something here broke it

- accounting for rounded down bytes

CMakeLists.txt

@@ -65,5 +65,7 @@ target_link_libraries(solanaceae_sha1_ngcft1 PUBLIC
 	sha1::sha1
 	solanaceae_tox_contacts
 	solanaceae_message3
+	solanaceae_object_store
+	solanaceae_file2
 )
 

solanaceae/ngc_ext/ngcext.cpp

@@ -3,8 +3,8 @@
 #include <iostream>
 
 NGCEXTEventProvider::NGCEXTEventProvider(ToxEventProviderI& tep) : _tep(tep) {
-	_tep.subscribe(this, Tox_Event::TOX_EVENT_GROUP_CUSTOM_PACKET);
+	_tep.subscribe(this, Tox_Event_Type::TOX_EVENT_GROUP_CUSTOM_PACKET);
-	_tep.subscribe(this, Tox_Event::TOX_EVENT_GROUP_CUSTOM_PRIVATE_PACKET);
+	_tep.subscribe(this, Tox_Event_Type::TOX_EVENT_GROUP_CUSTOM_PRIVATE_PACKET);
 }
 
 #define _DATA_HAVE(x, error) if ((data_size - curser) < (x)) { error; }
@@ -112,6 +112,8 @@ bool NGCEXTEventProvider::parse_ft1_init_ack(
 	_DATA_HAVE(sizeof(e.transfer_id), std::cerr << "NGCEXT: packet too small, missing transfer_id\n"; return false)
 	e.transfer_id = data[curser++];
 
+	e.max_lossy_data_size = 500-4; // -4 and 500 are hardcoded
+
 	return dispatch(
 		NGCEXT_Event::FT1_INIT_ACK,
 		e
@@ -224,6 +226,41 @@ bool NGCEXTEventProvider::parse_ft1_message(
 	);
 }
 
+bool NGCEXTEventProvider::parse_ft1_init_ack_v2(
+	uint32_t group_number, uint32_t peer_number,
+	const uint8_t* data, size_t data_size,
+	bool _private
+) {
+	if (!_private) {
+		std::cerr << "NGCEXT: ft1_init_ack_v2 cant be public\n";
+		return false;
+	}
+
+	Events::NGCEXT_ft1_init_ack e;
+	e.group_number = group_number;
+	e.peer_number = peer_number;
+	size_t curser = 0;
+
+	// - 1 byte (temporary_file_tf_id)
+	_DATA_HAVE(sizeof(e.transfer_id), std::cerr << "NGCEXT: packet too small, missing transfer_id\n"; return false)
+	e.transfer_id = data[curser++];
+
+	// - 2 byte (max_lossy_data_size)
+	if ((data_size - curser) >= sizeof(e.max_lossy_data_size)) {
+		e.max_lossy_data_size = 0;
+		for (size_t i = 0; i < sizeof(e.max_lossy_data_size); i++, curser++) {
+			e.max_lossy_data_size |= uint16_t(data[curser]) << (i*8);
+		}
+	} else {
+		e.max_lossy_data_size = 500-4; // default
+	}
+
+	return dispatch(
+		NGCEXT_Event::FT1_INIT_ACK,
+		e
+	);
+}
+
 bool NGCEXTEventProvider::handlePacket(
 	const uint32_t group_number,
 	const uint32_t peer_number,
@@ -247,7 +284,8 @@ bool NGCEXTEventProvider::handlePacket(
 		case NGCEXT_Event::FT1_INIT:
 			return parse_ft1_init(group_number, peer_number, data+1, data_size-1, _private);
 		case NGCEXT_Event::FT1_INIT_ACK:
-			return parse_ft1_init_ack(group_number, peer_number, data+1, data_size-1, _private);
+			//return parse_ft1_init_ack(group_number, peer_number, data+1, data_size-1, _private);
+			return parse_ft1_init_ack_v2(group_number, peer_number, data+1, data_size-1, _private);
 		case NGCEXT_Event::FT1_DATA:
 			return parse_ft1_data(group_number, peer_number, data+1, data_size-1, _private);
 		case NGCEXT_Event::FT1_DATA_ACK:

solanaceae/ngc_ext/ngcext.hpp

@@ -70,8 +70,6 @@ namespace Events {
 
 		// - X bytes (file_kind dependent id, differnt sizes)
 		std::vector<uint8_t> file_id;
-
-		// TODO: max supported lossy packet size
 	};
 
 	struct NGCEXT_ft1_init_ack {
@@ -81,7 +79,8 @@ namespace Events {
 		// - 1 byte (transfer_id)
 		uint8_t transfer_id;
 
-		// TODO: max supported lossy packet size
+		// - 2 byte (self_max_lossy_data_size)
+		uint16_t max_lossy_data_size;
 	};
 
 	struct NGCEXT_ft1_data {
@@ -163,6 +162,7 @@ enum class NGCEXT_Event : uint8_t {
 	// acknowlage init (like an accept)
 	// like tox ft control continue
 	// - 1 byte (transfer_id)
+	// - 2 byte (self_max_lossy_data_size) (optional since v2)
 	FT1_INIT_ACK,
 
 	// TODO: init deny, speed up non acceptance
@@ -263,6 +263,12 @@ class NGCEXTEventProvider : public ToxEventI, public NGCEXTEventProviderI {
 			bool _private
 		);
 
+		bool parse_ft1_init_ack_v2(
+			uint32_t group_number, uint32_t peer_number,
+			const uint8_t* data, size_t data_size,
+			bool _private
+		);
+
 		bool handlePacket(
 			const uint32_t group_number,
 			const uint32_t peer_number,

solanaceae/ngc_ft1/cca.hpp

@@ -38,22 +38,34 @@ struct CCAI {
 		//static_assert(maximum_segment_size == 574); // mesured in wireshark
 
 		// flow control
-		float max_byterate_allowed {10*1024*1024}; // 10MiB/s
+		//float max_byterate_allowed {100.f*1024*1024}; // 100MiB/s
+		float max_byterate_allowed {10.f*1024*1024}; // 10MiB/s
+		//float max_byterate_allowed {1.f*1024*1024}; // 1MiB/s
+		//float max_byterate_allowed {0.6f*1024*1024}; // 600KiB/s
+		//float max_byterate_allowed {0.5f*1024*1024}; // 500KiB/s
+		//float max_byterate_allowed {0.15f*1024*1024}; // 150KiB/s
+		//float max_byterate_allowed {0.05f*1024*1024}; // 50KiB/s
 
 	public: // api
 		CCAI(size_t maximum_segment_data_size) : MAXIMUM_SEGMENT_DATA_SIZE(maximum_segment_data_size) {}
+		virtual ~CCAI(void) {}
+
+		// returns current rtt/delay
+		virtual float getCurrentDelay(void) const = 0;
 
 		// return the current believed window in bytes of how much data can be inflight,
-		//virtual float getCWnD(void) const = 0;
+		virtual float getWindow(void) = 0;
 
 		// TODO: api for how much data we should send
 		// take time since last sent into account
 		// respect max_byterate_allowed
-		virtual size_t canSend(void) = 0;
+		virtual int64_t canSend(float time_delta) = 0;
 
 		// get the list of timed out seq_ids
 		virtual std::vector<SeqIDType> getTimeouts(void) const = 0;
 
+		virtual int64_t inFlightCount(void) const { return -1; }
+
 	public: // callbacks
 		// data size is without overhead
 		virtual void onSent(SeqIDType seq, size_t data_size) = 0;

solanaceae/ngc_ft1/cubic.cpp

@@ -3,14 +3,25 @@
 #include <cmath>
 #include <iostream>
 
+void CUBIC::updateReductionTimer(float time_delta) {
+	const auto now {getTimeNow()};
+
+	// only keep updating while the cca interaction is not too long ago
+	if (now - _time_point_last_update <= getCurrentDelay()*4.f) {
+		_time_since_reduction += time_delta;
+	}
+}
+
+void CUBIC::resetReductionTimer(void) {
+	_time_since_reduction = 0.f;
+}
+
 float CUBIC::getCWnD(void) const {
 	const double K = cbrt(
 		(_window_max * (1. - BETA)) / SCALING_CONSTANT
 	);
 
-	const double time_since_reduction = getTimeNow() - _time_point_reduction;
+	const double TK = _time_since_reduction - K;
-
-	const double TK = time_since_reduction - K;
 
 	const double cwnd =
 		SCALING_CONSTANT
@@ -33,29 +44,70 @@ float CUBIC::getCWnD(void) const {
 }
 
 void CUBIC::onCongestion(void) {
-	if (getTimeNow() - _time_point_reduction >= getCurrentDelay()) {
+	// 8 is probably too much (800ms for 100ms rtt)
-		const auto current_cwnd = getCWnD();
+	if (_time_since_reduction >= getCurrentDelay()*4.f) {
-		_time_point_reduction = getTimeNow();
+		const auto tmp_old_tp = _time_since_reduction;
-		_window_max = current_cwnd;
 
-		std::cout << "CONGESTION! cwnd:" << current_cwnd << "\n";
+		const auto current_cwnd = getCWnD(); // TODO: remove, only used by logging?
+		const auto current_wnd = getWindow(); // respects cwnd and fwnd
+
+		_bytes_leftover = 0;
+		resetReductionTimer();
+
+		if (current_cwnd < _window_max) {
+			// congestion before reaching the inflection point (prev window_max).
+			// reduce to wnd*beta to be fair
+			_window_max = current_wnd * BETA;
+		} else {
+			_window_max = current_wnd;
+		}
+
+		_window_max = std::max(_window_max, 2.0*MAXIMUM_SEGMENT_SIZE);
+
+#if 1
+		std::cout << "----CONGESTION!"
+			<< " cwnd:" << current_cwnd
+			<< " wnd:" << current_wnd
+			<< " cwnd_max:" << _window_max
+			<< " pts:" << tmp_old_tp
+			<< " rtt:" << getCurrentDelay()
+			<< "\n"
+		;
+#endif
 	}
 }
 
