Skip to content

v0.19.2 - Summary-Based Threshold Fitting (2026-03-18)

What Changed?

This release adds --use-summary option to fit_thresholds.py, enabling it to read pre-aggregated hit counts directly from thresholds_scanned.jsonl instead of loading individual event files. Output filename is also renamed from fitted_thresholds.json to thresholds_fitted.json for naming consistency.

What's New

Main Feature: --use-summary option in fit_thresholds.py

What it does: When --use-summary is specified, fit_thresholds.py loads thresholds_scanned.jsonl (generated by get_thresholds.py) directly. This is faster than reading all individual event files and requires no raw event data on disk.

How to use it:

# Traditional: read all individual event files
uv run fit_thresholds.py ./measurements

# New: read pre-aggregated summary file (faster)
uv run fit_thresholds.py ./measurements --use-summary

Installation

Quick Start

# Get the release
git checkout v0.19.2

# Setup
task env:setup

# Run CLI
uv run kazunoko --help

What's Different from the Last Version?

✅ Added

  • fit_thresholds.py: --use-summary option to load thresholds_scanned.jsonl directly
  • fit_thresholds.py: load_summary_file() and process_summary_data() functions

🔧 Changed

  • fit_thresholds.py: output filename renamed from fitted_thresholds.json to thresholds_fitted.json (aligns with thresholds_scanned.jsonl naming convention)
  • plot_thresholds.py, README.md, README.ja.md: updated references to thresholds_fitted.json

Is It Safe to Upgrade?

Backward Compatible: No (breaking change in output filename)

  • Output file renamed: fitted_thresholds.json → thresholds_fitted.json
  • Scripts that read fitted_thresholds.json (e.g. plot_thresholds.py) must use the new name
  • plot_thresholds.py is already updated in this release

Tests Passed

  • ✅ Builds without errors
  • ✅ commitizen consistency check passed
  • ✅ --use-summary verified with mock data

Release Details

  • Date: 2026-03-18
  • Version: v0.19.2
  • Files Changed: 4
  • Commits: 99bdf55, c3f8ea5, 2f9c4f6, a1846f7

Next Steps

  • Continue improving threshold analysis workflow