cmoriarty/braid
1
0
Fork 0
Code Issues 15 Pull requests Projects Releases Packages 1 Wiki Activity Actions

Review/Revise graph spine re-design #114

New issue
Closed
opened 2026-09-29 10:59:31 -04:00 by cmoriarty · 1 comment
cmoriarty commented 2026-09-29 10:59:31 -04:00
Owner
Copy link

The Review/Revise section of the graph spine is still a big confusing. I think I'd like to remove the agent.review and openspec.revise steps, since they are really non-functional steps, more like section headers. Instead how I would like to see it work is for there only to be a single set of agent.review, script.review-digest, openspec.revise, git. commit.revision. If a second round is going to start, then the graph should expand and add their steps, in a visually interesting way (subtle fade in, rather than jagged change).

image

The Review/Revise section of the graph spine is still a big confusing. I think I'd like to remove the agent.review and openspec.revise steps, since they are really non-functional steps, more like section headers. Instead how I would like to see it work is for there only to be a single set of agent.review, script.review-digest, openspec.revise, git. commit.revision. If a second round is going to start, then the graph should expand and add their steps, in a visually interesting way (subtle fade in, rather than jagged change). ![image](/attachments/7c1405da-afd4-4485-bfee-efe919d366a2)
image.png
221 KiB
cmoriarty commented 2026-10-03 11:07:21 -04:00
Author
Owner
Copy link

Shipped and live on production (9ae119c, deployed 2026-10-03).

  • The two summary rows (agent.review ×3, openspec.revise ×2) and their arrows are gone. The review is its first round as ordinary rows: one agent.review row for each lens, script.review-digest, openspec.revise and git.commit.revision. A revise or commit that was not needed stays a row marked not needed.
  • A later round is drawn when it starts: all its steps are added together, after the previous round's, once any of them is ready, running, waiting, failed or finished, or has begun an attempt. A round whose steps are all pending, skipped as not needed, or cancelled without starting is not drawn, so a review that needed one revision shows two sets and one that needed none shows one. A run recorded with the earlier step ids is drawn the same way.
  • The rows that join an open spine ease in (opacity and height, about 420 ms) while the rows below move down with them. Nothing fades when a run is opened, and nothing fades under reduced motion.
  • The step numbers have no gap, and a phase band's N of M and done/total count the rows drawn. The run view, the j/k walk and the run bar's stepper all use the same drawn list.

Checked with Playwright against the fixture (a clean review, a two-round review, a run stopped in round 2, and a run whose second round starts while the page is open: its new rows' opacity rises while the gate row below moves down over several frames) and, on production in the deployed console: a run with a revision and a second round draws round 1 and round 2 as rows and no round 3 (the REVIEW band at 16/16), and run 41, a clean first round, draws round 1 only (9/9), both with no summary row and no gap in the numbering. The fast and full lanes pass.

#126 changes what a later round is (one check in place of four lenses); this draws either shape. Change record: openspec/changes/archive/2026-10-03-review-rounds-grow-in-spine.

Shipped and live on production (`9ae119c`, deployed 2026-10-03). - The two summary rows (`agent.review ×3`, `openspec.revise ×2`) and their arrows are gone. The review is its first round as ordinary rows: one `agent.review` row for each lens, `script.review-digest`, `openspec.revise` and `git.commit.revision`. A revise or commit that was not needed stays a row marked not needed. - A later round is drawn when it starts: all its steps are added together, after the previous round's, once any of them is ready, running, waiting, failed or finished, or has begun an attempt. A round whose steps are all pending, skipped as not needed, or cancelled without starting is not drawn, so a review that needed one revision shows two sets and one that needed none shows one. A run recorded with the earlier step ids is drawn the same way. - The rows that join an open spine ease in (opacity and height, about 420 ms) while the rows below move down with them. Nothing fades when a run is opened, and nothing fades under reduced motion. - The step numbers have no gap, and a phase band's `N of M` and `done/total` count the rows drawn. The run view, the `j`/`k` walk and the run bar's stepper all use the same drawn list. Checked with Playwright against the fixture (a clean review, a two-round review, a run stopped in round 2, and a run whose second round starts while the page is open: its new rows' opacity rises while the gate row below moves down over several frames) and, on production in the deployed console: a run with a revision and a second round draws round 1 and round 2 as rows and no round 3 (the REVIEW band at 16/16), and run 41, a clean first round, draws round 1 only (9/9), both with no summary row and no gap in the numbering. The fast and full lanes pass. #126 changes what a later round is (one check in place of four lenses); this draws either shape. Change record: `openspec/changes/archive/2026-10-03-review-rounds-grow-in-spine`.
Sign in to join this conversation.
No Branch/Tag specified
Branches Tags
main
archive/prefect-kestra-experiments
Labels
Clear labels
No items
No labels
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
cmoriarty/braid#114
No description provided.