Skip to content

Chronaxis 0.1.0.dev2

Pre-release
Pre-release

Choose a tag to compare

@liuyaoze98 liuyaoze98 released this 05 Nov 15:16
· 10 commits to main since this release

Chronaxis 0.1.0.dev2

English

本版扩展配置校验、表达式与统计计算,增加高性能积分器,并同步中英文手册和随包 AI 指南。它仍属于开发预览版,运行环境为 Windows、Python 3.12。

安装或升级:

python -m pip install --upgrade "chronaxis==0.1.0.dev2"
python -m chronaxis guide

相比 0.1.0.dev1 的变化

功能 本版行为
配置诊断 一次返回多个可独立定位的配置错误,给出字段位置;依赖错误字段的后续检查仍可能需要先修正前置错误。
结果发布 可以重复使用输出目录;同次输出采用统一时间后缀,不覆盖已有文件,完成清单最后发布。
单位与表达式 分析配置使用版本 4,支持兼容单位自动换算、带单位常量、条件分支、数值与位运算、一维标定查表。
统计与时序变换 增加计数、方差、标准差、均方根、分位数和中位数,以及结果聚合、滚动统计和变化率。
自定义分析算子 宿主 Python 程序可显式注册表达式、统计和时序算子,配置按注册名称使用。
高性能积分器 使用独立的 fast validate/explain/run 和 fast batch 入口,对独立 ASC/BLF 文件中的同帧公式逐报文累计;批量任务支持按文件并行。它有明确的支持范围,不会自动替换通用分析。
使用资料 完善功能说明、配置示例及实际结果解释,提供中英文手册和可直接阅读的 AI 指南;教学结果在各分发形式中保留原始字节,并核对摘要。

高性能积分器直接累计每条匹配报文的表达式结果。由功率求能量时,表达式需要包含相应时间因子;如果输入已经是每条报文的能量贡献,则直接累计。性能取决于任务和输入,不能把该入口的测量结果推广到所有分析。

已完成真实数据数值和功能核对,但部分耗时、内存目标仍未达到。请按任务和可用资源选择高性能积分器;本次发布不表示全部性能目标已通过。

从旧版升级

本版只接受整数 config_version: 4。 版本 3 分析配置、旧 INI 专用导入入口及其映射配置已移除。独立批量清单仍使用 batch_version: 1;它与分析配置的版本号分别管理。

请参照当前示例核对旧分析配置,不能只把 3 改成 4 就认为结果含义不变:

  1. 核对信号的输入单位、目标单位和换算系数。conversion.scale 与 conversion.offset 都未填写时,程序按已登记的兼容单位自动换算;显式填写任一个系数时会校验完整换算,不再额外叠加一次自动换算。
  2. 核对表达式中的常量单位、比较与计算输出单位,按当前字段规则处理报错。先用已知输入和结果运行小例,再处理实际日志。
  3. 旧 INI 任务需要按当前分析配置重新表达。YAML/JSON 格式转换命令只转换文件格式,不会自动迁移配置版本或旧业务含义。
  4. 若其他程序读取输出文件,不要依赖固定文件名;使用本次返回的运行清单定位带时间后缀的结果。

字段和示例见信号绑定、表达式、统计与积分、时序变换、自定义分析算子及高性能积分器。

在线中文手册和英文手册对应当前发布版;历史版本手册保存在对应 GitHub Release 中。默认运行依赖与 0.1.0.dev1 相同,CAN 依赖无需额外安装组。

维护者:liuyaoze,work@liuyaoze.com。


Chronaxis 0.1.0.dev2

简体中文

This release expands configuration diagnostics, expressions and statistical calculations, adds the high-performance integrator, and updates the Chinese and English manuals and bundled AI guide. It remains a development preview for Windows and Python 3.12.

Install or upgrade:

python -m pip install --upgrade "chronaxis==0.1.0.dev2"
python -m chronaxis guide --language en

Changes since 0.1.0.dev1

Feature Behavior in this release
Configuration diagnostics Reports multiple independently identifiable configuration errors with field locations. Checks that depend on an invalid field may still require that error to be fixed first.
Result publication Reuses output directories, gives files from one publication a common timestamp suffix, preserves existing files, and publishes the completion manifest last.
Units and expressions Uses analysis configuration version 4, with automatic conversion of compatible units, unit-bearing constants, conditional expressions, numeric and bitwise operations, and one-dimensional calibration lookup.
Statistics and temporal transforms Adds counts, variance, standard deviation, RMS, quantiles and medians, as well as result aggregation, rolling statistics and rates of change.
Custom analysis operators A host Python program can explicitly register expression, statistical and temporal operators, then reference their registered names in configuration.
High-performance integrator Provides separate fast validate/explain/run and fast batch entries for same-frame per-message accumulation in independent ASC/BLF files. Batch tasks can process files in parallel. This entry has a defined supported scope and does not automatically replace general analysis.
User documentation Improves feature explanations, configuration examples and actual result interpretation, with bilingual manuals and an AI guide that can be read directly. Teaching results retain their original bytes across distribution formats and are checked against their summaries.

The high-performance integrator adds the expression result from each matching message. Calculating energy from power requires the appropriate time factor in the expression. Inputs that already represent per-message energy contributions are added directly. Performance depends on the task and input; measurements of this entry do not establish performance for all analyses.

Real-data numerical and functional checks are complete, but some runtime and memory targets remain unmet. Select this integrator according to the task and available resources; this release does not establish that all performance targets have passed.

Upgrading from an earlier version

This release accepts only the integer config_version: 4. Version 3 analysis configurations, the dedicated legacy INI import entry points and their mapping configurations have been removed. Independent batch manifests continue to use batch_version: 1, a separate version from analysis configuration.

Review older analysis configurations against the current examples. Changing 3 to 4 alone does not establish that the analysis retains the same meaning:

  1. Check input units, target units and conversion coefficients. When both conversion.scale and conversion.offset are omitted, registered compatible units are converted automatically. Supplying either coefficient makes the program validate the complete conversion; it does not apply a second automatic conversion.
  2. Check units on expression constants, comparisons and calculation outputs. Resolve errors using the current field rules, then run a small case with known inputs and results before processing actual logs.
  3. Express legacy INI tasks in the current analysis format. The YAML/JSON conversion command changes file format only; it does not migrate configuration versions or old business rules.
  4. If another program reads output files, locate the timestamped files through the returned run manifest instead of relying on fixed filenames.

See signal binding, expressions, statistics and integration, temporal transforms, custom analysis operators, and the high-performance integrator.

The online Chinese manual and English manual cover the current release. Manuals for earlier versions remain attached to their GitHub Releases. Default runtime dependencies are unchanged from 0.1.0.dev1, and CAN support needs no extra dependency group.

Maintainer: liuyaoze, work@liuyaoze.com.