Routing
Meshy supports two routing strategies for delivering direct messages: Classic and Smart. You can set a global default in Settings > Routing and Messaging and optionally override it for individual contacts in the contact detail dialog.
Classic Routing
Classic routing is very similar to the behavior of the official client. If there is no established path between you and the contact, the message is flood routed using the configured Flood Attempts (default 4). If an established path exists, it is used for the configured Path Attempts (default 4). After those attempts it falls back to flood routing using the configured Flood Attempts, unless Automatically Reset Path is disabled. If all attempts fail, the message is marked as failed.
If an established path exists, it is pre-filled in the Out Path field in the form of comma-separated hex hop identifiers (e.g., 9a,74 for 1-byte hashes, 9a3b,744c for 2-byte). The hash size is auto-detected from the hop length.
Classic routing allows you to:
- manually set a path by typing it in the Out Path field.
- reset the path when you know the established path is outdated and you would like to directly switch to flood routing to discover a new one.
- disable Automatically Reset Path when you don't want to automatically switch to flood routing after failed delivery attempts. Useful in situations when the established path is not stable, but you still trust it to deliver the message and want to avoid flood routing.
- force flood when you only want to use flood routing without establishing a path. Please note that the flood routing is very expensive for the mesh network and should be avoided unless necessary.
- show Path on Map, which visualizes the route on a map with colored markers at each hop.
Smart Routing
Smart routing tries to solve the main problem with classic routing, which is that it uses only one path for delivery attempts. Instead, it uses a list of recently successful paths and intelligently selects the best one for each attempt.
If there are no available paths on the list, it uses flood routing with the configured Flood Attempts (default 4), like the classic routing. If delivery is successful, the path used is added to the list with an initial weight of 2.5. Each additional successful delivery increases the path's weight by 0.5, while an unsuccessful delivery decreases it by 0.2. If a path is discovered by the firmware (e.g. via a message received from the contact), it is added to the list with an initial weight of 1.0. The weight is capped at 5.0. If a path's weight drops to 0 or below, it is removed from the list, and a path that has been removed this way is not re-added just because it still exists in the firmware's routing table. The weight serves as a health indicator -- it keeps useful paths in the pool and prunes dead ones, but it does not determine the order in which paths are tried.
If there are multiple paths on the list, smart routing selects the best path for each delivery attempt (up to the configured Path Attempts, default 4). The selection is based on a composite score that combines three factors:
| Factor | Weight | Formula |
|---|---|---|
| Reliability | 60% | (successes + 1) / (total_attempts + 2) -- Laplace-smoothed success rate |
| Latency | 20% | 1 / (1 + last_trip_ms / 10 000) -- lower latency -> higher score (0.5 if unknown) |
| Recency | 20% | 1 / (1 + age_days) -- days since last successful delivery (0 if never) |
score = reliability x 0.6 + latency x 0.2 + recency x 0.2
Each attempt uses a different path from the available pool, selected by highest score. If there are fewer paths than attempts, the best paths can be reused. After the path attempts are exhausted, smart routing falls back to flood using the configured Flood Attempts (default 4). If there are no paths left in the list, it automatically switches to flood routing and tries to find a new path.
Clicking Available Paths in the contact detail gives you a list of recently successful paths, showing their weights, the number of successful and unsuccessful attempts, the number of hops and the time taken for the last delivery. Each path is color-coded with a dot for easy identification. Clicking a path row expands it to show the full route with each hop resolved to a node name (unknown nodes fall back to their hash) and a copy button that copies either the raw hop prefixes or the resolved names. Clicking the map icon in the header bar opens a map dialog showing all available paths simultaneously, each drawn as a colored line matching its dot in the list. This allows you to visually compare paths and see which repeaters they pass through. If the list is outdated and no longer useful for routing, click the Clear History button. This will forget all paths and switch routing to flood to find new paths.
Choosing Between Classic and Smart
| Classic | Smart |
|---|---|
| Direct manual control over paths | Automatic path selection based on history |
| Better for mobile nodes | Better for stationary nodes |
| Useful for troubleshooting | Learns optimal routes over time |
| Single path at a time | Multiple paths with quality tracking |
Global Default and Per-Contact Override
The Default Routing toggle in Settings > Routing and Messaging sets the routing strategy used for all contacts. When you change a contact's routing mode in the contact detail dialog, it becomes a per-contact override that persists independently of the global setting. Changing the global default will not affect contacts that have been explicitly overridden.