-size_t CUBIC::canSend(void) {
+float CUBIC::getWindow(void) {
-	const auto fspace_pkgs = FlowOnly::canSend();
+	return std::min<float>(getCWnD(), FlowOnly::getWindow());
+}
+
+int64_t CUBIC::canSend(float time_delta) {
+	const auto fspace_pkgs = FlowOnly::canSend(time_delta);
+
+	updateReductionTimer(time_delta);
 
 	if (fspace_pkgs == 0u) {
 		return 0u;
 	}
 
-	const int64_t cspace_bytes = getCWnD() - _in_flight_bytes;
+	const auto window = getCWnD();
+	int64_t cspace_bytes = (window - _in_flight_bytes) + _bytes_leftover;
 	if (cspace_bytes < MAXIMUM_SEGMENT_DATA_SIZE) {
 		return 0u;
 	}
 
+	// also limit to max sendrate per tick, which is usually smaller than window
+	// this is mostly to prevent spikes on empty windows
+	const auto rate = window / getCurrentDelay();
+
+	// we dont want this limit to fall below atleast 1 segment
+	const int64_t max_bytes_per_tick = std::max<int64_t>(rate * time_delta + 0.5f, MAXIMUM_SEGMENT_SIZE);
+	cspace_bytes = std::min<int64_t>(cspace_bytes, max_bytes_per_tick);
+
 	// limit to whole packets
-	size_t cspace_pkgs = std::floor(cspace_bytes / MAXIMUM_SEGMENT_DATA_SIZE) * MAXIMUM_SEGMENT_DATA_SIZE;
+	int64_t cspace_pkgs = (cspace_bytes / MAXIMUM_SEGMENT_DATA_SIZE) * MAXIMUM_SEGMENT_DATA_SIZE;
+
+	_bytes_leftover = cspace_bytes - cspace_pkgs;
 
 	return std::min(cspace_pkgs, fspace_pkgs);
 }

solanaceae/ngc_ft1/cubic.hpp

@@ -8,7 +8,8 @@ struct CUBIC : public FlowOnly {
 	//using clock = std::chrono::steady_clock;
 
 	public: // config
-		static constexpr float BETA {0.7f};
+		//static constexpr float BETA {0.7f};
+		static constexpr float BETA {0.8f};
 		static constexpr float SCALING_CONSTANT {0.4f};
 		static constexpr float RTT_EMA_ALPHA = 0.1f; // 0.1 is very smooth, might need more
 
@@ -16,9 +17,14 @@ struct CUBIC : public FlowOnly {
 		// window size before last reduciton
 		double _window_max {2.f * MAXIMUM_SEGMENT_SIZE}; // start with mss*2
 		//double _window_last_max {2.f * MAXIMUM_SEGMENT_SIZE};
-		double _time_point_reduction {getTimeNow()};
+
+		double _time_since_reduction {12.f}; // warm start
+		int64_t _bytes_leftover {0};
 
 	private:
+		void updateReductionTimer(float time_delta);
+		void resetReductionTimer(void);
+
 		float getCWnD(void) const;
 
 		// moving avg over the last few delay samples
@@ -31,11 +37,14 @@ struct CUBIC : public FlowOnly {
 
 	public: // api
 		CUBIC(size_t maximum_segment_data_size) : FlowOnly(maximum_segment_data_size) {}
+		virtual ~CUBIC(void) {}
+
+		float getWindow(void) override;
 
 		// TODO: api for how much data we should send
 		// take time since last sent into account
 		// respect max_byterate_allowed
-		size_t canSend(void) override;
+		int64_t canSend(float time_delta) override;
 
 		// get the list of timed out seq_ids
 		//std::vector<SeqIDType> getTimeouts(void) const override;

solanaceae/ngc_ft1/flow_only.cpp

@@ -6,10 +6,16 @@
 #include <algorithm>
 
 float FlowOnly::getCurrentDelay(void) const {
-	return std::min(_rtt_ema, RTT_MAX);
+	// below 1ms is useless
+	return std::clamp(_rtt_ema, 0.001f, RTT_MAX);
 }
 
 void FlowOnly::addRTT(float new_delay) {
+	if (new_delay > _rtt_ema * RTT_UP_MAX) {
+		// too large a jump up, to be taken into account
+		return;
+	}
+
 	// lerp(new_delay, rtt_ema, 0.1)
 	_rtt_ema = RTT_EMA_ALPHA * new_delay + (1.f - RTT_EMA_ALPHA) * _rtt_ema;
 }
@@ -23,7 +29,40 @@ void FlowOnly::updateWindow(void) {
 	_fwnd = std::max(_fwnd, 2.f * MAXIMUM_SEGMENT_DATA_SIZE);
 }
 
-size_t FlowOnly::canSend(void) {
+void FlowOnly::updateCongestion(void) {
+	const auto tmp_window = getWindow();
+	// packet window * 0.3
+	// but atleast 4
+	int32_t max_consecutive_events = std::clamp<int32_t>(
+		(tmp_window/MAXIMUM_SEGMENT_DATA_SIZE) * 0.3f,
+		4,
+		50 // limit TODO: fix idle/time starved algo
+	);
+	// TODO: magic number
+
+#if 0
+	std::cout << "NGC_FT1 Flow: pkg out of order"
+		<< " w:" << tmp_window
+		<< " pw:" << tmp_window/MAXIMUM_SEGMENT_DATA_SIZE
+		<< " coe:" << _consecutive_events
+		<< " mcoe:" << max_consecutive_events
+		<< "\n";
+#endif
+
+	if (_consecutive_events > max_consecutive_events) {
+		//std::cout << "CONGESTION! NGC_FT1 flow: pkg out of order\n";
+		onCongestion();
+
+		// TODO: set _consecutive_events to zero?
+	}
+}
+
+float FlowOnly::getWindow(void) {
+	updateWindow();
+	return _fwnd;
+}
+
+int64_t FlowOnly::canSend(float time_delta) {
 	if (_in_flight.empty()) {
 		assert(_in_flight_bytes == 0);
 		return MAXIMUM_SEGMENT_DATA_SIZE;
@@ -31,16 +70,17 @@ size_t FlowOnly::canSend(void) {
 
 	updateWindow();
 
-	const int64_t fspace = _fwnd - _in_flight_bytes;
+	int64_t fspace = _fwnd - _in_flight_bytes;
 	if (fspace < MAXIMUM_SEGMENT_DATA_SIZE) {
 		return 0u;
 	}
 
-	// limit to whole packets
+	// also limit to max sendrate per tick, which is usually smaller than window
-	size_t space = std::floor(fspace / MAXIMUM_SEGMENT_DATA_SIZE)
+	// this is mostly to prevent spikes on empty windows
-		* MAXIMUM_SEGMENT_DATA_SIZE;
+	fspace = std::min<int64_t>(fspace, max_byterate_allowed * time_delta + 0.5f);
 
-	return space;
+	// limit to whole packets
+	return (fspace / MAXIMUM_SEGMENT_DATA_SIZE) * MAXIMUM_SEGMENT_DATA_SIZE;
 }
 
 std::vector<FlowOnly::SeqIDType> FlowOnly::getTimeouts(void) const {
@@ -49,7 +89,7 @@ std::vector<FlowOnly::SeqIDType> FlowOnly::getTimeouts(void) const {
 	// after 3 rtt delay, we trigger timeout
 	const auto now_adjusted = getTimeNow() - getCurrentDelay()*3.f;
 
-	for (const auto& [seq, time_stamp, size] : _in_flight) {
+	for (const auto& [seq, time_stamp, size, _] : _in_flight) {
 		if (now_adjusted > time_stamp) {
 			list.push_back(seq);
 		}
@@ -58,16 +98,28 @@ std::vector<FlowOnly::SeqIDType> FlowOnly::getTimeouts(void) const {
 	return list;
 }
 
+int64_t FlowOnly::inFlightCount(void) const {
+	return _in_flight.size();
+}
+
 void FlowOnly::onSent(SeqIDType seq, size_t data_size) {
 	if constexpr (true) {
 		for (const auto& it : _in_flight) {
-			assert(std::get<0>(it) != seq);
+			assert(it.id != seq);
 		}
 	}
 
-	_in_flight.push_back({seq, getTimeNow(), data_size + SEGMENT_OVERHEAD});
+	const auto& new_entry = _in_flight.emplace_back(
-	_in_flight_bytes += data_size + SEGMENT_OVERHEAD;
+		FlyingBunch{
-	//_recently_sent_bytes += data_size + SEGMENT_OVERHEAD;
+			seq,
+			static_cast<float>(getTimeNow()),
+			data_size + SEGMENT_OVERHEAD,
+			false
+		}
+	);
+	_in_flight_bytes += new_entry.bytes;
+
+	_time_point_last_update = getTimeNow();
 }
 
 void FlowOnly::onAck(std::vector<SeqIDType> seqs) {
@@ -78,28 +130,31 @@ void FlowOnly::onAck(std::vector<SeqIDType> seqs) {
 
 	const auto now {getTimeNow()};
 
+	_time_point_last_update = now;
+
 	// first seq in seqs is the actual value, all extra are for redundency
 	{ // skip in ack is congestion event
 		// 1. look at primary ack of packet
 		auto it = std::find_if(_in_flight.begin(), _in_flight.end(), [seq = seqs.front()](const auto& v) -> bool {
-			return std::get<0>(v) == seq;
+			return v.id == seq;
 		});
-		if (it != _in_flight.end()) {
+		if (it != _in_flight.end() && !it->ignore) {
-			if (it != _in_flight.begin()) {
+			// find first non ignore, it should be the expected
+			auto first_it = std::find_if_not(_in_flight.cbegin(), _in_flight.cend(), [](const auto& v) -> bool { return v.ignore; });
+
+			if (first_it != _in_flight.cend() && it != first_it) {
 				// not next expected seq -> skip detected
 
