Repository navigation
An issue tracker as a network hub: give work to a team from GitHub Issues, Jira or Linear (asking before any PR) #3117
Alejandro-SJ
started this conversation in
Ideas
Replies: 2 comments
|
Thanks, @Alejandro-SJ, for asking these scope questions before writing code, and for the clear picture of how your team works today. They're with the team now, and the scope answer will come here. |
0 replies
|
Thanks again, @Alejandro-SJ. Here are answers to your three questions.
None of this is a merge or release commitment. When your bridge or a doc change is ready, it goes through the usual review. |
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Hi Mike. Following the route in CONTRIBUTING ("integrations: ask first, often your own repository"), here is what I'm trying to do, what stops me, and three questions before I write code.
What I'm trying to do. We're a two-person team running one OpenRig instance with several teams at once. Our work already lives in an issue tracker. Today a person reads a ticket, retells it in Slack to the kernel operator, and then copies the PR link and outcome back to the ticket by hand. Any question the team asks goes to Slack, so the discussion ends up split between two places. I'd like the ticket to be the request, the record and the conversation: put it in "Ready", a team picks it up, progress and the result land on the ticket, and a question to a person is a comment there.
What stops me. Nothing in OpenRig reads a tracker, which is right: I don't think the core should embed Jira or GitHub SDKs. #1280 asks for a mission or slice to carry a ticket key, but that doesn't move the work itself.
What I found. The
rig networkconnector in 0.7.13 (#2993) already has the full lifecycle I need. It uses five MCP calls inpackages/daemon/src/domain/network/hub-client.ts:rig_queue_list {owned: true},rig_queue_claim,rig_queue_show,rig_queue_update {note}andrig_queue_update {state, closureReason, resolution}. It:network-hub-item:<id>So a small bridge that serves those five tools over a tracker is a hub, with no change to OpenRig:
rig_queue_list {owned: true}rig:ready(or in a Project's "Ready" column) for that participantrig_queue_claimrig:queuedand removerig:readyrig_queue_showgh:<owner>/<repo>#<n>rig_queue_update {note}rig_queue_update {state: done, resolution}The same five calls map onto Jira (JQL filter, transitions, comments), Linear and Trello. The plan is one bridge in my own repository with one adapter per tracker, GitHub first. OpenRig stays tracker-agnostic.
The one gap: questions. The connector sends nothing between pickup and the result ("No other note on the item is sent"). So when a seat needs a person, the question goes to Slack or the relay, not to the ticket. The relay can't find the ticket either: a non-digest outbound envelope has no
qitemIdor origin (ALLOWED_ENVELOPE_KEYSingateway/relay-outbound.ts). There are two small options I can see:member-workflow/v2. A person's reply on the hub returns the answer to the local item.originwith the root item'snetwork-hub-item:tag, so a relay bridge can post the question on the ticket and send the reply back through/api/gateway/relay/inbound.I lean to A: the ticket stays the one thread, and it reuses shapes OpenRig already has.
Questions before any code:
rig network, living in my own repository and linked from the docs if it's useful?rig networkis experimental, how far can a third-party hub rely on those five tools and the fieldshub-client.tsreads (qitemId,state,summary,body,resolution,closureReason)? Would you take a short hub-conformance page, likerunner-conformance, so other hubs can check themselves?Related: #1280 (ticket key on scope nodes), #2793 and #1802 (waking on outside events), #2329 (relay for people outside Slack).
All reactions