The clock rotates, and only thinking and the clock are terminated #815
No reviewers
Labels
No labels
move-to-repo
coilyco-bridge-deploy
move-to-repo
coilyco-flight-deck-agent-compose
move-to-repo
coilyco-gaming-eco-app
move-to-repo
coilysiren-inbox
move-to-repo
unknown
🔒⚠️📦⚠️🔒 SANDBOXED 🔒⚠️📦⚠️🔒
autonomy
async-consult
autonomy
epic
autonomy
headless
autonomy
live-collab
c#
priority
P0
priority
P1
priority
P2
priority
P3
priority
P4
role/ai
role/creator
role/design
role/director
role/engineer
role/exec
role/human
role/ops
role/qa
No milestone
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set
Reference
coilyco-gaming/sirens-echo!815
Loading…
Reference in a new issue
No description provided.
Delete branch "aos/claude/yt58-clock-rotation"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Two decisions Kai settled on #370. Both were previously held open deliberately, so this is implementation rather than a taste call.
What it renders
and the other three stages, now bare:
The clock rotates
An earlier round correctly treated that as an invitation rather than a spec and shipped one 🕐. Kai has now asked for it, so the twelve faces advance down the column as the lines pile up.
The rotation wraps, which never fires today:
maxProgressWaitLinesis 12 and there are twelve faces, so a capped turn uses each face exactly once. The modulo is there so the cap and the rotation length can move independently rather than silently coupling. A test pins both facts.One correction to the record. Quail's note on the issue, and Delphi's summary of it, both say a turn can emit seventeen elapsed lines at the 180 second p99. That is arithmetic from the p99 over the beat, and it does not hold:
maxProgressWaitLinesbounds the column at twelve and landed inf62ae55, the same commit as the elapsed lines and fourteen minutes before the note. No turn has ever been able to emit seventeen. The number is worth correcting because it was raised as a possible noise problem, and twelve is a third smaller than what was weighed.Only thinking and the clock are terminated
The ellipsis had been applied to all four stages on the reading that it meant still-happening. That reading is now reversed.
The icon and the ellipsis are separate fields, not one implying the other. It is tempting to derive the ellipsis from "has an icon", since thinking and the clock are exactly the iconned lines today, but they were named in two separate decisions. Deriving it means that the day you name an icon for
calling a tool, it silently grows an ellipsis you did not ask for. A test pins that an icon alone does not terminate a line.What is deliberately unchanged
The three stage icons are still unnamed.
reading recent messages,calling a tool,checking the replyrender plain. Delphi noted 🔨 already has an approved meaning for tool calls in the reaction vocabulary, so that one has an obvious answer whenever you want it, but you have not named it and I am not picking it.The brackets stay dropped.
still thinking 9 seconds...rather thanstill thinking... (9 seconds). Restoring them means wideningnoticeAllowed, which is a decision about the harness/model boundary rather than about formatting, and it is recorded as settled on the issue.Verified
Three new tests in
progresswait_test.goand two instagenotice_test.go: the hour hand advances across successive lines, the rotation wraps and holds twelve faces, only the clock line inside a non-thinking stage carries the ellipsis, the three bare stages render exactly as shown above, and an icon without a trailing flag does not terminate.Every existing shape test still passes, including the one asserting no rendered line escapes
noticeShape.ward exec gategreen.docs/sirens-echo-progress.mdhad a stale "what a member sees" block predating both the icon and the elapsed lines; it now shows what actually renders.History of this branch
It was first cut when
mainwas red, and carried a fix for that as its first commit (#813). #812 landed the same two fixes first, so this branch has been rebased onto currentmainand that commit dropped. The diff is now the progress change alone.c1ea1087ecb0e438d4e9