-				std::cout << "CONGESTION out of order\n";
+				_consecutive_events++;
-				onCongestion();
+				it->ignore = true; // only handle once
-				//if (getTimeNow() >= _last_congestion_event + _last_congestion_rtt) {
+
-					//_recently_lost_data = true;
+				updateCongestion();
-					//_last_congestion_event = getTimeNow();
-					//_last_congestion_rtt = getCurrentDelay();
-				//}
 			} else {
 				// only mesure delay, if not a congestion
-				addRTT(now - std::get<1>(*it));
+				addRTT(now - it->timestamp);
+				_consecutive_events = 0;
 			}
-		} else {
+		} else { // TOOD: if ! ignore too
 			// !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
 #if 0
 			// assume we got a duplicated packet
@@ -111,14 +166,14 @@ void FlowOnly::onAck(std::vector<SeqIDType> seqs) {
 
 	for (const auto& seq : seqs) {
 		auto it = std::find_if(_in_flight.begin(), _in_flight.end(), [seq](const auto& v) -> bool {
-			return std::get<0>(v) == seq;
+			return v.id == seq;
 		});
 
 		if (it == _in_flight.end()) {
 			continue; // not found, ignore
 		} else {
 			//most_recent = std::max(most_recent, std::get<1>(*it));
-			_in_flight_bytes -= std::get<2>(*it);
+			_in_flight_bytes -= it->bytes;
 			assert(_in_flight_bytes >= 0);
 			//_recently_acked_data += std::get<2>(*it);
 			_in_flight.erase(it);
@@ -128,8 +183,8 @@ void FlowOnly::onAck(std::vector<SeqIDType> seqs) {
 
 void FlowOnly::onLoss(SeqIDType seq, bool discard) {
 	auto it = std::find_if(_in_flight.begin(), _in_flight.end(), [seq](const auto& v) -> bool {
-		assert(!std::isnan(std::get<1>(v)));
+		assert(!std::isnan(v.timestamp));
-		return std::get<0>(v) == seq;
+		return v.id == seq;
 	});
 
 	if (it == _in_flight.end()) {
@@ -137,24 +192,27 @@ void FlowOnly::onLoss(SeqIDType seq, bool discard) {
 		return; // not found, ignore ??
 	}
 
-	std::cerr << "FLOW loss\n";
+	//std::cerr << "FLOW loss\n";
 
 	// "if data lost is not to be retransmitted"
 	if (discard) {
-		_in_flight_bytes -= std::get<2>(*it);
+		_in_flight_bytes -= it->bytes;
 		assert(_in_flight_bytes >= 0);
 		_in_flight.erase(it);
+	} else {
+		// and not take into rtt
+		it->timestamp = getTimeNow();
+		it->ignore = true;
 	}
-	// TODO: reset timestamp?
 
-#if 0 // temporarily disable ce for timeout
+	// usually after data arrived out-of-order/duplicate
-	// at most once per rtt?
+	if (!it->ignore) {
-	// TODO: use delay at event instead
+		it->ignore = true; // only handle once
-	if (getTimeNow() >= _last_congestion_event + _last_congestion_rtt) {
+		//_consecutive_events++;
-		_recently_lost_data = true;
+
-		_last_congestion_event = getTimeNow();
+		//updateCongestion();
-		_last_congestion_rtt = getCurrentDelay();
+		// this is usually a safe indicator for congestion/maxed connection
+		onCongestion();
 	}
-#endif
 }
 

solanaceae/ngc_ft1/flow_only.hpp

@@ -11,17 +11,10 @@ struct FlowOnly : public CCAI {
 		using clock = std::chrono::steady_clock;
 
 	public: // config
-		static constexpr float RTT_EMA_ALPHA = 0.1f; // might need over time
+		static constexpr float RTT_EMA_ALPHA = 0.001f; // might need change over time
+		static constexpr float RTT_UP_MAX = 3.0f; // how much larger a delay can be to be taken into account
 		static constexpr float RTT_MAX = 2.f; // 2 sec is probably too much
 
-		//float max_byterate_allowed {100.f*1024*1024}; // 100MiB/s
-		float max_byterate_allowed {10.f*1024*1024}; // 10MiB/s
-		//float max_byterate_allowed {1.f*1024*1024}; // 1MiB/s
-		//float max_byterate_allowed {0.6f*1024*1024}; // 600KiB/s
-		//float max_byterate_allowed {0.5f*1024*1024}; // 500KiB/s
-		//float max_byterate_allowed {0.05f*1024*1024}; // 50KiB/s
-		//float max_byterate_allowed {0.15f*1024*1024}; // 150KiB/s
-
 	protected:
 		// initialize to low value, will get corrected very fast
 		float _fwnd {0.01f * max_byterate_allowed}; // in bytes
@@ -30,11 +23,24 @@ struct FlowOnly : public CCAI {
 		float _rtt_ema {0.1f};
 
 		// list of sequence ids and timestamps of when they where sent (and payload size)
-		std::vector<std::tuple<SeqIDType, float, size_t>> _in_flight;
+		struct FlyingBunch {
+			SeqIDType id;
+			float timestamp;
+			size_t bytes;
+
+			// set to true if counted as ce or resent due to timeout
+			bool ignore {false};
+		};
+		std::vector<FlyingBunch> _in_flight;
 		int64_t _in_flight_bytes {0};
 
+		int32_t _consecutive_events {0};
+
 		clock::time_point _time_start_offset;
 
+		// used to clamp growth rate in the void
+		double _time_point_last_update {getTimeNow()};
+
 	protected:
 		// make values relative to algo start for readability (and precision)
 		// get timestamp in seconds
@@ -44,7 +50,9 @@ struct FlowOnly : public CCAI {
 
 		// moving avg over the last few delay samples
 		// VERY sensitive to bundling acks
-		float getCurrentDelay(void) const;
+		float getCurrentDelay(void) const override;
+
+		float getWindow(void) override;
 
 		void addRTT(float new_delay);
 
@@ -52,17 +60,23 @@ struct FlowOnly : public CCAI {
 
 		virtual void onCongestion(void) {};
 
+		// internal logic, calls the onCongestion() event
+		void updateCongestion(void);
+
 	public: // api
 		FlowOnly(size_t maximum_segment_data_size) : CCAI(maximum_segment_data_size) {}
+		virtual ~FlowOnly(void) {}
 
 		// TODO: api for how much data we should send
 		// take time since last sent into account
 		// respect max_byterate_allowed
-		size_t canSend(void) override;
+		int64_t canSend(float time_delta) override;
 
 		// get the list of timed out seq_ids
 		std::vector<SeqIDType> getTimeouts(void) const override;
 
+		int64_t inFlightCount(void) const override;
+
 	public: // callbacks
 		// data size is without overhead
 		void onSent(SeqIDType seq, size_t data_size) override;

solanaceae/ngc_ft1/ledbat.cpp

@@ -6,6 +6,7 @@
 #include <deque>
 #include <cstdint>
 #include <cassert>
+#include <tuple>
 
 #include <iomanip>
 #include <iostream>
@@ -19,7 +20,7 @@ LEDBAT::LEDBAT(size_t maximum_segment_data_size) : CCAI(maximum_segment_data_siz
 	_time_start_offset = clock::now();
 }
 
-size_t LEDBAT::canSend(void) {
+int64_t LEDBAT::canSend(float time_delta) {
 	if (_in_flight.empty()) {
 		return MAXIMUM_SEGMENT_DATA_SIZE;
 	}
@@ -34,9 +35,7 @@ size_t LEDBAT::canSend(void) {
 		return 0u;
 	}
 
-	size_t space = std::ceil(std::min<float>(cspace, fspace) / MAXIMUM_SEGMENT_DATA_SIZE) * MAXIMUM_SEGMENT_DATA_SIZE;
+	return std::ceil(std::min<float>(cspace, fspace) / MAXIMUM_SEGMENT_DATA_SIZE) * MAXIMUM_SEGMENT_DATA_SIZE;
-
-	return space;
 }
 
 std::vector<LEDBAT::SeqIDType> LEDBAT::getTimeouts(void) const {

solanaceae/ngc_ft1/ledbat.hpp

@@ -11,7 +11,7 @@
 // LEDBAT++: https://www.ietf.org/archive/id/draft-irtf-iccrg-ledbat-plus-plus-01.txt
 
 // LEDBAT++ implementation
-struct LEDBAT : public CCAI{
+struct LEDBAT : public CCAI {
 	public: // config
 #if 0
 		using SeqIDType = std::pair<uint8_t, uint16_t>; // tf_id, seq_id
@@ -47,21 +47,20 @@ struct LEDBAT : public CCAI{
 
 		//static constexpr size_t rtt_buffer_size_max {2000};
 
-		float max_byterate_allowed {10*1024*1024}; // 10MiB/s
-
 	public:
 		LEDBAT(size_t maximum_segment_data_size);
+		virtual ~LEDBAT(void) {}
 
 		// return the current believed window in bytes of how much data can be inflight,
 		// without overstepping the delay requirement
-		float getCWnD(void) const {
+		float getWindow(void) override {
 			return _cwnd;
 		}
 
 		// TODO: api for how much data we should send
 		// take time since last sent into account
 		// respect max_byterate_allowed
-		size_t canSend(void) override;
+		int64_t canSend(float time_delta) override;
 
 		// get the list of timed out seq_ids
 		std::vector<SeqIDType> getTimeouts(void) const override;
@@ -86,7 +85,7 @@ struct LEDBAT : public CCAI{
 
 		// moving avg over the last few delay samples
 		// VERY sensitive to bundling acks
-		float getCurrentDelay(void) const;
+		float getCurrentDelay(void) const override;
 
 		void addRTT(float new_delay);
 

solanaceae/ngc_ft1/ngcft1.cpp

@@ -1,9 +1,14 @@
 #include "./ngcft1.hpp"
 
-#include <solanaceae/toxcore/utils.hpp>
+#include "./flow_only.hpp"
+#include "./cubic.hpp"
+#include "./ledbat.hpp"
+
+#include <solanaceae/util/utils.hpp>
 
 #include <sodium.h>
 
+#include <cstdint>
 #include <iostream>
 #include <set>
 #include <algorithm>
@@ -71,6 +76,12 @@ bool NGCFT1::sendPKG_FT1_INIT_ACK(
 	pkg.push_back(static_cast<uint8_t>(NGCEXT_Event::FT1_INIT_ACK));
 	pkg.push_back(transfer_id);
 
+	// - 2 bytes max_lossy_data_size
+	const uint16_t max_lossy_data_size = _t.toxGroupMaxCustomLossyPacketLength() - 4;
+	for (size_t i = 0; i < sizeof(uint16_t); i++) {
+		pkg.push_back((max_lossy_data_size>>(i*8)) & 0xff);
+	}
+
 	// lossless
 	return _t.toxGroupSendCustomPrivatePacket(group_number, peer_number, true, pkg) == TOX_ERR_GROUP_SEND_CUSTOM_PRIVATE_PACKET_OK;
 }
@@ -87,6 +98,7 @@ bool NGCFT1::sendPKG_FT1_DATA(
 	// check header_size+data_size <= max pkg size
 
 	std::vector<uint8_t> pkg;
+	pkg.reserve(2048); // saves a ton of allocations
 	pkg.push_back(static_cast<uint8_t>(NGCEXT_Event::FT1_DATA));
 	pkg.push_back(transfer_id);
 	pkg.push_back(sequence_id & 0xff);
@@ -107,6 +119,7 @@ bool NGCFT1::sendPKG_FT1_DATA_ACK(
 	const uint16_t* seq_ids, size_t seq_ids_size
 ) {
 	std::vector<uint8_t> pkg;
+	pkg.reserve(1+1+2*32); // 32acks in a single pkg should be unlikely
 	pkg.push_back(static_cast<uint8_t>(NGCEXT_Event::FT1_DATA_ACK));
 	pkg.push_back(transfer_id);
 
