Issues

Setting DORA_OTLP_ENDPOINT exports zenoh's trace-level message dumps as spans, formatting every payload

bug

#3657

· ajaypen19 opened

on Sep 29, 2026

hub-client: index repo can truncate arbitrary user files via symlinked cache marker (`.dora-clone-complete` / `.dora-index-source`)

cli

python

rust

#3654

· phil-opp opened

on Sep 29, 2026

Cleanup: unused code, dependencies and stale docs (tracking)

#3637

· sanyyam0305 opened

on Sep 27, 2026

Node API scheduler: an input named `dora.non_input_event` merges with the control-event queue (queue_size ignored, fairness bypassed, flush/eviction can drop InputClosed)

rust

#3632

· phil-opp opened

on Sep 27, 2026

daemon/coordinator: resent finish report (#3612) is dropped when the daemon re-registers under a new DaemonId, so a normally-finished dataflow is stored as terminally Failed ("recovery timeout")

coordinator

daemon

#3631

· phil-opp opened

on Sep 27, 2026

daemon: #3610's `gave_up` clear condition in `note_drained` is equivalent to "clear on any take" in reachable states, so #3601's self-loop/cycle stall remains (test relies on an unreachable channel state)

daemon

#3630

· phil-opp opened

on Sep 27, 2026

daemon: the message that finds a receiver's event channel closed is dropped without being counted (escapes `fail_on_lost_backpressure_messages`)

daemon

rust

#3620

· phil-opp opened

on Sep 26, 2026

daemon: a held backpressure message can reach the receiver after `InputClosed`/`AllInputsClosed` when its producer crashes, and is then silently dropped (#3590)

bug

daemon

rust

#3619

· phil-opp opened

on Sep 26, 2026

`queue_size` is enforced only at `next()`, so a slow subscriber can pin up to 64 producer SHM chunks and exhaust the producer's pool

bug

daemon

rust

#3591

· ajaypen19 opened

on Sep 23, 2026

security: replace Zenoh's abandoned lockfree/owned-alloc dependency chain

dependencies

rust

security

#3577

· phil-opp opened

on Sep 21, 2026

ROS 2 bridge is Humble-only (YAML bridge + Python Ros2Context): expose distro feature, make Python bridge optional, per-distro builds?

#3554

· kahzmic opened

on Sep 18, 2026

Topic debug frames must not share the daemon→coordinator control channel

#3535

· phil-opp opened

on Sep 16, 2026

Navigation Menu

All issues

Setting DORA_OTLP_ENDPOINT exports zenoh's trace-level message dumps as spans, formatting every payload

hub-client: index repo can truncate arbitrary user files via symlinked cache marker (`.dora-clone-complete` / `.dora-index-source`)

Cleanup: unused code, dependencies and stale docs (tracking)

Node API scheduler: an input named `dora.non_input_event` merges with the control-event queue (queue_size ignored, fairness bypassed, flush/eviction can drop InputClosed)

daemon/coordinator: resent finish report (#3612) is dropped when the daemon re-registers under a new DaemonId, so a normally-finished dataflow is stored as terminally Failed ("recovery timeout")

daemon: #3610's `gave_up` clear condition in `note_drained` is equivalent to "clear on any take" in reachable states, so #3601's self-loop/cycle stall remains (test relies on an unreachable channel state)

daemon: the message that finds a receiver's event channel closed is dropped without being counted (escapes `fail_on_lost_backpressure_messages`)

daemon: a held backpressure message can reach the receiver after `InputClosed`/`AllInputsClosed` when its producer crashes, and is then silently dropped (#3590)

`queue_size` is enforced only at `next()`, so a slow subscriber can pin up to 64 producer SHM chunks and exhaust the producer's pool

security: replace Zenoh's abandoned lockfree/owned-alloc dependency chain

ROS 2 bridge is Humble-only (YAML bridge + Python Ros2Context): expose distro feature, make Python bridge optional, per-distro builds?

Topic debug frames must not share the daemon→coordinator control channel

Search results