 Jesse_ChenandClaude Opus 5
|
63c419f375
|
perf(consultation): 健康探测加 TTL、yoga 表达式只编译一次
Independent Staging Quality Gate / validate (push) Canceled after 27s
Independent Staging Quality Gate / publish (push) Canceled after 0s
BUG-734: probe_official_rest_health 不是 ping —— 它把一份虚构 smoke 排盘
POST 给官方 HoroscopePredictions(实测单次 1,034 ms),还经 call() 吃掉对方
每分钟 5 个限流令牌之一,而结论零 TTL。连续 6 轮前台请求下,第 6 次探测会被
本地限流器挡下、反过来报出假的 official_blocked,真正的业务调用还要和它抢令牌。
加进程级 TTL 缓存(成功 60 s / 失败 10 s,配置变即失效,缓存键只记 key 配没配
的布尔)。force_refresh 默认 True:诊断端点调的是裸 gateway_status() 且该文件
本轮不得改,默认实时才不会让运维看到旧结论;run_gateway_packet 显式走 False。
探测在锁外执行。6 轮实测:探测 6→1、握手 5→1、限流令牌 6→1。
连接复用(任务书 6.2)按让步顺序砍掉:实测 urllib.request 没有连接池,模块级
共享 opener 在 6 次请求下仍开 6 条 TCP 连接,按任务书写法只能骗过「同一个
opener 实例」的断言而省不掉任何握手。另立单。
BUG-735: yoga_engine._eval_custom 对静态规则表的表达式每请求重新编译。实测每次
检测 386 次源码编译(182 次 eval 字符串 + 204 次 ast.parse/compile)。拆出
_build_custom_exec_globals(每次新建,绑定当前盘 ctx)与 lru_cache 的
_compiled_eval_code / _compiled_exec_code,只缓存 code object,绝不缓存求值
结果。求值语义逐条对齐改前。第二次检测起编译 0 次,单次 27.4 ms → 3.3 ms。
等价用同进程差分(BUG-733 的做法):192 条去重表达式 × 3 张公开虚构示例盘,
与逐字复刻的改前实现同值,另有反向验证证明差分非恒真。
未改 scripts/jyotish_api_server.py。既有断言一条未改。新增 23 条测试。
快速门唯一红的 test_chat_page_uses_authenticated_cloud_persistence 经干净
origin/staging 检出复跑确认为基线红,记入 BLOCKED.md BLK-002,不顺手修。
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JUei7K13cYxLHE3Axe4A45
|
2026-09-16 02:01:33 +00:00 |
|
732642856
|
04d901e1be
|
Clarify benchmark boundaries and improve annual workflow surfaces
|
2026-06-27 19:09:31 +08:00 |
|
732642856
|
066c59bee3
|
v6.1.11: Solar/Lunar Yogas恢复 + 全量碎片审计
从Git dangling commit f19369c恢复遗失代码:
- scripts/yoga_engine.py: 新增_detect_solar_lunar_yogas()
- Solar Yogas: Veshi/Voshi/Ubhayachari (太阳第2/12宫判断)
- Lunar Yogas: Sunapha/Anapha/Durudhura (月亮第2/12宫判断)
- technique_registry: 新增solar_lunar_yogas条目 (covered)
全量碎片审计完成(28个文件 + Git + Desktop + Downloads + auto-generated)
- Brain: 5份P0文件提取待整合
- Git: 7个dangling commits确认 - f19369c代码已恢复,其余自然吸收
- New open-source: 12个项目发现 (vedic-calc/vedic-astro-skills/VedicAstro等)
|
2026-06-11 11:28:31 +08:00 |
|
732642856
|
0406723b7e
|
v6.1.3: D9/D60 engine upgrade + 5 bottleneck rules tuned
Engine (yoga_engine.py):
- Added D60/Shashtiamsa context injection (d60_house_of, d60_sign_of, etc.)
- Added D9 dignity queries (is_exalted_in_d9, is_own_sign_in_d9, etc.)
