Status line: band shape, drop token_in/token_out segments

Band merges the segment groups into one row above the editor and paints
the labeled inline context meter (60% - 200K); the session title rides
the working row. Cumulative token totals live in /session - off the bar.
Docs and changelog updated to the band final state.
This commit is contained in:
Frank
2026-10-18 00:58:21 -07:00
parent 04d3628c94
commit 246189f054
3 changed files with 22 additions and 19 deletions
+2 -2
View File
@@ -15,8 +15,8 @@ timestamp: 2026-08-28
Interactive tune-up of the omp status bar (grilling session, live loop: config edit → watcher repaint → niri screenshot verify). New knowledge page: [OMP Status Line Customization](/pi/statusline.md).
* **Mechanics discovered:** `leftSegments`/`rightSegments` are inert unless `preset: custom` (fleet was on `nerd` with dead arrays); `composer.shape` routes them — in `claude`, right segments paint the top rule chip, left segments the bottom strip. The red rule **is** the context meter (length = usage); `contextLine` label variants only render in `box`/`band`. Working row (`⠿ Working… + tok/s`) is hardcoded — model/thinking can never join it (Tern-only). Eight composer shapes registered in the bundle; box/band previews run live and rejected.
* **Final bar:** chip = `token_in · token_out · cost · context_pct · time_spent` (stable width, pure numbers); strip = `session_name · vim · mode · hostname · model · path · git(no branch text) · pr`. Cut: tok/s duplicate, clock, cache trio, `context_total`, `subagents`, `session` hash. `path`: logical `project/worktree` label (`stripWorkPrefix: true`, kept after a live `false` trial of the full cwd); branch text dropped (worktree branches are autogenerated), dirty badges kept.
* **Mechanics discovered:** `leftSegments`/`rightSegments` are inert unless `preset: custom` (fleet was on `nerd` with dead arrays); `composer.shape` routes them — `band` merges both arrays into one row above the editor (title rides the working row), `claude` splits them into top chip + bottom strip. The `claude` red rule **is** the context meter (length = usage); `contextLine` label variants render in `box`/`band` — `band` paints the labeled inline meter (`60% ——— × 200K`). Working row (`⠿ Working… + tok/s`) is hardcoded — model/thinking can never join it (Tern-only). Eight composer shapes registered in the bundle.
* **Final bar (`band`, after a `claude` sojourn):** left = `session_name · vim · mode · hostname · model · path · git(no branch text) · pr`; right = `cost · context_pct · time_spent` with the labeled inline context meter between the groups. Cut: tok/s duplicate, clock, cache trio, `context_total`, `subagents`, `session` hash, and `token_in`/`token_out` (cumulative totals live in `/session`). `path`: logical `project/worktree` label (`stripWorkPrefix: true`, kept after a live `false` trial of the full cwd); branch text dropped (worktree branches are autogenerated), dirty badges kept.
* **models.yml:** frank-archon display `name` slimmed to `"frank-archon"` (endpoint suffix dropped; `id` untouched).
* **Out of scope noted:** worktree name generation (`wt-<date>-<hash>`) is omp-internal; possible upstream feature request.
+19 -14
View File
@@ -33,9 +33,9 @@ with any named preset they are silently ignored. The fleet config runs
| shape | leftSegments | rightSegments |
|---|---|---|
| `claude` (fleet) | bottom strip, left-aligned | top rule chip, right-aligned |
| `band` (fleet) | merged single row above editor | same row, right side (title skipped — it rides the working row) |
| `claude` | bottom strip, left-aligned | top rule chip, right-aligned |
| `rule` | same as claude, gap above strip | same |
| `band` | merged single row above editor | same row, right side (title skipped — it rides the working row) |
| `box` | box top border | box top border |
| `borderless` / `pi` / `field` / `rail` | single bottom bar | single bottom bar |
@@ -58,12 +58,14 @@ staged/unstaged/untracked badges (branch text optional via
## The context meter
The red rule in the `claude` shape **is** the context meter: its length tracks
context usage. `statusLine.contextLine` (`off` / `percentage` / `annotated` /
`embedded`) only changes rendering in `box`/`band` — in `claude` all non-`off`
values look identical. Docs' claim that contextLine is "box composer only" is
wrong for the meter itself; the labeled inline variant (`57%×—200K←`) is the
box/band-only part.
In the `band` shape (fleet) the meter is the labeled inline line between the
segment groups (`60% ———— × 200K`). In the `claude` shape the red rule left of
the right-segment chip **is** the meter: its length tracks context usage.
`statusLine.contextLine` (`off` / `percentage` / `annotated` / `embedded`)
only changes rendering in `box`/`band` — in `claude` all non-`off` values
look identical. Docs' claim that contextLine is "box composer only" is
wrong for the meter itself; the labeled inline variant is the box/band-only
part.
## Path segment and `stripWorkPrefix`
@@ -85,13 +87,16 @@ modes. The fleet runs `true` — logical label, worktree glyph (verified live
## Fleet configuration (2026-10-18)
- Chip (right): `session_name` was moved to the strip; chip = `token_in`,
`token_out`, `cost`, `context_pct`, `time_spent` — pure numbers, stable width.
- Strip (left): `session_name`, `vim`, `mode`, `hostname`, `model`, `path`,
`git` (no branch text — worktree branches are autogenerated names), `pr`.
- Shape `band`: one merged status row. Left group: `session_name`, `vim`,
`mode`, `hostname`, `model`, `path`, `git` (no branch text — worktree
branches are autogenerated names), `pr`. Right group: `cost`,
`context_pct`, `time_spent`, with the labeled inline context meter
between the groups. `session_name` renders on the working row, not the
status row.
- Deliberately cut: `token_rate` (lives on the working row), `time` (panel
clock), `cache_*` (noise), `context_total` (implied by `x/200K`),
`subagents`, `session`.
clock), `cache_*` (noise), `context_total` (implied by the meter),
`subagents`, `session`, `token_in`/`token_out` (cumulative totals are
in `/session` anyway).
- Model display names slimmed in `models.yml` (`frank-archon`, no endpoint
suffix); `name:` is free-form display, `id:` is the wiring — renaming ids is
the cutover procedure, see [Pi Models](/pi/models.md).