Skip to content

chore(release): 0.5.0 - #19

Merged
sangowu merged 1 commit into
mainfrom
chore/release-0.5.0
Aug 7, 2026
Merged

sangowu merged 1 commit into
mainfrom
chore/release-0.5.0

Conversation

@sangowu

@sangowu sangowu commented Aug 7, 2026

Copy link
Copy Markdown
Owner

发布 0.5.0,并按你的要求重写了 highlights 的收录规则。

为什么是 minor 而不是 patch

两条都不满足 patch 的"行为不变"承诺:

  • 新增了公开命令 jobradar cache prune-scores
  • 行为变更用户可感知且影响成本:换 CV 后第一次搜索会重新评估缓存职位;UI 分数来源从混合分制变为统一 0~100;此前被旧 CV 拒绝而不可见的职位会重新出现

收录规则改写

原文(三语):

From 0.5.0 onward each release is compared against the previous release.

问题在于它只说了"要对比",没说什么样的版本该收录,字面读下来像是每个版本都得进这个文件——包括纯重构和纯文档版本。改为:

From 0.5.0 onward, a release appears here when it changes performance or user-facing behavior, and is described against the previous release. Which kind of change it is determines what gets compared: performance work is compared on measured figures, behavior work on what the tool now does differently. A release that changes neither is not listed.

三段逻辑:何时收录对比什么(按改动性质分流)何时不收录。中西文同步改写,非直译。

0.5.0 的 highlights 没有性能数字

这批改动是正确性修复,没跑基准,也不该造数字(文件开头就写着"本文件不产生任何自己的数字")。所以 0.5.0 一节明确写成行为版本,并附成本说明:本版本增加而非减少评估量

连带处理了一个会误导人的地方:三个 README 的 Performance 段落原本写"最新版本 0.4.0",若不动,读者会以为 0.4.0 仍是最新版。改为说明"最近一次性能改动来自 0.4.0,当前版本 0.5.0 改的是行为而非速度"。

CHANGELOG

[Unreleased]## [0.5.0] — 2026-08-07,并新增 ### Validation 条目 f14e369b47dfc0,沿用既有格式(两个 arm、held constant、数字、明确 verdict):

  • assess 可捞出的职位:0 → 205
  • 换 CV 无法重开的缓存职位:95 → 0
  • 当前 CV 下有 match 的职位:93 → 293(补跑 200 条,0 失败,173 s)
  • 参与 effective_score 的分制:2 → 1
  • 缓存命中判定实测:传 cv_hash 得 0 skipped / 5 re-assessed,legacy 路径同样数据得 5 / 0
  • Verdict 明确声明不主张任何性能或成本改进,延迟未测量

highlights 里的每个数字都能追溯到这一条,符合文件自身的规则。

版本号

pyproject.toml 0.4.0 → 0.5.0,uv lock 已同步(避免重演 #14 那次不一致),jobradar --version 实测输出 v0.5.0

全量 241 passed

🤖 Generated with Claude Code

https://claude.ai/code/session_012DbPx4FbswofbGn41anE1q

Scoring became CV-aware: a match score is owned by the CV that produced it, so
changing CV re-opens every cached job instead of reusing a verdict reached for
another profile. Minor rather than patch because the release adds a public
command (`cache prune-scores`) and changes observable behavior and cost.

Also restates the release-highlights inclusion rule. It read "from 0.5.0 onward
each release is compared against the previous release", which said nothing about
releases with nothing to compare, and implied every release belongs in the file.
It now states when a release is listed, that the kind of change decides what is
compared -- measured figures for performance work, changed behavior otherwise --
and that a release altering neither is left out.

The 0.5.0 highlights therefore carry no performance figures, and the README
Performance sections say so explicitly rather than letting 0.4.0's numbers read
as current.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012DbPx4FbswofbGn41anE1q
@sangowu
sangowu merged commit e863ed0 into main Aug 7, 2026
2 checks passed
@sangowu
sangowu deleted the chore/release-0.5.0 branch August 7, 2026 22:06
sangowu added a commit that referenced this pull request Aug 7, 2026
Scoring became CV-aware: a match score is owned by the CV that produced it, so
changing CV re-opens every cached job instead of reusing a verdict reached for
another profile. Minor rather than patch because the release adds a public
command (`cache prune-scores`) and changes observable behavior and cost.

Also restates the release-highlights inclusion rule. It read "from 0.5.0 onward
each release is compared against the previous release", which said nothing about
releases with nothing to compare, and implied every release belongs in the file.
It now states when a release is listed, that the kind of change decides what is
compared -- measured figures for performance work, changed behavior otherwise --
and that a release altering neither is left out.

The 0.5.0 highlights therefore carry no performance figures, and the README
Performance sections say so explicitly rather than letting 0.4.0's numbers read
as current.

Claude-Session: https://claude.ai/code/session_012DbPx4FbswofbGn41anE1q

Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
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.

1 participant