Page MenuHomePhorge

Starmap
Open, Needs TriagePublic

Description

Phorge needs to populate a starmap identifying the many different applications and a general focus of what updates/changes are expected for them. Unlike a roadmap the starmap does not identify priorities but only provides vision for the individual applications.

The current starmap document

The last Phabricator starmap prior to deletion is located here:
https://secure.phabricator.com/w/starmap/?v=20

Phorge's initial starmap will likely align closely with Phabricator's, at least initially.

Event Timeline

speck updated the task description. (Show Details)
speck added subscribers: Matthew, Bezalel.

I'd suggest starting of the starmap by shedding some/most of the applications that were not part of the core of Phabricator or have been in the Prototype phase for a long time, allowing the project to regain focus and improve on the things that make Phorge unique. This would provide a number of advantages:

  1. Phorge is a young project with a small team and community inheriting a large code base. This means that new contributors will have to get acquainted with the code.
  2. Documentation on many of the smaller applications is very sparse. It's hard to understand the vision.
  3. Reiterating. The small team, with a smaller code base would have focus, allowing to pour more resources into the remaining applications.
  4. Many of the applications that Phabricator intended to provide are available from vendors that already provided by third parties. While I understand and respect the vision behind Phabricator's one-stop-shop approach, it's just (IMO) not feasible for a project with this scope and a small team. Instead we could focus on making Phorge extensible by third parties.

I'd also like to bring up stronger integrations with other services. Some I'd suggest

  1. Github / Gitlab Pull Requests: These provide overlapping functionality to Phabricator. If we were able to bridge Pull requests with Differential, the projects we host inside of Phabricator would not feel so isolated when using it to host an open source project.
  2. CI / CD: Currently, I feel most CI/CD providers do not work. The documentation on Harbormaster is basically unavailable and many of it's features are unimplemented.
  3. Unit testing / Linting: suffers from the same issues that CI has. Common drivers do not work and Phabricator tries to enforce it's own tools instead.

I understand that most of these suggestions require a "Philosophy change" to the project, at least when compared with Phabricator. I understand that Phabricator was all about providing the "toolbox and the tools", while these suggestions would make it more of a toolbox, and you add the tools you prefer. I have personally struggled a lot with some of the tools that Phabricator brings, and that's why I wouldn't be sad to see some of them go, but I definitely love Phabricator as a toolbox.