SHOULD, and SHOULD NOT are to be interpreted
as described in [RFC2119].
Objects
Objects are the core concept around which both ActivityPub and ForgeFed are
built. Examples of Objects are [Note], [Ticket][type-ticket], [Image],
[Create], [Push][act-push]. Some objects are resources, which are objects that
contain or represent information and user made or program made content, and
some objects are helpers that exist as implementation detail aren't necessarily
exposed to humans or are useful to humans. But everything is an [Object],
represented as compacted JSON-LD.
ForgeFed is an ActivityPub extension, and communication between ForgeFed
implementations occurs using activity objects sent to actor inboxes and
outboxes.
There are 4 kinds of objects in ForgeFed:
- Activities: These are objects that describe actions (actions that happened,
or actions that are happening, or a request to perform an action), and their
primary use is for S2S interaction between actors, by being sent to an
actor's inbox, and C2S interaction between a person or a program and actor
they control, by being sent to the actor's outbox. Activities can also
appear or be linked inside other objects and activities and be listed in
Collections.
- Actors: These are static persistent objects that have an [inbox] and can be
directly interacted with by POSTing activities to it. Their primary use is
to contain or represent information and output of user actions or program
actions, and to manage access this information and to modification of it.
- Child objects: These are persistent objects which, like actors, contain or
represent information and output of user actions or program actions, but
they don't have their own [inbox] and aren't directly interacted with. A
managed static object always has a parent object, which is an actor, and
that actor's inbox is the way to interact with the child object. The parent
actor manages access and modification of the child object.
- Global helper objects: These are objects that don't belong to any actor and
don't need any interaction through activities. As such, they don't exactly
fit into the actor model, but may be involved in implementation details and
practical considerations.
Actors, children and globals are referred in ForgeFed as the static objects,
while activities are the dynamic objects (the terms constant and variable
are used for stating whether an object changes during its lifetime or not).
Static objects, in addition to being an actor or child or global, also have a
resource/helper distinction:
- Resource: Contains or represents information and user made or program made
content, usually belongs to the domain model of version control systems and
project management.
- Helper: Used for running things behind the scenes, not exposed directly as
user content, may be transient or auto generated, usually related to
implementation detail and not to concepts of version control and project
management.
This specification doesn't mandate which types and objects should be actors,
but it does provide guidelines that implementations SHOULD follow:
- Resource objects that have self-contained stand-alone meaning should be
actors
- Objects that handle access control for updates of themselves should be actors
- Objects that need to be able to send activities should be actors
- Objects whose meaning is inherently tied to a parent object, or whose access
control is managed by a parent object, can have all their interactions done
via the parent object, and not be actors themselves
- If an object doesn't need to send or receive activities, even if it's self
contained, there's probably no need to make it an actor, because it
practically doesn't participate in actor-model communication
Here are some examples and their rationale:
- A ticket/issue/bug is created with respect to some project, repo, software,
system, the ticket is inherently a part of that parent object, so tickets
would generally not be actors
- A project or repository are generally self-contained entities, and even if
some forge has users as top-level namespace and repos are created under
users, the user managing/owning/sharing a repo is just a matter of access
control and authority, it isn't a part of the meaning of the repo itself,
and the repo could easily change hands and change maintainers while remaining
the same repo, same software, same content, same meaning. So, repos and
projects would generally be actors.
- A group/organization/team is a self-contained object, a set of users along
with access control and roles and so on, and it needs to be able to receive
update activities that update the team members list, structure and access and
so on, even though a team isn't a user and probably doesn't publish
activities. So, teams would generally be actors.
The proposal here is that the following types typically be actors:
- Person
- Project
- Repository
- Group/Organization/Team
And other types such as these typically not be actors:
- Commit
- Ticket
- Merge request
- Patch
- Diff
- Discussion thread
Actors
A ForgeFed implementation MUST provide an Actor of type Repository
for every
repository that should support federation.
A ForgeFed implementation SHOULD provide an Actor of type Person
for every user
of the platform.
Client to Server Interactions
ForgeFed uses Activities for client to server interactions, as described by
ActivityPub. A client will send objects (eg. a Ticket) wrapped in a Activity
(eg. Create) to an actor's outbox, and in turn the server will take care of
delivery.
Follow Activity
The Follow activity is used to subscribe to the activities of a Repository.
The client MUST send a Follow activity to the Person's outbox. The server
in turn delivers the message to the destination inbox.
Push Activity
The Push activity is used to notify followers when somebody has pushed changes
to a Repository.
The client MUST send a Push activity to the Repository's outbox. The server
in turn delivers the message to the Repository followers.
Server to Server Interactions
Follow Activity
The server receiving a Follow activity in a Repository's inbox SHOULD add the
sender actor to the Repository's followers collection.
Acknowledgements