Новый модуль src/radio (radio/radio_audio/radio_headless) — push-to-talk
радио поверх topo-групп, доставка аудио в канал, headless-режим и тесты.
Интеграция в chatgui (ptt_hotkey_manager, radio_audio_engine, радионастройки,
вызовы) и chatgui-android (RadioAudioEngine, RadioOverlayService, UI).
- media cache: удаление сирот на старте (сверка ФС с сообщениями) + лимит
storage_total_size (удаление самых старых файлов, только старше backfill_days)
- ma_dir_walk(): кроссплатформенный рекурсивный обход каталога (POSIX/Windows)
- async_dns: первый драйв slow-пути отложен через call_soon (исключён
double-free при синхронном завершении запроса); cancel унифицирован
- chat: видимый WARN 'media too large' + честный счётчик started/skipped
- chatgui: просмотрщик изображений (ImageViewerWindow) с зумом/поворотом
- chatgui: отправка картинок как image с определением разрешения и типа
- chatgui: у видео-пузыря строка info (res·size·bitrate) и бейдж только длительность
- chatgui: контекстное меню медиа (сохранить в файл/скачать)
- chatgui-android: большие emoji-only сообщения, битрейт/разрешение у видео
- chatgui-android: сохранение неактивных интерфейсов как закомментированные секции
- chatgui-android: отправка картинок по расширению (jpg/png/gif/bmp/webp)
- photo_sender: content-type и расширение по типу файла
Добавлен 64-битный reset_id (эпоха ресета) в INIT-фреймы и STCP-handshake:
- генерится один раз при создании conn, меняется только при фатальном ресете
(normalizer desync / bad fragment size);
- при приёме чужого id slave принимает эпоху пира и ресетится, master держит свою;
- первый id принимается без ресета (свежие при init);
- «tcp peer clean» реинит гейтится по смене id — нет повторного сброса seq.
Это устраняет коллизию seq=1 (RX dup) из-за двойного реинита нескольких TCP-линков,
из-за которой NODEINFO пира терялся и member не возвращался online.
Также: chat member online теперь из topo_group (без БД/merkle), тест test_etcp_seq_collision.
- u_async.c: process immediate_queue BEFORE epoll_wait in uasync_poll
- stcp_link.c: defensive NULL etcp_conn/etcp_link in stcp_link_close
- utun_instance.c: move stcp_server_list_destroy_all to Phase J (before ETCP),
add deferred drain in both Phase J and L
- stcp.h/c: STCP_HS_ENC_CLIENT 6→38, STCP_HS_ENC_SERVER 7→39,
exchange ed25519_pubkey in both handshake directions
- etcp_connections.c: copy peer_ed25519 from STCP to ETCP_CONN in
etcp_link_enter_ready_tcp; guard link->etcp in tcp_link_close_cb
- chat_sync.c: fix lk->conn NULL dereference for TCP links
- test_stcp.c: update stcp_server_create/stcp_client_connect callsites
- tools/chat_tcp_test: change transport udp→tcp
- node_put/node_update_verified: set online=1 on INSERT (nodes created when conn is UP)
- channel_peers_all: fix peer_needed calculation (atl byte was missing)
- cs_send_channel_join: add proto byte to address serialization matching all parsers
Android chatgui:
- New MemberListScreen with member list, detail card, edit dialog
- Link status dots (green/red/grey) per address in detail card (1s poll)
- Self-node name coloured green
- ChatScreen: people icon in TopAppBar
IPv6 autoconnect fixes (3 bugs in conn_mgr/topo):
- topo_node_sqlite_node_load: load v6 addresses from DB
- topo_node_registry_store: merge v6_addrs into existing node
- cm_sock_v6_classify: wildcard [::] treated as CM_V6_DIR
- cm_add_v6_link: htons(port) was missing
Memory fixes:
- strdup→u_strdup in bridge (7 instances)
- u_realloc cap update before success check (3 instances)
- char sql[256]→[512] for truncated SELECT
Chat member hash now includes addresses for change detection
Linux chatgui:
- Link status dots in AccountList detail panel (1s poll)
- Self-node name in green in member list
lottie_animation_from_data uses 'key' as a cache key for parsed models.
All LottieIcon instances were created with key="", causing rlottie's
model cache to return the first loaded animation for all subsequent
requests. Only visible on Windows where rlottie cache is enabled by
default.
Fix: pass unique animPath as key to each LottieIcon constructor.
Also replaced fprintf(stderr) diag with GUI_INFO logging via
AnimTimer::logRegisteredIcons() in debug_ui.h.
- Add last_msg_at column to channels table
- Update last_msg_at on every message insert (send/receive)
- Parameterize getChannels() with ChannelSortMode enum (Alphabetical/Recent)
- Add getChannelLastMessage() for preview text in channel list
- Add sorting combo to Profile settings page (stored in ui_state)
- Reload channels on message receive when sort is recent
- AnimTimer::shutdown(): stop QTimer, clear m_icons before icon destruction
- ~MainWindow: AnimTimer::shutdown() -> cleanupAnimations() (correct order)
- main: replace ExitProcess with TerminateProcess(GetCurrentProcess,0)
TerminateProcess skips DLL_PROCESS_DETACH, avoiding libwinpthread
pthread_join hang on rlottie/miniaudio threads.
rlottie C API has no global shutdown - internal RleTaskScheduler threads
cannot be stopped through public API. Same for miniaudio WASAPI threads.
All owned resources already freed before TerminateProcess.
Root cause: librlottie RleTaskScheduler creates 32 worker threads on first
lottie_animation_from_data call. These pthreads wait in WaitForMultipleObjects(INFINITE)
and never exit. On MinGW, exit()/pthread_join blocks forever.
Changes:
- messagelist: s_icons moved to file-static; cleanupAnimations() clears them,
calling lottie_animation_destroy BEFORE static destruction
- mainwindow: ~MainWindow calls MessageList::cleanupAnimations() after node stop
- main: ShutdownSentinel counts threads at exit; if >4 remaining, calls
ExitProcess(0) to bypass CRT pthread_join hang
- main: if rlottie properly shuts down (<=4 threads), normal return (no ExitProcess)
Signal handler via g_shutdown + uasync_wakeup ensures clean uasync poll exit.
updateReadPosition() now saves both lastReadMsgId and lastPosMsgId on every
scroll stop (300ms debounce), not just when read position changes.
connect(stopped, quit) triggers graceful Qt shutdown after uasync cleanup.
QString::fromUtf8(ch_id) read past length-prefixed string, capturing garbage.
Use fromUtf8(ch_id, ch_id_len) — length already available from bridge but was unused.
Messages now appear on refresh and unread badge correctly matches channel.
evgeny
Evgeny