Run routerHBP group controls before bridge scans, pass UDP receive time
into dmrd_received for rate/timeout parity with legacy, and set RX_START
on new HBP streams in the router. Talker Alias behaviour is unchanged.
Add idle (5s) and max (180s) recording timers, synthetic VTERM when the
hotspot ends without terminator, and non-blocking reactor playback.
Handle hotspot re-key mid-PTT without a second VHEAD and renumber playback
seq so bridge ingress does not drop frames. Restore prohibited static bridge
legs (TG 9990 ECHO) after bridge reset and options refresh.
Align OpenBridge group voice with bridge_master.routerOBP: use the legacy
RATE DROP formula, run loop control only from the second packet onward, and
send BCSQ once on loop loser instead of periodic retransmits.
Notify report clients immediately after SIGHUP reload, emit END events when
removing systems, and keep OPTIONS/TS1_STATIC/TS2_STATIC when merging YAML.
Reload GLOBAL, REPORTS, ALIASES and SYSTEMS without full restart:
add/remove UDP listeners (OBP, GENERATOR expansion), update bind
addresses, preserve PEERS/STATS and active streams on unchanged ports.
Inject Talker Alias on HBP bridge/repeat via standalone DMRA packets
and embedded LC in forwarded DMRD voice (FLCO 4–7) for MMDVMHost/
DMRGateway. UTF-8 encoding (format 2) and LC/TA superframe alternation
for the full stream; clear embed state on VTERM.
When LOGGER.ENABLED is false, setup_logging attaches only a NullHandler
so legacy configs without the key keep current behavior. Document the
option in EN/ES configuration guides and example YAML templates.
Use a flat OBP STATUS dict like legacy routerOBP, iterate it in
stream_trimmer_loop, stop HBP STATUS[stream_id] writes, cap BCSQ log
dedup with a bounded deque, and sweep stale voice-announcement entries.
Add reopen_file_handlers() and register SIGUSR2 after logging setup in
main and parrot. Document create/postrotate/kill -USR2 in monitoring
guide (EN/ES). SIGUSR2 only reopens file handlers; does not reload YAML.
Skip options_config_for_system when static TS/timer fingerprint matches the last
successful RPTO apply to avoid log spam and redundant work.
Document why make_single_bridge must not rebuild 9990–9999 when the bridge
already exists: it drops the ECHO ACTIVE row and breaks the parrot path.
When BRIDGES[tg] existed without TS1/TS2 rows for a MASTER, make_static_tg
only updated existing rows and that hotspot never received traffic. Append
inactive legs like make_stat_bridge / legacy UA re-add.
- Compute rate from elapsed time since stream START (not packets/START).
- Drop duplicate frames by CRC even when seq is 0 (retransmissions).
- Remove _log_trace helper and TRACE calls for BCSQ/BCKA/ACL skips.
- on_obp_bcsq_received: clear forward STATUS only; do not send_bridge_event
- Loop-control loser: report END,TX instead of END,RX (no LH 0s / global OBP pop)
- Sync TARGET_SOCK from actual peer when RELAX_CHECKS accepts DMRD/DMRE
- Fallback TARGET_IP/PORT for BCSQ send; warn if no destination
- Resend BCSQ every 2s while loop-control loser (UDP loss)
- Emit END,RX on first loser transition for monitor
- Robust _bcsq TG key matching in to_target
- Emit END,TX for OBP forward destinations when source stream times out in trimmer
- After VTERM END,RX, sweep remaining forward legs so monitor state stays consistent
- Log MAX HOPS drops at DEBUG to avoid per-packet warning floods
- Drop obsolete OBP_BRIDGE_MESH_OWNER_BYPASS_BCSQ comments from example yaml
Match bridge_master/hblink behaviour: BRIDGES table iteration for
to_target, OBP VTERM/_fin timing, stream seq/crc handling, and
document related GLOBAL options.
- Match BRIDGES source rules with effective TS1 for OpenBridge (has_source).
- Normalize DMRE receive slot to 1 for OPENBRIDGE before STATUS/dmrd_received.
- Dedupe OBP (SYSTEM, TS) per packet like legacy sysIgnore (TG + reflector lists).
- Run dynamic-bridge deactivate for dst 4000 before ACL (group/unit/vcsbk; master+peer)
- Send PRIVATE VOICE,END,RX to report; set slot RX_START on private stream start; fix slot STATUS setdefault
- dmrd_received: return early for private call to 4000 (legacy parity)
- Replace _broadcast_active with _broadcast_active_tgs (set)
- Serialize only same-TG broadcasts; different TGs run in parallel
- Add _mark_slots_busy/_mark_slots_free for TS conflict prevention
- Add ffmpeg system dependency note in requirements.txt
Voice words and ondemand files for en_GB, es_ES, cy_GB, de_DE, el_GR, fr_FR, pt_PT, th_TH, CW, etc.
Remove Audio from .gitignore to track these assets.
- Use DefaultVoiceProvider when Audio dir exists (was gated by ANNOUNCEMENT_LANGUAGES)
- Create Audio dir at startup when voice config present
- ANNOUNCEMENT_LANGUAGES optional (ident only); each item has own LANGUAGE
- ident: fallback to per-system ANNOUNCEMENT_LANGUAGE when empty
- disconnected_voice: use system language directly
- Clarify ENABLED, ANNOUNCEMENT_LANGUAGES in adn-voice.example.yaml
- Add Audio/.gitkeep, update .gitignore to keep folder structure
- Port playback.py to clean architecture: PlaybackUseCases, parrot_main.py, adn-parrot.py
- Fix critical PEER auth bug: RPTACK/MSTNAK data offsets were off by 1 (5→6) preventing PEER login
- Extract config normalization to shared config_normalizer.py (used by both main and parrot)
Rodrigo Pérez