Round 3 · notes · nonfinal
Three more directions, built to broaden the range rather than refine what exists. These notes cover what the round set out to do, how 7–9 differ, how the five corrections are applied, issues inherited by concepts 1–6, invented fixture values, state coverage and the questions we need you to answer. No direction is selected.
Fictional demonstration data — not real equipment
mainskeeper status CLI preview with a NO_COLOR rendition.
A light-first working instrument. A datum is the fixed level you measure from; one level line runs through the drawn wordmark. The home is a fleet line, an incident strip and a sortable 13-column grid of all six UPSs, with an overlay inspector and a compare dock.
Answers “which piece of equipment, and how does it compare?” Restrained: greys, graphite and a single copper accent for “you are here”.
A dark-first map of who watches what: collectors → UPSs → equipment in four site clusters, with reads, powers and checks links. Selecting anything lights its path; a coverage toggle shows what nobody watches.
Answers “what depends on what, and where are the blind spots?” Symbol-led: mint for live, sodium for selection.
A power weather service: sites as stations, readings as observations, UPS runtime as a labelled forecast, late records as late reports and incidents as plain-headline bulletins, on a station chart that says it is not a map.
Answers “what are conditions, and what has been reported late?” The visual risk: sage paper, weather-chart colours, a wind-barb mark.
Five corrections to the shared scenario apply this round: recovery is independent of the history upload; open incidents include acknowledged ones, and severity, acknowledgment, recovery and sync are separate attributes; site summaries show mixed states; the first observation after a gap does not establish when power returned; completed episodes show when the start and the end were each received. Pointers are to the incident state of each dashboard (file:line in the concept folder; “fixture” means the line holds the concept’s data object).
| Correction | 7 Datum | 8 Lattice | 9 Synoptic |
|---|---|---|---|
| Recovery stands on its own; the upload backlog is a separate sync attribute | Duplex reads “Recovered 15:02 · uploading backlog, 1,248 pending” in the fleet line; the incident strip has separate Recovery and Sync fields.g/dashboard.html:482–493 | Duplex cluster: “Recovered 15:02” with “Sync (separate) 382 of 1,630 uploaded · 1,248 pending”; detail panel repeats the split.h/dashboard.html:350 (fixture) | Duplex plate: “Recovered 15:02 · history still uploading”; the bulletin lists Recovery and Sync as separate attributes.i/dashboard.html:664–668, 735 |
| Open includes acknowledged; severity, acknowledgment, recovery and sync are separate | “Open incidents 3 · incl. 1 acknowledged”; each incident shows severity, state, ack, recovery, sync and notifications as separate fields.g/dashboard.html:493 · incident strip | Incident rail entries and the detail panel carry each attribute separately; Duplex stays open while acknowledged.h/dashboard.html:350 (fixture: “3 open · 1 critical · 1 acknowledged”) | Bulletin attribute list: Severity, State, Acknowledged (jp 13:52), Recovery, Sync, Notifications.i/dashboard.html:735 |
| Mixed site states (Broadwater) | Fleet line: “1 of 2 UPSs on battery · 1 on utility (separate circuit)”; both UPS rows in the grid.g/dashboard.html:478 | Broadwater cluster summary uses the same wording; the two UPS nodes carry their own states.h/dashboard.html:350 (fixture) | Broadwater station plate with two station plots in different states.i/dashboard.html:541 |
| First observation after a gap does not establish return time | Duplex timeline: “Utility first observed 14:52, return time unknown between 14:30 and 14:52.”g/dashboard.html:544 | Duplex evidence timeline: “Utility first observed present at 14:52 (first record after the gap). Return time unknown — sometime between 14:30 and 14:52.”h/dashboard.html:350 (fixture) | Duplex report (“The Duplex story”) and gaps list state the same, with the 14:52 timestamp marked uncertain ±90 s.i/dashboard.html · station report |
| Completed episodes show start receipt and end receipt separately | Inspector “Recent episodes”: yesterday 21:14–21:31 with start record received 21:14:03 and end record received 21:31:05; Home 06:42 likewise.g/dashboard.html:603, 611 | Duplex and Home evidence: “Start record received 21:14:03; end record received 21:31:05.”h/dashboard.html:350 (fixture) | Late reports table lists event time, received time and lag for start and end of each episode (06:42:49, 21:14:03, 21:31:05).i/dashboard.html:603–608, 712–715, 834 |
Concepts 1–6 are locked and were not rebuilt. Reading their sources against the corrections and the round-two inventory found the following. Paths are under concepts/; 1–3 use the round-one scenario.
sev: "recovering")No winner is selected and no concepts are combined this round.