- Added functional_malefics() / functional_benefics()
- Added is_shashtiamsa_evil/good() and vaiseshikamsa_score()
- Exposed all new methods to rule expression sandbox
Data (_compute_one_chart.py):
- Generate D60 positions alongside D1/D9
- Inject D60 into context for all computed charts
Rules (yoga_rules.json):
- thrikaala_gnana: D1 strong = exalted/own/moola only (kendra/trikona removed)
+ D9 strong check (at least 1 of 3 in D9 exalted/own/moola)
FP: 12→0, FN: stable
- dharidhra: restored method1 (L2/L11 in dusthana + unfriendly navamsa + malefic aspect)
FN reduced
- bandhubhisthyaktha: added is_shashtiamsa_evil(lord_4)
- nishkapata: added vaiseshikamsa_score(lord(1)) >= 15
- kapata: reverted functional_malefics (caused 22 FN), back to MALEFICS
Validation:
- FP: 45→39 (-6)
- FN: 72→67 (-5)
- F1: 94.35%→94.88% (+0.53%)
- Precision: 95.60%→96.18% (+0.58%)
New scripts:
- scripts/add_d60_to_test_charts.py: batch D60 augmentation
- scripts/ashtottari_dasha.py, yogini_dasha.py, kalachakra_dasha.py
- scripts/report_orchestrator.py
D60 data added to all 60 standard test charts.
|
2026-06-07 14:44:53 +08:00 |
|
732642856
|
bc07add030
|
Release v6.0.52: fix same_house(self) bug globally in yoga_engine\n\n- Removed 'if a==b: return False' guard from same_house() function\n- A planet is trivially in the same house as itself (lord(X) == planet case)\n- Prior workaround in mathibhramana variation_2 now redundant but harmless\n- FN: 79->74 (-5). Recall: 92.47%->92.95% (+0.48%). F1: 93.54%->93.80% (+0.26%)\n- kapata: 8->5 FN (3 auto-fixed from global fix)\n- Other rules with same_house(lord, planet) patterns also benefit
|
2026-06-07 09:07:33 +08:00 |
|
732642856
|
26835f80b8
|
Release v6.0.43 Yoga F1 93.76%
- Fix 3 PyJHora source bugs (kalaanidhi Yoga JUPITER_ID->VENUS_ID x2,
krisanga Yoga missing return)
- Implement planet_in_house_with_aspect_or_conjunction semantic condition
- Fix lagna_lord_in_dry_sign: dry_signs set correction (Capricorn->Taurus),
add ll_owner_dry check using dynamic co-lord resolution
- Regenerate reference data against fixed PyJHora
- F1 93.28% -> 93.76% (FN 96->86, FP 44 unchanged)
|
2026-06-06 09:58:28 +08:00 |
|
732642856
|
86e7d03c4a
|
Release v6.0.41 Yoga source parity
|
2026-06-05 23:20:53 +08:00 |
|
732642856
|
ede8b96d2c
|
Release v6.0.40 Yoga source parity
|
2026-06-05 22:37:25 +08:00 |
|
732642856
|
94be318d56
|
Release v6.0.39 Yoga source parity
|
2026-06-05 22:03:06 +08:00 |
|
732642856
|
58fc615b16
|
fix: align yoga source rules
|
2026-06-05 20:31:14 +08:00 |
|
732642856
|
975f00f5f2
|
fix: improve context-aware yoga precision
|
2026-06-05 18:00:00 +08:00 |
|
732642856
|
93610326fe
|
feat: add yoga context validation fixtures
|
2026-06-05 17:16:23 +08:00 |
|
732642856
|
4b840b5c2a
|
fix: tighten yoga precision pass
|
2026-06-05 16:39:05 +08:00 |
|
732642856
|
44221191f6
|
fix: align high-impact yoga logic
|
2026-06-05 16:17:38 +08:00 |
|
732642856
|
8da45c56b3
|
feat: add yoga logic validation and varga fixes
|
2026-06-05 14:30:20 +08:00 |
|