@@ -143,7 +156,7 @@ bool NGCFT1::sendPKG_FT1_MESSAGE(
 	return _t.toxGroupSendCustomPacket(group_number, true, pkg) == TOX_ERR_GROUP_SEND_CUSTOM_PACKET_OK;
 }
 
-void NGCFT1::updateSendTransfer(float time_delta, uint32_t group_number, uint32_t peer_number, Group::Peer& peer, size_t idx, std::set<CCAI::SeqIDType>& timeouts_set) {
+void NGCFT1::updateSendTransfer(float time_delta, uint32_t group_number, uint32_t peer_number, Group::Peer& peer, size_t idx, std::set<CCAI::SeqIDType>& timeouts_set, int64_t& can_packet_size) {
 	auto& tf_opt = peer.send_transfers.at(idx);
 	assert(tf_opt.has_value());
 	auto& tf = tf_opt.value();
@@ -175,22 +188,36 @@ void NGCFT1::updateSendTransfer(float time_delta, uint32_t group_number, uint32_
 			}
 			//break;
 			return;
-		case State::SENDING: {
+		case State::FINISHING: // we still have unacked packets
+			tf.ssb.for_each(time_delta, [&](uint16_t id, const std::vector<uint8_t>& data, float& time_since_activity) {
+				if (can_packet_size >= data.size() && timeouts_set.count({idx, id})) {
+					sendPKG_FT1_DATA(group_number, peer_number, idx, id, data.data(), data.size());
+					peer.cca->onLoss({idx, id}, false);
+					time_since_activity = 0.f;
+					timeouts_set.erase({idx, id});
+					can_packet_size -= data.size();
+				}
+			});
+			if (tf.time_since_activity >= sending_give_up_after) {
+				// no ack after 30sec, close ft
+				// TODO: notify app
+				std::cerr << "NGCFT1 warning: sending ft finishing timed out, deleting\n";
+
+				// clean up cca
 				tf.ssb.for_each(time_delta, [&](uint16_t id, const std::vector<uint8_t>& data, float& time_since_activity) {
-					// no ack after 5 sec -> resend
+					peer.cca->onLoss({idx, id}, true);
-					//if (time_since_activity >= ngc_ft1_ctx->options.sending_resend_without_ack_after) {
+					timeouts_set.erase({idx, id});
-					if (timeouts_set.count({idx, id})) {
-						// TODO: can fail
-						sendPKG_FT1_DATA(group_number, peer_number, idx, id, data.data(), data.size());
-						peer.cca->onLoss({idx, id}, false);
-						time_since_activity = 0.f;
-						timeouts_set.erase({idx, id});
-					}
 				});
 
-				if (tf.time_since_activity >= sending_give_up_after) {
+				tf_opt.reset();
+			}
+			break;
+		case State::SENDING: {
+				// first handle overall timeout (could otherwise do resends directly before, which is useless)
+				// timeout increases with active transfers (otherwise we could starve them)
+				if (tf.time_since_activity >= (sending_give_up_after * peer.active_send_transfers)) {
 					// no ack after 30sec, close ft
-					std::cerr << "NGCFT1 warning: sending ft in progress timed out, deleting\n";
+					std::cerr << "NGCFT1 warning: sending ft in progress timed out, deleting (ifc:" << peer.cca->inFlightCount() << ")\n";
 					dispatch(
 						NGCFT1_Event::send_done,
 						Events::NGCFT1_send_done{
@@ -210,22 +237,23 @@ void NGCFT1::updateSendTransfer(float time_delta, uint32_t group_number, uint32_
 					return;
 				}
 
+				// do resends
+				tf.ssb.for_each(time_delta, [&](uint16_t id, const std::vector<uint8_t>& data, float& time_since_activity) {
+					if (can_packet_size >= data.size() && time_since_activity >= peer.cca->getCurrentDelay() && timeouts_set.count({idx, id})) {
+						// TODO: can fail
+						sendPKG_FT1_DATA(group_number, peer_number, idx, id, data.data(), data.size());
+						peer.cca->onLoss({idx, id}, false);
+						time_since_activity = 0.f;
+						timeouts_set.erase({idx, id});
+						can_packet_size -= data.size();
+					}
+				});
+
 				// if chunks in flight < window size (2)
-				//while (tf.ssb.size() < ngc_ft1_ctx->options.packet_window_size) {
-				int64_t can_packet_size {static_cast<int64_t>(peer.cca->canSend())};
-				//if (can_packet_size) {
-					//std::cerr << "FT: can_packet_size: " << can_packet_size;
-				//}
-				size_t count {0};
 				while (can_packet_size > 0 && tf.file_size > 0) {
 					std::vector<uint8_t> new_data;
 
-					// TODO: parameterize packet size? -> only if JF increases lossy packet size >:)
-					//size_t chunk_size = std::min<size_t>(496u, tf.file_size - tf.file_size_current);
-					//size_t chunk_size = std::min<size_t>(can_packet_size, tf.file_size - tf.file_size_current);
 					size_t chunk_size = std::min<size_t>({
-						//496u,
-						//996u,
 						peer.cca->MAXIMUM_SEGMENT_DATA_SIZE,
 						static_cast<size_t>(can_packet_size),
 						tf.file_size - tf.file_size_current
@@ -237,14 +265,6 @@ void NGCFT1::updateSendTransfer(float time_delta, uint32_t group_number, uint32_
 
 					new_data.resize(chunk_size);
 
-					//ngc_ft1_ctx->cb_send_data[tf.file_kind](
-						//tox,
-						//group_number, peer_number,
-						//idx,
-						//tf.file_size_current,
-						//new_data.data(), new_data.size(),
-						//ngc_ft1_ctx->ud_send_data.count(tf.file_kind) ? ngc_ft1_ctx->ud_send_data.at(tf.file_kind) : nullptr
-					//);
 					assert(idx <= 0xffu);
 					// TODO: check return value
 					dispatch(
@@ -258,45 +278,19 @@ void NGCFT1::updateSendTransfer(float time_delta, uint32_t group_number, uint32_
 					);
 
 					uint16_t seq_id = tf.ssb.add(std::move(new_data));
-					sendPKG_FT1_DATA(group_number, peer_number, idx, seq_id, tf.ssb.entries.at(seq_id).data.data(), tf.ssb.entries.at(seq_id).data.size());
+					const bool sent = sendPKG_FT1_DATA(group_number, peer_number, idx, seq_id, tf.ssb.entries.at(seq_id).data.data(), tf.ssb.entries.at(seq_id).data.size());
-					peer.cca->onSent({idx, seq_id}, chunk_size);
+					if (sent) {
-
+						peer.cca->onSent({idx, seq_id}, chunk_size);
-#if defined(EXTRA_LOGGING) && EXTRA_LOGGING == 1
+					} else {
-					fprintf(stderr, "FT: sent data size: %ld (seq %d)\n", chunk_size, seq_id);
+						std::cerr << "NGCFT1: failed to send packet (queue full?) --------------\n";
-#endif
+						peer.cca->onLoss({idx, seq_id}, false); // HACK: fake congestion event
+						// TODO: onCongestion
+						can_packet_size = 0;
+					}
 
 					tf.file_size_current += chunk_size;
 					can_packet_size -= chunk_size;
-					count++;
 				}
-				//if (count) {
-					//std::cerr << " split over " << count << "\n";
-				//}
-			}
-			break;
-		case State::FINISHING: // we still have unacked packets
-			tf.ssb.for_each(time_delta, [&](uint16_t id, const std::vector<uint8_t>& data, float& time_since_activity) {
-				// no ack after 5 sec -> resend
-				//if (time_since_activity >= ngc_ft1_ctx->options.sending_resend_without_ack_after) {
-				if (timeouts_set.count({idx, id})) {
-					sendPKG_FT1_DATA(group_number, peer_number, idx, id, data.data(), data.size());
-					peer.cca->onLoss({idx, id}, false);
-					time_since_activity = 0.f;
-					timeouts_set.erase({idx, id});
-				}
-			});
-			if (tf.time_since_activity >= sending_give_up_after) {
-				// no ack after 30sec, close ft
-				// TODO: notify app
-				std::cerr << "NGCFT1 warning: sending ft finishing timed out, deleting\n";
-
-				// clean up cca
-				tf.ssb.for_each(time_delta, [&](uint16_t id, const std::vector<uint8_t>& data, float& time_since_activity) {
-					peer.cca->onLoss({idx, id}, true);
-					timeouts_set.erase({idx, id});
-				});
-
-				tf_opt.reset();
 			}
 			break;
 		default: // invalid state, delete
@@ -308,12 +302,33 @@ void NGCFT1::updateSendTransfer(float time_delta, uint32_t group_number, uint32_
 }
 
 void NGCFT1::iteratePeer(float time_delta, uint32_t group_number, uint32_t peer_number, Group::Peer& peer) {
-	auto timeouts = peer.cca->getTimeouts();
+	if (peer.cca) {
-	std::set<CCAI::SeqIDType> timeouts_set{timeouts.cbegin(), timeouts.cend()};
+		auto timeouts = peer.cca->getTimeouts();
+		std::set<CCAI::SeqIDType> timeouts_set{timeouts.cbegin(), timeouts.cend()};
 
