Files
TeslaRel410/emulator
CydandClaude Fable 5 b499c88cb7 BT410 5.3.137: the ammo cooks off -- a mech can finally be gutted by its own magazine
The blocking identity is settled: DamageZoneClassID == 78 == 0x4e,
computed from our own VDATA enum, so the binary's collector filter
selects DAMAGE ZONES and the blast is shared across the mech's zones.

Landed: AmmoBin::AmmoBinSimulation -- a Performance the class never had,
installed under the same master+dynamic gate the binary's ctor uses --
and AmmoBin::CookOff.  A bin whose fed weapon reaches heat failure
latches a randomised fuse; when it burns down the bin is destroyed and
its remaining rounds go off inside the mech, damage scaled by rounds
remaining and SHARED across the zones, every victim announced.

Correcting my own 5.3.136 write-up: heat clearing does NOT cancel the
countdown.  Both the arm gate and the cancel read the same cell -- the
ammo alarm settled on spent -- so once armed, only spending or destroying
the bin calls it off.  Letting the gun cool does not save you.

Live on the rig: three bins arm, the fuse spread sends them off on
different frames rather than together, 20 rounds x 25 = 500 damage over
22 zones from one bin and 16 x 50 = 800 from another, feeding straight
into the existing destruction cascade -- Searchlight and ThermalSight
destroyed as a consequence.

[T1] recorded: the per-round charge is a stand-in.  The binary reads it
from the bin's streamed explosion record, in the block our cookOffState
reserves and nothing parses yet; we use the fed weapon's authored
per-shot damage until that record is read.

And a mistake worth keeping in the notes: the first attempt sourced the
charge from heatPerRound, a heat figure of about 3e-9, so the detonation
computed 4.8e-08 of damage and did precisely nothing -- a mechanic that
fires, announces itself, and is silently inert.  Only printing the
arithmetic caught it.  Log the numbers, not just the event.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-08-12 23:27:09 -05:00
..