No 1995 L4SPLR.CPP survives, but its header does, and the direct descendant
source survives in the RP411 tree. This is that 741-line file back-dated
onto the 1995 engine APIs the header declares -- the BACKFILLS method -- and
the ENGINE COUNT MOVES: 181 -> 182 TUs, link clean, zero unresolved.
What the brick is: the whole mission-review spool layer --
SpoolingInterestManager every interest packet -> the spool
L4SpoolingNetworkManager normal connect + the spool PREAMBLE (app ID,
data version, per-host remote/local + HostID)
L4PlaybackNetworkManager "connect" FROM the spool: verify the preamble,
rebuild the host table with no wire, post Load
L4SpoolingApplication the recorder app -- renderless makers, deferred
stop (spool the ending exactly once), hand the
finished spool to the application manager
SpoolerTask the playback pump: egg packets while waiting,
interest packets while loading, then everything
doled out on recorded timestamps + GetTimeBias()
Back-dating deltas, each named in the header block: the HINSTANCE/HWND ctor
reverts to (ResourceFile*, ApplicationID); SOCKADDR_IN + two-arg
ResolveAddress revert to the 1995 unsigned-long address; L4Host construction
uses NullNetworkAddress/0 (L4NET.CPP's own idiom); PostQuitMessage becomes
Exit_Code = AbortExitCodeID (the loop gate ExecuteForeground already tests);
accessor-style statics revert to the header's static members; and the
later-era MakeGaugeRenderer/Shutdown parameters drop.
Two BC4.52 notes in source: the StopMissionMessage re-post carries an
explicit (Receiver::Message *) cast for the compiler's nested-scope upcast
quirk (NetworkClient derives Receiver, so the type IS right), and the
foreground tick casts to LBE4ControlsManager::Execute exactly as L4APP does
(the 1995 base ControlsManager has no Execute).
KNOWN DEBT, deliberately left: L4PlaybackNetworkManager::DefaultData is
referenced by its ctor but defined nowhere yet -- the linker does not pull
the member because nothing references the playback manager until BTL4PB.CPP
exists. That definition belongs with the playback app increment, which is
next. Regression run clean (same y-track terrain walk).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
CydandClaude Fable 5
f85e5f5bcd