From e693b481b3a5cab0ba394d9bbd64be49e033ea09 Mon Sep 17 00:00:00 2001
From: Joe DiPrima <epilectrik@gmail.com>
Date: Thu, 13 Aug 2026 02:15:45 -0500
Subject: [PATCH] commit-hygiene sweep: (1) the #170 L4STEAMNET.h contract
 comment never landed in 2fe58b8 -- the git add used the lowercase filename
 and the case-mismatch left the tracked (uppercase) path unstaged, silently;
 (2) the gauges-hud #172 correction was a silent no-op -- a python str.replace
 whose multi-line needle didn't match printed 'corrected' anyway. Both applied
 for real now (verified by grep, not by exit code). Note the pattern: two
 silent tool no-ops in one evening -- verify edits by READING the result, not
 by the tool's success message

---
 context/gauges-hud.md        | 15 +++++++++++----
 engine/MUNGA_L4/L4STEAMNET.h |  8 +++++---
 2 files changed, 16 insertions(+), 7 deletions(-)

diff --git a/context/gauges-hud.md b/context/gauges-hud.md
index d3c5756..894f148 100644
--- a/context/gauges-hud.md
+++ b/context/gauges-hud.md
@@ -836,10 +836,17 @@ were three different presentation facts:
   authored item codes 0x80+0x87 collapse to that one lamp, mode 0xffffffff = the ALWAYS-ACTIVE
   sentinel, never mode-gated). NO MFD carries any generator-leak indicator -- binary-authentic
   (@004cc2fc). In glass surround that rail cell renders at the minimumStrip floor (~6px,
-  L4RIOBANK.cpp) -- Elengil's fatal lamp FLASHED for 28.4s, effectively invisible. FIX (glass-
-  only deviation, #154-class): the ALERT BLOOM -- an alternating lamp paints a grown halo
-  (+8px) under the surfaces in BTDrawCockpitPanels; only the outward part shows. Receipt:
-  [lamppx] alert ON/OFF with cell size, per edge.
+  L4RIOBANK.cpp) -- Elengil's fatal lamp FLASHED for 28.4s. ~~"effectively invisible"~~
+  **CORRECTED SAME DAY (user challenge upheld, bench-verified):** the strip PAINTS AND BLINKS
+  correctly -- [lampblink] receipts show the painted shade alternating at a metronomic 125ms,
+  full frame rate, in a loop code-identical to 913's; Elengil's layout receipts show the stock
+  surround with the lamp on screen (~126px wide x ~6px strip @4Hz). The failure is
+  EXPECTATION, not visibility: testers watch the MFDs, and no MFD carries a generator-leak
+  indicator (authentic). Resolution: the handout's WHERE-TO-LOOK entry; the ALERT RING
+  deviation exists but is GATED OFF (BT_ALERT_RING=1) pending Friday evidence that testers
+  miss the lamp even knowing its location. Receipts: [lamppx] alert edges, [lampblink] shade
+  transitions (BT_LAMP_LOG), [seqloop] voice Start/Stop. Era question for Oracle: how
+  prominent was the pod's physical generator button light?
 - **Condenser (loop) leak -> the per-loop button** (@0051d058, 1-based '2F 2E 2D 2B 2A 29';
   the 0x2c gap is AUTHENTIC -- 0x2c belongs to the DISTINCT coolingLoop set of the code<0x80
   path). Also always-active. No gate, immediate flash.
diff --git a/engine/MUNGA_L4/L4STEAMNET.h b/engine/MUNGA_L4/L4STEAMNET.h
index e5552b4..500f958 100644
--- a/engine/MUNGA_L4/L4STEAMNET.h
+++ b/engine/MUNGA_L4/L4STEAMNET.h
@@ -33,9 +33,11 @@ int
 void
 	BTSteamNet_Pump();					// callbacks + rx drain (game thread)
 void
-	BTSteamNet_ShutdownAll();			// #163: ordered transport death for
-										// process exit -- close every
-										// connection, then SteamAPI_Shutdown
+	BTSteamNet_ShutdownAll();			// #163/#170: transport QUIESCE for
+										// process exit -- gate the seam, close
+										// every connection (linger), drain.
+										// NO SteamAPI_Shutdown (#170: NULL-call
+										// crash); caller MUST TerminateProcess
 
 //
 // The wire-seam surface (BTNet* wrappers in L4NET.CPP + the marshal).