A device which doesn't hear the welcome repeats its greeting. Minting a
fresh session for every repeat left both ends chasing each other's
session ids: the device eventually answered with the id it had been
given, the simulator had already moved on and rejected it as
INVALID_SESSION, the device restarted the handshake, and so on twice a
second for as long as the link was up.
The same greeting arriving on a session which has not been used yet is
now answered with the session it already has, and only a different
device, or one whose session had progressed, gets a new id. Repeated
greetings are counted as handshakes and shown in the debug panel, so the
condition which causes them stays visible instead of flooding the log.
Three additions, all driven from the simulator debug panel.
DEVICE_LOG lets a controller send text lines; they reach the simulator log
and a per connection view in the panel which keeps the last 64 of them.
Diagnostic routines: a controller lists the procedures it offers, the
operator picks one from a drop down and starts it, and the controller
reports progress and the result. While a routine runs the controller can
ask the simulator to show a message box with its own text and buttons and
gets back which one was pressed, which is what a valve or gauge
calibration needs. The message box appears whether the debug panel is
open or not.
Links can now be disconnected and reconnected by hand: disconnecting
closes the serial port so the controller can be reprogrammed, and the
session is rebuilt from scratch on reconnect. Ports which appear while
the simulation runs can be picked from a refreshed list, given a baud
rate and added as a new controller, so the manager now starts even with
no uart2 entries in eu07.ini.
The zmq interface kept its own copy of the state field list, its own name
table and its own packer. All three are now the shared registry, which
resolves the same channel names and returns the same values, so the wire
format is unchanged. New state exposed to external interfaces is added in
one place from now on.
The registry reports the raw velocity rather than zeroing it while the
simulation is paused, and a symbol which doesn't apply to the current
vehicle keeps its measurement and says so through its quality flag.
The snapshot sent right after SUBSCRIBE was taken from the previous
frame, which on the first frame of a session meant every value came back
as UNAVAILABLE. The link now publishes the state of the current frame
before the received frames are processed.
Adds the WITH_HARDWARE_PROTOCOL_V2 build option and the eu07.ini entries
which configure it: uart2 declares a serial link and may be repeated for
several controllers, the remaining uart2* entries tune frame size,
heartbeat, rate limits and logging. UART v1 is untouched and both may be
active at the same time on different ports.
The driver mode starts the manager when at least one link is configured
and pumps it from the input polling loop, and the debug panel gains a
Hardware Protocol v2 section reporting link and session condition,
packet rates, round trip time, protocol and transport error counters and
whatever diagnostics the device itself reports, with switches for
protocol and frame logging.
Adds the logical protocol layer on top of the frame codec: session
handshake with negotiated frame size, session scoped symbol handles,
state subscriptions with periodic, on change and deadband filtering,
command events with duplicate protection and continuous controls with
stale update rejection, plus per connection rate limits and diagnostics.
A single worker thread owns all of the blocking transport work and
exchanges finished frames with the simulation thread through short locked
queues, so a stalled, flooding or unplugged controller cannot hold up the
simulation.
Adds the transport-independent core of the MaSzyna Hardware Protocol v2:
COBS framing with a CRC32C protected 24 byte header, bounds checked
payload readers and writers, and a resynchronising stream decoder.
The shared TrainState registry describes simulator state in engineering
units with per-value quality, the command registry generates the command
namespace from Commands_descriptions, and the symbol registry merges both
into the public, name-addressed hardware api.
make_event() lowercased the event name twice: cParser::getToken() already
folds ASCII via toLowerChar() (which intentionally leaves bytes >= 0x80
alone), then ToLower() ran std::tolower() over every byte on top of that.
With a non-"C" global locale, that second pass remaps cp1250 uppercase
accented letters (e.g. 0xAF -> 0xBF). Every lookup path - TTrack::AssignEvents(),
child-event resolution in multi_event, event-launcher name resolution -
normalizes only through the parser, so a registered name containing such a
letter can no longer be found: the track logs "can't find assigned event"
and the binding is silently dropped.
Removing the extra ToLower() makes the registration key match every lookup
site (ASCII-only folding, non-ASCII bytes compared verbatim), and matches
the direction toLowerChar()'s own comment already points at for UTF-8.
A whois with no triggering vehicle previously just wrote nothing to
the target cell with no trace in the log, which is a real pain to
diagnose from the scenery side (this is exactly the kind of thing that
cost real debugging time on a station script recently).
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
whois_event::run_() dereferenced m_activator unconditionally in every
branch (this predates this PR). getvalues_event::run_() and
putvalues_event::run_() already guard the same way elsewhere in this
file; whois_event was missing it. Single early return covers every
branch, old and new.
The +8/+16/+24/+32/+40 (and the new +48) modes all pick the reporting
vehicle the same way: the activator's own Mechanik if it's the primary
driver, otherwise the consist's controlling vehicle (ctOwner) - so the
event still resolves correctly if a wagon happens to trigger it ahead
of its locomotive.
+0 (train name / stations left / stop flag) predates that pattern and
never got it: it required m_activator->Mechanik directly and silently
wrote nothing otherwise. A whois hook placed where the loco doesn't
necessarily lead (e.g. a train that reverses direction, or one being
propelled) can have a wagon trigger the event first and the memcell
stays empty, with no error - hard to diagnose from scenery side.
Same owner-selection, same guard semantics (still nothing written if
neither the activator nor its consist has a controller), only the
source of the vehicle it reads from is widened.
When the Autorewident function is enabled, the EP fuse switch will no longer automatically engage or disengage if the player is controlling the train. This would happen for example when approaching a station.
