Refactor bullet mechanics: add HEAT/SUBCALIBER/FRAG enum types, penetration loss, and fragmentation damage#3
Conversation
…ration loss, and fragmentation damage - Add HEAT, SUBCALIBER, and FRAG to BulletType enum, deprecate isHeat boolean (backwards compatible) - Add fragmentation parameters: fragConeAngle, fragHitProbability, fragDamage to JSONBullet - Implement penetration loss for kinetic (non-HEAT) rounds based on armor thickness ratio - Implement fragmentation damage to internal parts and crew when FRAG bullet penetrates armor https://claude.ai/code/session_01L7GeMnAYtRwxzvzDwnTyRH
…ll-rework
- Fix frag crew damage: use PacketEntityBulletHitExternalEntity for proper
client/server handling (was using server-only rider.attack() which never
executed since bullet checks run on client)
- Adapt HEAT shell behavior from heat-shell-rework branch:
- HEAT shells always detonate on armor contact (return ARMOR, stop flying)
- On successful penetration, apply post-pen damage proportional to residual
penetration ratio to both vehicle entity and all crew members
- Add missing PacketEntityBulletHitExternalEntity import
https://claude.ai/code/session_01L7GeMnAYtRwxzvzDwnTyRH
… messages - coneRange was diameter/100 (0.75m for 75mm) — far too small to reach any internal parts. Changed to diameter/10 (7.5m for 75mm) - Removed distToPart > 0 check that filtered parts at the hit point - Added zero-length vector guard to prevent NaN from acos - Added FRAG ACTIVATED debug message at block entry to confirm code is reached - Enhanced hit debug messages with distance and angle info for diagnostics https://claude.ai/code/session_01L7GeMnAYtRwxzvzDwnTyRH
- Add MISSILE type for guided projectiles and BOMB type for unguided drops - Simplify all new debug messages to match mod's terse style https://claude.ai/code/session_01L7GeMnAYtRwxzvzDwnTyRH
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: bd72b0275f
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| InterfaceManager.packetInterface.sendToServer(new PacketEntityBulletHitEntity(bullet.gun, part, fragDamage)); | ||
| } else { | ||
| EntityBullet.performEntityHitLogic(part, fragDamage); | ||
| } |
There was a problem hiding this comment.
This branch applies damage to part while attackProjectile is iterating allParts, and APart.attack() can call removePart(), which rebuilds allParts. In server-side bullet simulation (for isLongRange bullets or NPC-fired bullets), a FRAG hit that destroys a part will mutate the collection mid-iteration and can raise ConcurrentModificationException, aborting hit processing. Iterate over a snapshot/list copy (or pre-collect targets) before applying fragment damage.
Useful? React with 👍 / 👎.
| //If the bullet has FRAG type and penetrated armor, deal fragmentation damage to internals. | ||
| if (bullet.definition.bullet.types.contains(BulletType.FRAG) && bullet.definition.bullet.fragDamage > 0) { | ||
| float coneAngle = bullet.definition.bullet.fragConeAngle > 0 ? bullet.definition.bullet.fragConeAngle : 45.0f; | ||
| float hitProbability = bullet.definition.bullet.fragHitProbability > 0 ? bullet.definition.bullet.fragHitProbability : 0.5f; |
There was a problem hiding this comment.
Using > 0 to decide whether to apply a default makes fragHitProbability: 0.0 impossible: it is treated as "unset" and replaced with 0.5, so fragments still hit 50% of the time. That breaks the documented 0.0–1.0 behavior and can cause unexpected internal/crew damage when pack authors intentionally configure zero chance. Defaulting should use a distinct sentinel (for example negative values) or clamp the provided value directly.
Useful? React with 👍 / 👎.
2 participants
https://claude.ai/code/session_01L7GeMnAYtRwxzvzDwnTyRH