-	for (size_t idx = 0; idx < peer.send_transfers.size(); idx++) {
+		int64_t can_packet_size {peer.cca->canSend(time_delta)}; // might get more space while iterating (time)
-		if (peer.send_transfers.at(idx).has_value()) {
+
-			updateSendTransfer(time_delta, group_number, peer_number, peer, idx, timeouts_set);
+		// get number current running transfers TODO: improve
+		peer.active_send_transfers = 0;
+		for (const auto& it : peer.send_transfers) {
+			if (it.has_value()) {
+				peer.active_send_transfers++;
+			}
+		}
+
+		// change iterate start position to not starve transfers in the back
+		size_t iterated_count = 0;
+		bool last_send_found = false;
+		for (size_t idx = peer.next_send_transfer_send_idx; iterated_count < peer.send_transfers.size(); idx++, iterated_count++) {
+			idx = idx % peer.send_transfers.size();
+
+			if (peer.send_transfers.at(idx).has_value()) {
+				if (!last_send_found && can_packet_size <= 0) {
+					peer.next_send_transfer_send_idx = idx;
+					last_send_found = true; // only set once
+				}
+				updateSendTransfer(time_delta, group_number, peer_number, peer, idx, timeouts_set, can_packet_size);
+			}
 		}
 	}
 
@@ -333,15 +348,42 @@ NGCFT1::NGCFT1(
 	_neep.subscribe(this, NGCEXT_Event::FT1_DATA_ACK);
 	_neep.subscribe(this, NGCEXT_Event::FT1_MESSAGE);
 
-	_tep.subscribe(this, Tox_Event::TOX_EVENT_GROUP_PEER_EXIT);
+	_tep.subscribe(this, Tox_Event_Type::TOX_EVENT_GROUP_PEER_EXIT);
 }
 
-void NGCFT1::iterate(float time_delta) {
+float NGCFT1::iterate(float time_delta) {
+	bool transfer_in_progress {false};
 	for (auto& [group_number, group] : groups) {
 		for (auto& [peer_number, peer] : group.peers) {
 			iteratePeer(time_delta, group_number, peer_number, peer);
+
+			// find any active transfer
+			if (!transfer_in_progress) {
+				for (const auto& t : peer.send_transfers) {
+					if (t.has_value()) {
+						transfer_in_progress = true;
+						break;
+					}
+				}
+			}
+			if (!transfer_in_progress) {
+				for (const auto& t : peer.recv_transfers) {
+					if (t.has_value()) {
+						transfer_in_progress = true;
+						break;
+					}
+				}
+			}
 		}
 	}
+
+	if (transfer_in_progress) {
+		// ~15ms for up to 1mb/s
+		// ~5ms for up to 4mb/s
+		return 0.005f; // 5ms
+	} else {
+		return 1.f; // once a sec might be too little
+	}
 }
 
 void NGCFT1::NGC_FT1_send_request_private(
@@ -487,7 +529,7 @@ bool NGCFT1::onEvent(const Events::NGCEXT_ft1_init& e) {
 
 bool NGCFT1::onEvent(const Events::NGCEXT_ft1_init_ack& e) {
 //#if !NDEBUG
-	std::cout << "NGCFT1: FT1_INIT_ACK\n";
+	std::cout << "NGCFT1: FT1_INIT_ACK mds:" << e.max_lossy_data_size << "\n";
 //#endif
 
 	// we now should start sending data
@@ -507,10 +549,28 @@ bool NGCFT1::onEvent(const Events::NGCEXT_ft1_init_ack& e) {
 
 	using State = Group::Peer::SendTransfer::State;
 	if (transfer.state != State::INIT_SENT) {
-		std::cerr << "NGCFT1 error: inti_ack but not in INIT_SENT state\n";
+		std::cerr << "NGCFT1 error: init_ack but not in INIT_SENT state\n";
 		return true;
 	}
 
+	// negotiated packet_data_size
+	const auto negotiated_packet_data_size = std::min<uint32_t>(e.max_lossy_data_size, _t.toxGroupMaxCustomLossyPacketLength()-4);
+	// TODO: reset cca with new pkg size
+	if (!peer.cca) {
+		// make random max of [1020-1220]
+		const uint32_t random_max_data_size = (1024-4) + _rng()%201;
+		const uint32_t randomized_negotiated_packet_data_size = std::min(negotiated_packet_data_size, random_max_data_size);
+
+		peer.max_packet_data_size = randomized_negotiated_packet_data_size;
+
+		std::cerr << "NGCFT1: creating cca with max:" << peer.max_packet_data_size << "\n";
+
+		peer.cca = std::make_unique<CUBIC>(peer.max_packet_data_size);
+		//peer.cca = std::make_unique<LEDBAT>(peer.max_packet_data_size);
+		//peer.cca = std::make_unique<FlowOnly>(peer.max_packet_data_size);
+		//peer.cca->max_byterate_allowed = 1.f *1024*1024;
+	}
+
 	// iterate will now call NGC_FT1_send_data_cb
 	transfer.state = State::SENDING;
 	transfer.time_since_activity = 0.f;
@@ -520,7 +580,7 @@ bool NGCFT1::onEvent(const Events::NGCEXT_ft1_init_ack& e) {
 
 bool NGCFT1::onEvent(const Events::NGCEXT_ft1_data& e) {
 #if !NDEBUG
-	std::cout << "NGCFT1: FT1_DATA\n";
+	//std::cout << "NGCFT1: FT1_DATA\n";
 #endif
 
 	if (e.data.empty()) {
@@ -599,7 +659,8 @@ bool NGCFT1::onEvent(const Events::NGCEXT_ft1_data_ack& e) {
 
 	Group::Peer& peer = groups[e.group_number].peers[e.peer_number];
 	if (!peer.send_transfers[e.transfer_id].has_value()) {
-		std::cerr << "NGCFT1 warning: data_ack for unknown transfer\n";
+		// we delete directly, packets might still be in flight (in practice they are when ce)
+		//std::cerr << "NGCFT1 warning: data_ack for unknown transfer\n";
 		return true;
 	}
 
@@ -625,7 +686,7 @@ bool NGCFT1::onEvent(const Events::NGCEXT_ft1_data_ack& e) {
 
 	// delete if all packets acked
 	if (transfer.file_size == transfer.file_size_current && transfer.ssb.size() == 0) {
-		std::cout << "NGCFT1: " << int(e.transfer_id) << " done\n";
+		std::cout << "NGCFT1: " << int(e.transfer_id) << " done. wnd:" << peer.cca->getWindow() << "\n";
 		dispatch(
 			NGCFT1_Event::send_done,
 			Events::NGCFT1_send_done{
@@ -711,7 +772,7 @@ bool NGCFT1::onToxEvent(const Tox_Event_Group_Peer_Exit* e) {
 	}
 
 	// reset cca
-	peer.cca = std::make_unique<CUBIC>(500-4); // TODO: replace with tox_group_max_custom_lossy_packet_length()-4
+	peer.cca.reset(); // dont actually reallocate
 
 	return false;
 }

solanaceae/ngc_ft1/ngcft1.hpp

@@ -2,22 +2,22 @@
 
 // solanaceae port of tox_ngc_ft1
 
-#include <solanaceae/toxcore/tox_interface.hpp>
 #include <solanaceae/toxcore/tox_event_interface.hpp>
+#include <solanaceae/toxcore/tox_interface.hpp>
 
 #include <solanaceae/ngc_ext/ngcext.hpp>
-#include "./cubic.hpp"
+#include "./cca.hpp"
-//#include "./flow_only.hpp"
-//#include "./ledbat.hpp"
 
 #include "./rcv_buf.hpp"
 #include "./snd_buf.hpp"
 
 #include "./ngcft1_file_kind.hpp"
 
+#include <cstdint>
 #include <map>
 #include <set>
 #include <memory>
+#include <random>
 
 namespace Events {
 
@@ -133,15 +133,19 @@ class NGCFT1 : public ToxEventI, public NGCEXTEventI, public NGCFT1EventProvider
 	ToxEventProviderI& _tep;
 	NGCEXTEventProviderI& _neep;
 
+	std::default_random_engine _rng{std::random_device{}()};
+
 	// TODO: config
 	size_t acks_per_packet {3u}; // 3
-	float init_retry_timeout_after {5.f}; // 10sec
+	float init_retry_timeout_after {4.f};
-	float sending_give_up_after {30.f}; // 30sec
+	float sending_give_up_after {15.f}; // 30sec (per active transfer)
 
 
 	struct Group {
 		struct Peer {
-			std::unique_ptr<CCAI> cca = std::make_unique<CUBIC>(500-4); // TODO: replace with tox_group_max_custom_lossy_packet_length()-4
+			uint32_t max_packet_data_size {500-4};
+			//std::unique_ptr<CCAI> cca = std::make_unique<CUBIC>(max_packet_data_size); // TODO: replace with tox_group_max_custom_lossy_packet_length()-4
+			std::unique_ptr<CCAI> cca;
 
 			struct RecvTransfer {
 				uint32_t file_kind;
@@ -188,6 +192,9 @@ class NGCFT1 : public ToxEventI, public NGCEXTEventI, public NGCFT1EventProvider
 			};
 			std::array<std::optional<SendTransfer>, 256> send_transfers;
 			size_t next_send_transfer_idx {0}; // next id will be 0
+			size_t next_send_transfer_send_idx {0};
+
+			size_t active_send_transfers {0};
 		};
 		std::map<uint32_t, Peer> peers;
 	};
@@ -201,7 +208,7 @@ class NGCFT1 : public ToxEventI, public NGCEXTEventI, public NGCFT1EventProvider
 		bool sendPKG_FT1_DATA_ACK(uint32_t group_number, uint32_t peer_number, uint8_t transfer_id, const uint16_t* seq_ids, size_t seq_ids_size);
 		bool sendPKG_FT1_MESSAGE(uint32_t group_number, uint32_t message_id, uint32_t file_kind, const uint8_t* file_id, size_t file_id_size);
 
-		void updateSendTransfer(float time_delta, uint32_t group_number, uint32_t peer_number, Group::Peer& peer, size_t idx, std::set<CCAI::SeqIDType>& timeouts_set);
+		void updateSendTransfer(float time_delta, uint32_t group_number, uint32_t peer_number, Group::Peer& peer, size_t idx, std::set<CCAI::SeqIDType>& timeouts_set, int64_t& can_packet_size);
 		void iteratePeer(float time_delta, uint32_t group_number, uint32_t peer_number, Group::Peer& peer);
 
 	public:
@@ -211,7 +218,7 @@ class NGCFT1 : public ToxEventI, public NGCEXTEventI, public NGCFT1EventProvider
 			NGCEXTEventProviderI& neep
 		);
 
