All prior measurements ran on server 0.35.5. The default is now the current
1.3.1 jar, with the legacy jar kept and switchable via TR_SERVER_JAR (no code
edit). test_gauntlet_5bots.nim no longer clobbers a caller's TR_SERVER_JAR -
it used to putEnv() unconditionally, so an override was silently ignored.
RE-BASELINE (controlled RulesProbe battle, stationary bot, powers 0.1/0.5/1/2/3):
dimension 1.3.1 0.35.5 verdict
bullet damage per hit 0.4/2/4/10/16 identical SAME
bullet speed (20-3p) within noise within noise SAME
post-fire gun heat (1+p/5) identical identical SAME
cooling 0.1/tick 0.1/tick SAME
bulletDamage SCORE exactly 100/round 104..113/round DIFFERENT
bulletKillBonus (20%) 20/round 20..23/round DIFFERENT
LOUD FINDING - a SCORING rule changed, physics did not: 0.35.5 credits
OVERKILL to bulletDamage (the killing bullet's full damage even past 0 energy);
1.3.1 caps it at the energy actually removed. Every 0.35.5 score is therefore
inflated ~5-6%, and bulletKillBonus inherits the inflation. Gauntlet totals
shift accordingly (SittingDuck 1936 -> 1800, WaveSurfer 1886 -> 1669).
Consequence: score-based numbers recorded on 0.35.5 are NOT comparable to 1.3.1.
Our gun A/Bs used real HIT RATE, not score, so those conclusions stand.
Runner 1.0.2 (unchanged, no newer one on the box) is measured compatible with
the 1.3.1 server. Note TrBattleCapture uses the runner's EMBEDDED server, which
is 1.0.2 - so the capture path still runs an older engine than the gauntlet.
Also re-ran acceptance_offline_vs_online on the new default: 12/12.
P0. Three of the five launchers ran ./<Bot> (a tracked binary at the bot root)
while config.nims sets outdir=out and both the test framework's compileBots and
a manual 'nim c src/<Bot>.nim' write to out/. SittingDuck and OscillatorBot
correctly ran ./out/<Bot>; RandomMover, PatternMover and WaveSurfer did not.
cmp confirms the root and out binaries differed for all three.
Consequence: the previous session's adversary fixes were compiled into out/ and
never executed. Every gauntlet and every capture ran the OLD code. This is
almost certainly why the user's instinct that these bots were still bugged was
correct while the code claimed otherwise.
Fixed by pointing all five launchers at ./out/<Bot>, and by deleting the three
stale root binaries so the trap cannot recur. Verified end to end through the
booter: WaveSurfer went from standing still 96.2% of ticks with a 1398-tick
longest standstill, to rest 12.3% / mean speed 6.69 / longest zero run 18 /
perpendicular 0.845.
Also honours GUN_STATS_PATH in test_gauntlet_5bots.nim (same knob ModularBot
reads) so pooled gauntlet runs append to one file instead of clobbering the
default.
NOTE for a follow-up: the out/ binaries are still TRACKED build artifacts, which
is the same class of hazard that caused this. Untracking them (as was done for
ModularBot_garage/ModularBot) would remove the failure mode entirely.
SirStone