| epriestley | |
| Mar 7 2016, 11:12 PM |
| F1164892: Screen Shot 2016-03-09 at 5.29.33 AM.png | |
| Mar 9 2016, 1:31 PM |
| F1164851: Screen Shot 2016-03-09 at 4.48.28 AM.png | |
| Mar 9 2016, 12:51 PM |
| F1164848: Screen Shot 2016-03-09 at 4.46.30 AM.png | |
| Mar 9 2016, 12:51 PM |
If we build a tremendously complicated, expensive, difficult-to-maintain bridge between Phabricator and GitHub, it is possible that we will reduce the activation energy required to move a project from GitHub to Phabricator enough to have a substantial impact on growth. Or maybe not.
Our current plan is to build a minimal version of this and see if it gets any traction.
| Status | Assigned | Task | ||
|---|---|---|---|---|
| Open | epriestley | T8783 Unprototype Nuance (v1) | ||
| Open | None | T12739 Nuance: As an External Bridge, to GitHub and other similar systems | ||
| Open | epriestley | T10538 Bridge GitHub to Phabricator |
Alright, so here's the first real issue. Here's what I did on GitHub:
I also edited a comment and subscribed/unsubscribed in this sequence.
Here's what the GitHub Events API gives me for that:
That is, I only get the major events (comment, close, reopen).
There's also a separate "Issue Events API", which I haven't looked at yet. So it's possible we can just poll both APIs, which is messy but not especially problematic. I'll see how bad Pull Request-related events look first.
The "Issue Events API" appears to provide the remaining events in a different format with different rules. It does not provide any duplicate events, so, e.g., one API retrieves issues being created and commented on while a different API retrieves labels, milestones, etc., except that both report opens and closes, I think. Anyway, seems mostly workable.
Where does this feature stand? I wasn't able to find any documentation on how to use this, but it also looks that a non-trivial amount of work has been completed towards this goal. Is this something still in the immediate roadmap? Being able to easily integrate GitHub issues into Phabricator (and back) is a big roadblock to adoption for us.
Hard to say since it fell off the roadmap a year ago, but we do have rough plans to ship Nuance (pre-requisite) in the shorter term.