-		void iterate(float delta);
+		float iterate(float delta);
 
 	public: // ft1 api
 		// TODO: public variant?

solanaceae/ngc_ft1_sha1/file_rw_mapped.hpp

@@ -1,58 +1,76 @@
 #pragma once
 
-#include <solanaceae/message3/file.hpp>
+#include <solanaceae/file/file2.hpp>
 
 #include "./mio.hpp"
 
 #include <filesystem>
 #include <fstream>
+#include <iostream>
 #include <cstring>
 
-struct FileRWMapped : public FileI {
+struct File2RWMapped : public File2I {
 	mio::ummap_sink _file_map;
 
 	// TODO: add truncate support?
-	FileRWMapped(std::string_view file_path, uint64_t file_size) {
+	// TODO: rw always true?
-		_file_size = file_size;
+	File2RWMapped(std::string_view file_path, int64_t file_size = -1) : File2I(true, true) {
+		std::filesystem::path native_file_path{file_path};
 
-		if (!std::filesystem::exists(file_path)) {
+		if (!std::filesystem::exists(native_file_path)) {
-			std::ofstream(std::string{file_path}) << '\0'; // force create the file
+			std::ofstream(native_file_path) << '\0'; // force create the file
+		}
+
+		_file_size = std::filesystem::file_size(native_file_path);
+		if (file_size >= 0 && _file_size != file_size) {
+			_file_size = file_size;
+			std::filesystem::resize_file(native_file_path, file_size); // ensure size, usually sparse
 		}
-		std::filesystem::resize_file(file_path, file_size); // ensure size, usually sparse
 
 		std::error_code err;
 		// sink, is also read
-		_file_map.map(std::string{file_path}, 0, file_size, err);
+		_file_map.map(native_file_path.u8string(), 0, _file_size, err);
 
 		if (err) {
-			// TODO: errro
+			std::cerr << "FileRWMapped error: mapping file failed " << err << "\n";
 			return;
 		}
 	}
 
-	virtual ~FileRWMapped(void) override {}
+	virtual ~File2RWMapped(void) override {}
 
 	bool isGood(void) override {
 		return _file_map.is_mapped();
 	}
 
-	std::vector<uint8_t> read(uint64_t pos, uint64_t size) override {
+	bool write(const ByteSpan data, int64_t pos = -1) override {
-		if (pos+size > _file_size) {
+		// TODO: support streaming write
-			//assert(false && "read past end");
+		if (pos < 0) {
-			return {};
-		}
-
-		return {_file_map.data()+pos, _file_map.data()+(pos+size)};
-	}
-
-	bool write(uint64_t pos, const std::vector<uint8_t>& data) override {
-		if (pos+data.size() > _file_size) {
 			return false;
 		}
 
-		std::memcpy(_file_map.data()+pos, data.data(), data.size());
+		if (data.empty()) {
+			return true; // false?
+		}
+
+		// file size is fix for mmaped files
+		if (pos+data.size > _file_size) {
+			return false;
+		}
+
+		std::memcpy(_file_map.data()+pos, data.ptr, data.size);
 
 		return true;
 	}
+
+	ByteSpanWithOwnership read(uint64_t size, int64_t pos = -1) override {
+		if (pos+size > _file_size) {
+			//assert(false && "read past end");
+			return ByteSpan{};
+		}
+
+		// return non-owning
+		return ByteSpan{_file_map.data()+pos, size};
+	}
 };
 

solanaceae/ngc_ft1_sha1/sha1_ngcft1.cpp

@@ -1,14 +1,12 @@
 #include "./sha1_ngcft1.hpp"
 
-#include <solanaceae/toxcore/utils.hpp>
+#include <solanaceae/util/utils.hpp>
 
 #include <solanaceae/contact/components.hpp>
 #include <solanaceae/tox_contacts/components.hpp>
 #include <solanaceae/message3/components.hpp>
 #include <solanaceae/tox_messages/components.hpp>
 
-#include <solanaceae/message3/file_r_file.hpp>
-
 #include "./ft1_sha1_info.hpp"
 #include "./hash_utils.hpp"
 
@@ -26,11 +24,11 @@
 
 namespace Message::Components {
 
-	using Content = ContentHandle;
+	using Content = ObjectHandle;
 
 } // Message::Components
 
-// TODO: rename to content components
+// TODO: rename to object components
 namespace Components {
 
 	struct Messages {
@@ -112,8 +110,7 @@ static size_t chunkSize(const FT1InfoSHA1& sha1_info, size_t chunk_index) {
 	}
 }
 
-void SHA1_NGCFT1::queueUpRequestChunk(uint32_t group_number, uint32_t peer_number, ContentHandle content, const SHA1Digest& hash) {
+void SHA1_NGCFT1::queueUpRequestChunk(uint32_t group_number, uint32_t peer_number, ObjectHandle content, const SHA1Digest& hash) {
-	// TODO: transfers
 	for (auto& [i_g, i_p, i_m, i_h, i_t] : _queue_requested_chunk) {
 		// if already in queue
 		if (i_g == group_number && i_p == peer_number && i_h == hash) {
@@ -123,6 +120,32 @@ void SHA1_NGCFT1::queueUpRequestChunk(uint32_t group_number, uint32_t peer_numbe
 		}
 	}
 
+	// check for running transfer
+	if (_sending_transfers.count(combineIds(group_number, peer_number))) {
+		for (const auto& [_, transfer] : _sending_transfers.at(combineIds(group_number, peer_number))) {
+			if (std::holds_alternative<SendingTransfer::Info>(transfer.v)) {
+				// ignore info
+				continue;
+			}
+
+			const auto& t_c = std::get<SendingTransfer::Chunk>(transfer.v);
+
+			if (content != t_c.content) {
+				// ignore different content
+				continue;
+			}
+
+			auto chunk_idx_vec = content.get<Components::FT1ChunkSHA1Cache>().chunkIndices(hash);
+
+			for (size_t idx : chunk_idx_vec) {
+				if (idx == t_c.chunk_index) {
+					// already sending
+					return; // skip
+				}
+			}
+		}
+	}
+
 	// not in queue yet
 	_queue_requested_chunk.push_back(std::make_tuple(group_number, peer_number, content, hash, 0.f));
 }
@@ -131,7 +154,7 @@ uint64_t SHA1_NGCFT1::combineIds(const uint32_t group_number, const uint32_t pee
 	return (uint64_t(group_number) << 32) | peer_number;
 }
 
-void SHA1_NGCFT1::updateMessages(ContentHandle ce) {
+void SHA1_NGCFT1::updateMessages(ObjectHandle ce) {
 	assert(ce.all_of<Components::Messages>());
 
 	for (auto msg : ce.get<Components::Messages>().messages) {
@@ -160,7 +183,7 @@ void SHA1_NGCFT1::updateMessages(ContentHandle ce) {
 	}
 }
 
-std::optional<std::pair<uint32_t, uint32_t>> SHA1_NGCFT1::selectPeerForRequest(ContentHandle ce) {
+std::optional<std::pair<uint32_t, uint32_t>> SHA1_NGCFT1::selectPeerForRequest(ObjectHandle ce) {
 	// get a list of peers we can request this file from
 	// TODO: randomly request from non SuspectedParticipants
 	std::vector<std::pair<uint32_t, uint32_t>> tox_peers;
@@ -194,6 +217,10 @@ std::optional<std::pair<uint32_t, uint32_t>> SHA1_NGCFT1::selectPeerForRequest(C
 					continue;
 				}
 
+				if (_cr.all_of<Contact::Components::TagSelfStrong>(child)) {
+					continue; // skip self
+				}
+
 				if (_cr.all_of<Contact::Components::ToxGroupPeerEphemeral>(child)) {
 					const auto& tgpe = _cr.get<Contact::Components::ToxGroupPeerEphemeral>(child);
 					un_tox_peers.push_back(tgpe.peer_number);
@@ -219,11 +246,13 @@ std::optional<std::pair<uint32_t, uint32_t>> SHA1_NGCFT1::selectPeerForRequest(C
 }
 
 SHA1_NGCFT1::SHA1_NGCFT1(
+	ObjectStore2& os,
 	Contact3Registry& cr,
 	RegistryMessageModel& rmm,
 	NGCFT1& nft,
 	ToxContactModel2& tcm
 ) :
+	_os(os),
 	_cr(cr),
 	_rmm(rmm),
 	_nft(nft),
@@ -266,8 +295,9 @@ void SHA1_NGCFT1::iterate(float delta) {
 			for (auto it = peer_it->second.begin(); it != peer_it->second.end();) {
 				it->second.time_since_activity += delta;
 
-				// if we have not heard for 10sec, timeout
+				// if we have not heard for 2min, timeout (lower level event on real timeout)
-				if (it->second.time_since_activity >= 10.f) {
+				// TODO: do we really need this if we get events?
+				if (it->second.time_since_activity >= 120.f) {
 					std::cerr << "SHA1_NGCFT1 warning: sending tansfer timed out " << "." << int(it->first) << "\n";
 					it = peer_it->second.erase(it);
 				} else {
@@ -320,8 +350,8 @@ void SHA1_NGCFT1::iterate(float delta) {
 		}
 
 		{ // requested info timers
-			std::vector<Content> timed_out;
+			std::vector<Object> timed_out;
-			_contentr.view<Components::ReRequestInfoTimer>().each([delta, &timed_out](Content e, Components::ReRequestInfoTimer& rrit) {
+			_os.registry().view<Components::ReRequestInfoTimer>().each([delta, &timed_out](Object e, Components::ReRequestInfoTimer& rrit) {
 				rrit.timer += delta;
 
 				// 15sec, TODO: config
@@ -331,12 +361,13 @@ void SHA1_NGCFT1::iterate(float delta) {
 			});
 			for (const auto e : timed_out) {
 				// TODO: avoid dups
-				_queue_content_want_info.push_back({_contentr, e});
+				_queue_content_want_info.push_back(_os.objectHandle(e));
-				_contentr.remove<Components::ReRequestInfoTimer>(e);
+				_os.registry().remove<Components::ReRequestInfoTimer>(e);
+				// TODO: throw update?
 			}
 		}
 		{ // requested chunk timers
-			_contentr.view<Components::FT1ChunkSHA1Requested>().each([delta](Components::FT1ChunkSHA1Requested& ftchunk_requested) {
+			_os.registry().view<Components::FT1ChunkSHA1Requested>().each([delta](Components::FT1ChunkSHA1Requested& ftchunk_requested) {
 				for (auto it = ftchunk_requested.chunks.begin(); it != ftchunk_requested.chunks.end();) {
 					it->second += delta;
 
@@ -522,11 +553,11 @@ bool SHA1_NGCFT1::onEvent(const Message::Events::MessageUpdated& e) {
 	// first, open file for write(+readback)
 	std::string full_file_path{e.e.get<Message::Components::Transfer::ActionAccept>().save_to_path};
 	// TODO: replace with filesystem or something
-	// TODO: ensure dir exists
 	if (full_file_path.back() != '/') {
 		full_file_path += "/";
 	}
 
+	// ensure dir exists
 	std::filesystem::create_directories(full_file_path);
 
 	const auto& info = ce.get<Components::FT1InfoSHA1>();
@@ -534,14 +565,13 @@ bool SHA1_NGCFT1::onEvent(const Message::Events::MessageUpdated& e) {
 
 	ce.emplace<Message::Components::Transfer::FileInfoLocal>(std::vector{full_file_path});
 
-	std::unique_ptr<FileRWMapped> file_impl;
 	const bool file_exists = std::filesystem::exists(full_file_path);
-
+	std::unique_ptr<File2I> file_impl = std::make_unique<File2RWMapped>(full_file_path, info.file_size);
-	file_impl = std::make_unique<FileRWMapped>(full_file_path, info.file_size);
 
 	if (!file_impl->isGood()) {
 		std::cerr << "SHA1_NGCFT1 error: failed opening file '" << full_file_path << "'!\n";
-		//e.e.remove<Message::Components::Transfer::ActionAccept>(); // stop
+		// we failed opening that filepath, so we should offer the user the oportunity to save it differently
+		e.e.remove<Message::Components::Transfer::ActionAccept>(); // stop
 		return false;
 	}
 
@@ -557,22 +587,24 @@ bool SHA1_NGCFT1::onEvent(const Message::Events::MessageUpdated& e) {
 			// iterate existing file
 			for (size_t i = 0; i < info.chunks.size(); i++) {
 				const uint64_t chunk_size = info.chunkSize(i);
-				auto existing_data = file_impl->read(i*uint64_t(info.chunk_size), chunk_size);
+				auto existing_data = file_impl->read(chunk_size, i*uint64_t(info.chunk_size));
-				assert(existing_data.size() == chunk_size);
 
-				// TODO: avoid copy
+				assert(existing_data.size == chunk_size);
+				if (existing_data.size == chunk_size) {
+					const auto data_hash = SHA1Digest{hash_sha1(existing_data.ptr, existing_data.size)};
+					const bool data_equal = data_hash == info.chunks.at(i);
 
-				const auto data_hash = SHA1Digest{hash_sha1(existing_data.data(), existing_data.size())};
+					cc.have_chunk.push_back(data_equal);
-				const bool data_equal = data_hash == info.chunks.at(i);
 
-				cc.have_chunk.push_back(data_equal);
+					if (data_equal) {
-
+						cc.have_count += 1;
-				if (data_equal) {
+						bytes_received += chunk_size;
-					cc.have_count += 1;
+						//std::cout << "existing i[" << info.chunks.at(i) << "] == d[" << data_hash << "]\n";
-					bytes_received += chunk_size;
+					} else {
-					//std::cout << "existing i[" << info.chunks.at(i) << "] == d[" << data_hash << "]\n";
+						//std::cout << "unk i[" << info.chunks.at(i) << "] != d[" << data_hash << "]\n";
+					}
 				} else {
-					//std::cout << "unk i[" << info.chunks.at(i) << "] != d[" << data_hash << "]\n";
+					// error reading?
 				}
 
 				_chunks[info.chunks[i]] = ce;
@@ -801,9 +833,7 @@ bool SHA1_NGCFT1::onEvent(const Events::NGCFT1_recv_data& e) {
 		for (const auto chunk_index : std::get<ReceivingTransfer::Chunk>(tv).chunk_indices) {
 			const auto offset_into_file = chunk_index* ce.get<Components::FT1InfoSHA1>().chunk_size;
 
-			// TODO: avoid temporary copy
+			if (!file->write({e.data, e.data_size}, offset_into_file + e.data_offset)) {
-			// TODO: check return
-			if (!file->write(offset_into_file + e.data_offset, {e.data, e.data + e.data_size})) {
 				std::cerr << "SHA1_NGCFT1 error: writing file failed o:" << offset_into_file + e.data_offset << "\n";
 			}
 		}
@@ -841,14 +871,17 @@ bool SHA1_NGCFT1::onEvent(const Events::NGCFT1_send_data& e) {
 		auto& chunk_transfer = std::get<SendingTransfer::Chunk>(transfer.v);
 		const auto& info = chunk_transfer.content.get<Components::FT1InfoSHA1>();
 		// TODO: should we really use file?
-		const auto data = chunk_transfer.content.get<Message::Components::Transfer::File>()->read((chunk_transfer.chunk_index * uint64_t(info.chunk_size)) + e.data_offset, e.data_size);
+		const auto data = chunk_transfer.content.get<Message::Components::Transfer::File>()->read(
+			e.data_size,
+			(chunk_transfer.chunk_index * uint64_t(info.chunk_size)) + e.data_offset
+		);
 
 		// TODO: optimize
-		for (size_t i = 0; i < e.data_size && i < data.size(); i++) {
+		for (size_t i = 0; i < e.data_size && i < data.size; i++) {
 			e.data[i] = data[i];
 		}
 
-		chunk_transfer.content.get_or_emplace<Message::Components::Transfer::BytesSent>().total += data.size();
+		chunk_transfer.content.get_or_emplace<Message::Components::Transfer::BytesSent>().total += data.size;
 		// TODO: add event to propergate to messages
 		//_rmm.throwEventUpdate(transfer); // should we?
 
@@ -929,10 +962,11 @@ bool SHA1_NGCFT1::onEvent(const Events::NGCFT1_recv_done& e) {
 		const auto chunk_size = info.chunkSize(chunk_index);
 		assert(offset_into_file+chunk_size <= info.file_size);
 
-		const auto chunk_data = ce.get<Message::Components::Transfer::File>()->read(offset_into_file, chunk_size);
+		const auto chunk_data = ce.get<Message::Components::Transfer::File>()->read(chunk_size, offset_into_file);
+		assert(!chunk_data.empty());
 
 		// check hash of chunk
-		auto got_hash = hash_sha1(chunk_data.data(), chunk_data.size());
+		auto got_hash = hash_sha1(chunk_data.ptr, chunk_data.size);
 		if (info.chunks.at(chunk_index) == got_hash) {
 			std::cout << "SHA1_NGCFT1: got chunk [" << SHA1Digest{got_hash} << "]\n";
 
@@ -954,7 +988,7 @@ bool SHA1_NGCFT1::onEvent(const Events::NGCFT1_recv_done& e) {
 							// HACK: remap file, to clear ram
 
 							// TODO: error checking
-							ce.get<Message::Components::Transfer::File>() = std::make_unique<FileRWMapped>(
+							ce.get<Message::Components::Transfer::File>() = std::make_unique<File2RWMapped>(
 								ce.get<Message::Components::Transfer::FileInfoLocal>().file_list.front(),
 								info.file_size
 							);
@@ -962,7 +996,7 @@ bool SHA1_NGCFT1::onEvent(const Events::NGCFT1_recv_done& e) {
 
 						// good chunk
 						// TODO: have wasted + metadata
-						ce.get_or_emplace<Message::Components::Transfer::BytesReceived>().total += chunk_data.size();
+						ce.get_or_emplace<Message::Components::Transfer::BytesReceived>().total += chunk_data.size;
 					}
 				}
 			} else {
@@ -1042,19 +1076,29 @@ bool SHA1_NGCFT1::onEvent(const Events::NGCFT1_recv_message& e) {
 	//reg.emplace<Components::TimestampWritten>(new_msg_e, 0);
 	reg.emplace<Message::Components::Timestamp>(new_msg_e, ts); // reactive?
 
+	reg.emplace<Message::Components::TagUnread>(new_msg_e);
+
 	{ // by whom
-		auto& synced_by = reg.get_or_emplace<Message::Components::SyncedBy>(new_msg_e).list;
+		reg.get_or_emplace<Message::Components::SyncedBy>(new_msg_e).ts.try_emplace(self_c, ts);
-		synced_by.emplace(self_c);
+	}
+
+	{ // we received it, so we have it
+		auto& rb = reg.get_or_emplace<Message::Components::ReceivedBy>(new_msg_e).ts;
+		rb.try_emplace(c, ts);
+		// TODO: how do we handle partial files???
+		// tox ft rn only sets self if the file was received fully
+		rb.try_emplace(self_c, ts);
 	}
 
 	// check if content exists
 	const auto sha1_info_hash = std::vector<uint8_t>{e.file_id, e.file_id+e.file_id_size};
-	ContentHandle ce;
+	ObjectHandle ce;
 	if (_info_to_content.count(sha1_info_hash)) {
 		ce = _info_to_content.at(sha1_info_hash);
 		std::cout << "SHA1_NGCFT1: new message has existing content\n";
 	} else {
-		ce = {_contentr, _contentr.create()};
+		// TODO: backend
+		ce = {_os.registry(), _os.registry().create()};
 		_info_to_content[sha1_info_hash] = ce;
 		std::cout << "SHA1_NGCFT1: new message has new content\n";
 
@@ -1142,9 +1186,7 @@ bool SHA1_NGCFT1::sendFilePath(const Contact3 c, std::string_view file_name, std
 		file_name_ = std::string(file_name),
 		file_path_ = std::string(file_path)
 	]() mutable {
-		// TODO: rw?
+		auto file_impl = std::make_unique<File2RWMapped>(file_path_, -1);
-		// TODO: memory mapped would be king
-		auto file_impl = std::make_unique<FileRFile>(file_path_);
 		if (!file_impl->isGood()) {
 			{
 				std::lock_guard l{self->_info_builder_queue_mutex};
@@ -1163,19 +1205,19 @@ bool SHA1_NGCFT1::sendFilePath(const Contact3 c, std::string_view file_name, std
 		FT1InfoSHA1 sha1_info;
 		// build info
 		sha1_info.file_name = file_name_;
-		sha1_info.file_size = file_impl->_file_size;
+		sha1_info.file_size = file_impl->_file_size; // TODO: remove the reliance on implementation details
 
