fix: 再部署重置浮游倍率;roster-026、solo-HARD-1、solo-ABYSS-1 不再沿用旧缓存

This commit is contained in:
星闌 committed 2026-10-11 16:40:10 +08:00
1 parent bc50cac01f
commit 7434e9eaf0
8 files changed
+210 -11

No files matched your search

+21
View File
@@ -274,3 +274,24 @@ enemy: picking emitted a unit click with no definition, instead of the tile clic
engine-canvas regression retains the card/100 HP assertions and also covers another field refresh, live HP loss and
closing the card when the crate is destroyed. Device metadata supplied by the new entry still wins; changing fields
does not carry old devices across. Gameplay and allied targeting rules are unchanged.
### 29.15 Funnel ramps on redeployment
PRTS [branch traits, 驭械术师](https://prts.wiki/w/分支特性信息/data#驭械术师) resets drone damage to its initial
scale when the operator redeploys. The profession installer previously initialized the ramp only once, so a
knocked-out Rockrock revived by Ermengard could keep 110% against the same surviving enemy. Normal deployments now
reset the funnel target/scale and the per-target map used by Minimalist and Kjera. Other trait state is retained;
continuous attacks, target changes and the existing free-move carry decision (§22.3) keep their rules. No operator
kit or attack/chase calculation changes.
The regression uses the production Match with seed 8, NORMAL solo mode and Ermengard: buy Caper in round 1,
upgrade in round 2, buy Rockrock in round 5 and leave both fighting normally. In round 6 a real enemy knocks out
Rockrock; automatic revival must start the same living target at 20%. Controlled Battle tests separately cover
normal/elite Rockrock and hidden Kjera, Minimalist/Kjera per-target maps, uninterrupted ramping, target changes,
other-unit deployments, unrelated trait state and free movement.
Golden updates: `roster-026` reduces Minimalist's damage from 3936 to 2397 and updates the affected enemy,
player and snapshot totals; `solo-HARD-1` and `solo-ABYSS-1` reduce the AI lineup's damage after funnel
redeployments (boss damage decreases by 572 and 413 respectively). Their victory, round, RNG and error outcomes
are unchanged. These three scenarios now start each normal deployment with a fresh ramp instead of carrying
the previous deployment's per-target cache.