fix: -q flag suppresses sparkline data in HTML reports

[decko]

Bug

Running raki run -q (quiet mode) generates an HTML report without trend indicator badges, even when history data exists. Running without -q on the same data produces badges correctly.

Root cause

Needs investigation — the sparkline build block runs regardless of -q, but the HTML report ends up without badges when quiet is set.

Reproduction

raki run -m manifest.yaml -o out/ --force -q   # no badges in HTML
raki run -m manifest.yaml -o out/ --force       # badges appear

[decko]

Spec

Problem

Running raki run -q produces an HTML report without trend delta badges, even when history data exists and the same command without -q produces badges correctly.

Root cause (investigation needed)

The sparkline build block in cli.py is not gated by quiet — run_sparklines is built unconditionally when _resolved_history_path is not None, and write_html_report() always receives run_sparklines. The -q flag only gates print_summary() and status messages.

Likely candidates to investigate:

1. Exception silently swallowed: The except Exception: run_sparklines = None block at line 642 may be catching a real error (e.g. import failure, missing history file) that only manifests in the -q path due to a subtle import ordering or side-effect difference.

2. load_history() returns empty on first --force run: If the reproduction requires --force on the very first run (no prior history), then history_entries is empty, run_sparklines stays None, and no badges appear — identically with or without -q. The reporter may have run without -q on a subsequent run (history now has an entry), making it appear -q is the cause.

3. Console-level suppression: Console(quiet=True) or similar Rich behavior suppressing template rendering output — unlikely but worth ruling out.

4. -q interacts with another flag: The reproduction uses --force -q together; --force bypasses duplicate detection. Verify whether the bug reproduces with -q alone (no --force).

Fix (after root cause confirmed)

• If cause is (1): add targeted exception logging and surface the real error.
• If cause is (2): this is not a real bug — document expected behavior.
• If cause is (3) or (4): fix the specific interaction.

The fix must not move sparkline building inside the if not quiet: block — sparklines must always be passed to write_html_report() regardless of quiet mode.

Acceptance Criteria

• Root cause identified and documented in this issue
• raki run -q produces HTML badges identical to raki run (no -q) when history has ≥ 2 entries
• A regression test reproduces the original failure and passes after the fix
• No change to sparkline data passed to write_html_report() under any quiet mode
• Towncrier fragment 365.fix

---

Plan

Step 0 — Reproduce and isolate

Before writing code, confirm the exact reproduction path:

# Run 1: populate history (no -q, no --force)
raki run -m raki.yaml -o out/

# Run 2: reproduce the bug
raki run -m raki.yaml -o out/ --force -q

# Check: does the HTML in out/ contain delta-badge elements?
grep -c "delta-badge" out/*.html

Also reproduce without --force:

raki run -m raki.yaml -o out/ -q
grep -c "delta-badge" out/*.html

And check whether an exception is being swallowed by adding temporary logging to the except Exception block at cli.py:642.

Step 1 — Add exception logging (always)

File: src/raki/cli.py, the except Exception block in the sparkline build section.

Replace the silent run_sparklines = None fallback with a logged warning, even in quiet mode (warnings go to stderr):

except Exception as exc:
    if verbose:
        out.print(f"[yellow]Warning: Failed to build sparkline data: {exc}[/yellow]")
    run_sparklines = None

This ensures silent failures surface under --verbose and does not affect quiet mode output.

Step 2 — Fix the identified root cause

Apply the targeted fix once root cause is confirmed from Step 0. Update this plan with the specific change before implementing.

Step 3 — Add regression test

File: tests/test_cli.py or tests/test_365_quiet_sparklines.py

Write a test that:

1. Pre-populates a JSONL history file with ≥ 2 entries for a known manifest
2. Runs raki run -q via CliRunner
3. Reads the generated HTML and asserts delta-badge elements are present

Step 4 — Towncrier fragment

File: changes/365.fix

The ``-q`` / ``--quiet`` flag no longer suppresses trend delta badges in generated HTML reports.

Estimated scope

• Investigation: 30–60 min
• Code change: likely ≤ 20 lines depending on root cause
• Test: ~30 lines
• 1 changelog fragment