 		{ // build chunks
 			// HACK: load file fully
-			// TODO: the speed is truly horrid
+			// ... its only a hack if its not memory mapped, but reading in chunk_sized chunks is probably a good idea anyway
-			const auto file_data = file_impl->read(0, file_impl->_file_size);
+			const auto file_data = file_impl->read(file_impl->_file_size, 0);
 			size_t i = 0;
-			for (; i + sha1_info.chunk_size < file_data.size(); i += sha1_info.chunk_size) {
+			for (; i + sha1_info.chunk_size < file_data.size; i += sha1_info.chunk_size) {
-				sha1_info.chunks.push_back(hash_sha1(file_data.data()+i, sha1_info.chunk_size));
+				sha1_info.chunks.push_back(hash_sha1(file_data.ptr+i, sha1_info.chunk_size));
 			}
 
-			if (i < file_data.size()) {
+			if (i < file_data.size) {
-				sha1_info.chunks.push_back(hash_sha1(file_data.data()+i, file_data.size()-i));
+				sha1_info.chunks.push_back(hash_sha1(file_data.ptr+i, file_data.size-i));
 			}
 		}
 
@@ -1194,7 +1236,7 @@ bool SHA1_NGCFT1::sendFilePath(const Contact3 c, std::string_view file_name, std
 			]() mutable { //
 				// back on iterate thread
 
-				auto file_impl = std::make_unique<FileRFile>(file_path_);
+				auto file_impl = std::make_unique<File2RWMapped>(file_path_, sha1_info.file_size);
 				if (!file_impl->isGood()) {
 					std::cerr << "SHA1_NGCFT1 error: failed opening file '" << file_path_ << "'!\n";
 					return;
@@ -1211,7 +1253,7 @@ bool SHA1_NGCFT1::sendFilePath(const Contact3 c, std::string_view file_name, std
 				std::cout << "SHA1_NGCFT1 sha1_info_hash: " << bin2hex(sha1_info_hash) << "\n";
 
 				// check if content exists
-				ContentHandle ce;
+				ObjectHandle ce;
 				if (self->_info_to_content.count(sha1_info_hash)) {
 					ce = self->_info_to_content.at(sha1_info_hash);
 
@@ -1280,7 +1322,8 @@ bool SHA1_NGCFT1::sendFilePath(const Contact3 c, std::string_view file_name, std
 						it = self->_queue_content_want_chunk.erase(it);
 					}
 				} else {
-					ce = {self->_contentr, self->_contentr.create()};
+					// TODO: backend
+					ce = {self->_os.registry(), self->_os.registry().create()};
 					self->_info_to_content[sha1_info_hash] = ce;
 
 					ce.emplace<Components::FT1InfoSHA1>(sha1_info);
@@ -1325,6 +1368,7 @@ bool SHA1_NGCFT1::sendFilePath(const Contact3 c, std::string_view file_name, std
 				reg_ptr->emplace<Message::Components::ContactTo>(msg_e, c);
 				reg_ptr->emplace<Message::Components::ContactFrom>(msg_e, c_self);
 				reg_ptr->emplace<Message::Components::Timestamp>(msg_e, ts); // reactive?
+				reg_ptr->emplace<Message::Components::Read>(msg_e, ts);
 
 				reg_ptr->emplace<Message::Components::Transfer::TagHaveAll>(msg_e);
 				reg_ptr->emplace<Message::Components::Transfer::TagSending>(msg_e);
@@ -1357,10 +1401,6 @@ bool SHA1_NGCFT1::sendFilePath(const Contact3 c, std::string_view file_name, std
 					// TODO: check return
 					self->_nft.NGC_FT1_send_message_public(group_number, message_id, static_cast<uint32_t>(NGCFT1_file_kind::HASH_SHA1_INFO), sha1_info_hash.data(), sha1_info_hash.size());
 					reg_ptr->emplace<Message::Components::ToxGroupMessageID>(msg_e, message_id);
-
-					// TODO: generalize?
-					auto& synced_by = reg_ptr->emplace<Message::Components::SyncedBy>(msg_e).list;
-					synced_by.emplace(c_self);
 				} else if (
 					// non online group
 					self->_cr.any_of<Contact::Components::ToxGroupPersistent>(c)
@@ -1368,12 +1408,11 @@ bool SHA1_NGCFT1::sendFilePath(const Contact3 c, std::string_view file_name, std
 					// create msg_id
 					const uint32_t message_id = randombytes_random();
 					reg_ptr->emplace<Message::Components::ToxGroupMessageID>(msg_e, message_id);
-
-					// TODO: generalize?
-					auto& synced_by = reg_ptr->emplace<Message::Components::SyncedBy>(msg_e).list;
-					synced_by.emplace(c_self);
 				}
 
+				reg_ptr->get_or_emplace<Message::Components::SyncedBy>(msg_e).ts.try_emplace(c_self, ts);
+				reg_ptr->get_or_emplace<Message::Components::ReceivedBy>(msg_e).ts.try_emplace(c_self, ts);
+
 				self->_rmm.throwEventConstruct(*reg_ptr, msg_e);
 
 				// TODO: place in iterate?

solanaceae/ngc_ft1_sha1/sha1_ngcft1.hpp

@@ -2,6 +2,7 @@
 
 // solanaceae port of sha1 fts for NGCFT1
 
+#include <solanaceae/object_store/object_store.hpp>
 #include <solanaceae/contact/contact_model3.hpp>
 #include <solanaceae/message3/registry_message_model.hpp>
 #include <solanaceae/tox_contacts/tox_contact_model2.hpp>
@@ -20,11 +21,9 @@
 #include <mutex>
 #include <list>
 
-enum class Content : uint32_t {};
-using ContentRegistry = entt::basic_registry<Content>;
-using ContentHandle = entt::basic_handle<ContentRegistry>;
-
 class SHA1_NGCFT1 : public RegistryMessageModelEventI, public NGCFT1EventI {
+	ObjectStore2& _os;
+	// TODO: backend abstraction
 	Contact3Registry& _cr;
 	RegistryMessageModel& _rmm;
 	NGCFT1& _nft;
@@ -32,21 +31,18 @@ class SHA1_NGCFT1 : public RegistryMessageModelEventI, public NGCFT1EventI {
 
 	std::minstd_rand _rng {1337*11};
 
-	// registry per group?
-	ContentRegistry _contentr;
-
 	// limit this to each group?
-	entt::dense_map<SHA1Digest, ContentHandle> _info_to_content;
+	entt::dense_map<SHA1Digest, ObjectHandle> _info_to_content;
 
 	// sha1 chunk index
 	// TODO: optimize lookup
 	// TODO: multiple contents. hashes might be unique, but data is not
-	entt::dense_map<SHA1Digest, ContentHandle> _chunks;
+	entt::dense_map<SHA1Digest, ObjectHandle> _chunks;
 
 	// group_number, peer_number, content, chunk_hash, timer
-	std::deque<std::tuple<uint32_t, uint32_t, ContentHandle, SHA1Digest, float>> _queue_requested_chunk;
+	std::deque<std::tuple<uint32_t, uint32_t, ObjectHandle, SHA1Digest, float>> _queue_requested_chunk;
 	//void queueUpRequestInfo(uint32_t group_number, uint32_t peer_number, const SHA1Digest& hash);
-	void queueUpRequestChunk(uint32_t group_number, uint32_t peer_number, ContentHandle content, const SHA1Digest& hash);
+	void queueUpRequestChunk(uint32_t group_number, uint32_t peer_number, ObjectHandle content, const SHA1Digest& hash);
 
 	struct SendingTransfer {
 		struct Info {
@@ -56,7 +52,7 @@ class SHA1_NGCFT1 : public RegistryMessageModelEventI, public NGCFT1EventI {
 		};
 
 		struct Chunk {
-			ContentHandle content;
+			ObjectHandle content;
 			size_t chunk_index; // <.< remove offset_into_file
 			//uint64_t offset_into_file;
 			// or data?
@@ -72,14 +68,14 @@ class SHA1_NGCFT1 : public RegistryMessageModelEventI, public NGCFT1EventI {
 
 	struct ReceivingTransfer {
 		struct Info {
-			ContentHandle content;
+			ObjectHandle content;
 			// copy of info data
 			// too large?
 			std::vector<uint8_t> info_data;
 		};
 
 		struct Chunk {
-			ContentHandle content;
+			ObjectHandle content;
 			std::vector<size_t> chunk_indices;
 			// or data?
 			// if memmapped, this would be just a pointer
@@ -93,8 +89,8 @@ class SHA1_NGCFT1 : public RegistryMessageModelEventI, public NGCFT1EventI {
 	entt::dense_map<uint64_t, entt::dense_map<uint8_t, ReceivingTransfer>> _receiving_transfers;
 
 	// makes request rotate around open content
-	std::deque<ContentHandle> _queue_content_want_info;
+	std::deque<ObjectHandle> _queue_content_want_info;
-	std::deque<ContentHandle> _queue_content_want_chunk;
+	std::deque<ObjectHandle> _queue_content_want_chunk;
 
 	std::atomic_bool _info_builder_dirty {false};
 	std::mutex _info_builder_queue_mutex;
@@ -108,20 +104,21 @@ class SHA1_NGCFT1 : public RegistryMessageModelEventI, public NGCFT1EventI {
 
 	static uint64_t combineIds(const uint32_t group_number, const uint32_t peer_number);
 
-	void updateMessages(ContentHandle ce);
+	void updateMessages(ObjectHandle ce);
 
-	std::optional<std::pair<uint32_t, uint32_t>> selectPeerForRequest(ContentHandle ce);
+	std::optional<std::pair<uint32_t, uint32_t>> selectPeerForRequest(ObjectHandle ce);
 
 	public: // TODO: config
 		bool _udp_only {false};
 
-		size_t _max_concurrent_in {4};
+		size_t _max_concurrent_in {6};
-		size_t _max_concurrent_out {6};
+		size_t _max_concurrent_out {8};
 		// TODO: probably also includes running transfers rn (meh)
 		size_t _max_pending_requests {32}; // per content
 
 	public:
 		SHA1_NGCFT1(
+			ObjectStore2& os,
 			Contact3Registry& cr,
 			RegistryMessageModel& rmm,
 			NGCFT1& nft,