Skip to content

UI phase 3: processing queue, live view, match results and field map - #41

Open
sushilm20 wants to merge 26 commits into
mainfrom
claude/ui-phase3-processing
Open

sushilm20 wants to merge 26 commits into
mainfrom
claude/ui-phase3-processing

Conversation

@sushilm20

Copy link
Copy Markdown
Contributor

UI phase 3 from the 23-page UI guide: Processing queue, Processing live, Match results and Field map (guide pages 13–16). Built on main after #34, #37 and #38.

What changes

  • Processing queue (Ctrl+9). A new processing_jobs table keeps the queue across restarts. You can add matches, reorder, remove, clear finished jobs and retry failed ones, and start or pause the queue.
    • A job that can't run fails with its reason. For "Calibration missing" and "Robot identities missing", the reason comes with a one-click fix. Failed jobs are never retried automatically.
    • Jobs that were running when the app closed go back into the queue, and the app asks before resuming them.
    • Each job runs through the same MainWindow.process() path as a single match, with that job's sampling rate, tracking mode and model. Your own settings are restored afterwards. Re-seeds from Store reviewer re-seed clicks and apply them on the next run #38 therefore apply to queued jobs as well.
  • Processing live. Progress, sampled frames, frames per second, time left and video time come from the run's own progress reports and log. The page also shows:
    • Phase (AUTO, TRANSITION, TELEOP, POST-MATCH), computed from video time and match_periods_s. The page says it assumes the video starts at match start.
    • Robot detections on the current frame.
    • ID switches, counted from "Recovered robot" and "Re-seeded robot" log lines.
    • The picture is the frame being analysed, without boxes: the pipeline doesn't hand boxes to the app until it finishes, and the page says so.
    • Stopping asks first, and nothing from a stopped run is saved (unchanged).
  • Match results. Our red and blue totals use match_scoreboard (confirmed events only, the same rules as Score and export). Each total sits next to the official score you enter, never merged, with a ±10% check.
    • A review badge says the score isn't final while events are flagged.
    • Robot cards show attributed points, cycles and how many events need review. Absent robots show 0 points.
    • Hand-count accuracy shows "Not yet checked", because no hand count is stored.
  • Field map and trails. Trails come from the stored track positions (feet, origin at the first clicked corner) and break at gaps.
    • Layers: trails, heatmap, zones and the field picture. Ball positions are disabled ("not stored by this build").
    • Each robot's legend entry shows distance and gap count.
    • The scrubber shows the video frame and the trail up to that time. Clicking a trail point jumps there.
  • Sidebar: the "Match workflow" section gains the four pages. Ctrl+9 is now Processing queue, and Review queue keeps its place without a shortcut.
  • No changes to the pipeline, scoring or training. docs/USER_GUIDE.md has a new "Processing and results screens" section.

Screenshots

These use demo jobs, positions, events and log lines on the real footage, so the layout can be seen without a full run. The numbers in them are not results.

Processing queue
Processing live
Match results
Field map

Tests

tests/test_processing_pages.py covers:

  • job storage, ordering, recovery and retry
  • run-log parsing and phase tracking
  • the queue page, the live page, results scoring and the official check
  • field map trails, gaps and nearest point
  • the window's queue runner

The suites in test_modern_ui and test_setup_pages still pass.

Next

The next UI PR moves the window to FluentWindow and these pages onto the qfluentwidgets components from #40.

Copilot AI balanced review requested due to automatic review settings November 17, 2026 17:07
@sushilm20

Copy link
Copy Markdown
Contributor Author

@loopyreviews-bot review

Copilot AI left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Copilot was unable to review this pull request because the user who requested the review has insufficient remaining quota. Try requesting a review at a lower effort level.

An organization owner can choose the repository organization as the billing source for reviews requested by members with a paid Copilot license. Learn about Copilot code review billing.

@sushilm20

Copy link
Copy Markdown
Contributor Author

@loopyreviews-bot review

Keep both guide additions: the Later corrections paragraph from main and the
processing and results screens section.
@sushilm20

Copy link
Copy Markdown
Contributor Author

@loopyreviews-bot review

Comment thread control_vision/app/runlog.py Resolved
Comment thread control_vision/app/main_window.py Resolved
@loopyreviews-bot

loopyreviews-bot Bot commented Nov 17, 2026 •

Copy link
Copy Markdown

Loopy review

Confidence: 5/5 — No open findings.

Reviewed 4 of 10 changed files.

UI phase 3 adds four pages (Processing queue, Processing live, Match results, Field map) and wires them into MainWindow. A persistent processing_jobs queue runs each job through the existing process() path, with the job's sampling rate, tracking mode and model. The run log is relayed to the live page, and the sidebar and shortcuts are updated. I reviewed main_window.py and fieldmap_page.py and found no new defects beyond those already dismissed.

  • MainWindow: queue runner (start, pause, next job, finish), settings save and restore around each job, run-log handler attach and detach, and hooks in processing_done and processing_failed.
  • FieldMapPage: trails built from stored track positions with gap-aware segments, heatmap, zoom and pan, scrubber, and click-to-seek.

Review passes: 4 ran (general ×2, failure-paths, state) · 16 unit and lens pair(s) carried over from earlier reviews · 6 planned unit and lens pair(s) left out by the budget.

Ran on: Claude on a Mac. Muse on a Mac: no Mac took the review in time.

Reviews: 6 · last reviewed d051943 · harness claude/claude-sonnet-5-5 · Reply @loopyreviews-bot review to re-run · @loopyreviews-bot remember <rule> to teach Loopy

The pipeline writes "Processed 1,234 of 5,000 sampled frames"; the live page lost
its frame count and time left once a run passed 999 sampled frames.
Cancel or Stop processing on the running job no longer starts the next match;
Start queue carries on.
@sushilm20

Copy link
Copy Markdown
Contributor Author

@loopyreviews-bot review

Comment thread control_vision/app/main_window.py Resolved
Comment thread control_vision/app/main_window.py Resolved
…not start

Start queue while another run is active pauses the queue and says so. An error
while a job starts fails that job with the reason, restores the user's settings
and pauses the queue instead of leaving the job running forever.
@sushilm20

Copy link
Copy Markdown
Contributor Author

@loopyreviews-bot review

Comment thread control_vision/app/main_window.py Resolved
Comment thread control_vision/app/pages/queue_page.py Resolved
… run

The failure goes through processing_failed, so the match is not left running,
Process is enabled again and the run log is detached and closed.
@sushilm20

Copy link
Copy Markdown
Contributor Author

@loopyreviews-bot review

Comment thread control_vision/app/pages/queue_page.py Resolved
A match that already waits (retried from the live page) is skipped, and a match
that failed more than once gets one job.
@sushilm20

Copy link
Copy Markdown
Contributor Author

@loopyreviews-bot review

@sushilm20

Copy link
Copy Markdown
Contributor Author

@loopyreviews-bot review

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants