Undertow
Undertow 2.4.0 clears three CVEs and finally lands long-open HTTP/2 and timeout requests
A side-by-side editorial comparison of Betaflight and NumPy — release velocity, themes, recent moves, and the top alternatives to consider.
Betaflight grew a real autopilot: waypoint missions, geofence RTH and MAVLink ground control
Betaflight runs two lines at once. The 2025.12.x branch is pure maintenance — UART interrupt handling, MSP race conditions, failsafe clamping, and periodic board-target refreshes — while the 2026.6 line, which reached 2026.6.1 in August, carries the largest feature set in this window by a wide margin. That feature set is dominated by one thread: an AUTOPILOT mode with flight-plan configuration, CLI waypoint commands, in-flight waypoint capture on a switch, geofence return-to-home, LAND loiter and TAKEOFF patterns, plus OSD waypoint elements and a navigation minimap. Alongside it sits a DroneCAN protocol stack with ESC and GNSS providers, CAN peripheral support on three MCU families, MAVLink mission upload and download, and STM32N6 platform bring-up.
NumPy cut distutils loose and is quietly rebuilding for free-threaded Python.
NumPy is in the maintenance rhythm of a foundational library: a transitional minor release followed by a run of patch releases cleaning up what it broke. 2.5.0 removed distutils, expired a large batch of 2.0-era deprecations, and dropped Python 3.11. The patch line since has been about compatibility surfaces — a Cython datetime API fix so downstream can still target pre-2.5, a GCC minimum bump, and wheels for Python 3.15 release candidates.
Betaflight runs two lines at once. The 2025.12.x branch is pure maintenance — UART interrupt handling, MSP race conditions, failsafe clamping, and periodic board-target refreshes — while the 2026.6 line, which reached 2026.6.1 in August, carries the largest feature set in this window by a wide margin. That feature set is dominated by one thread: an AUTOPILOT mode with flight-plan configuration, CLI waypoint commands, in-flight waypoint capture on a switch, geofence return-to-home, LAND loiter and TAKEOFF patterns, plus OSD waypoint elements and a navigation minimap. Alongside it sits a DroneCAN protocol stack with ESC and GNSS providers, CAN peripheral support on three MCU families, MAVLink mission upload and download, and STM32N6 platform bring-up.
The autopilot work is not a single feature but a rebuild of how the firmware handles position: the release notes show position control unified on a velocity primitive and the existing GPS rescue reimplemented as a flight-plan mission rather than a special case. That direction pulls hardware integration along with it — DroneCAN node allocation, GNSS Fix2, ESC telemetry and MAVLink ground-station handshaking are all things a manually flown craft does not need. The SITL harness gained fidelity and per-run session ids in the same releases, which is what you build when autonomous behaviour has to be regression-tested rather than test-flown.
With GPS rescue now running through the mission engine, the next releases on this line should concentrate on hardening that path — the release candidates already added an rx-loss policy, sanity checks and an arming gate for flight plans, and safety scaffolding of that kind usually keeps accumulating after the feature ships.
NumPy is in the maintenance rhythm of a foundational library: a transitional minor release followed by a run of patch releases cleaning up what it broke. 2.5.0 removed distutils, expired a large batch of 2.0-era deprecations, and dropped Python 3.11. The patch line since has been about compatibility surfaces — a Cython datetime API fix so downstream can still target pre-2.5, a GCC minimum bump, and wheels for Python 3.15 release candidates.
Two forces are steering releases. One is Python itself: NumPy is tracking 3.15 before it ships and steadily improving free-threading support, including fixing an ABI leak in the free-threading-compatible stable ABI. The other is the array-api standard, which is pulling NumPy's own semantics into line — descending sorts landed for exactly that reason.
Expect the 2.5.x line to keep absorbing free-threading and Python 3.15 fallout; the entries suggest the interesting work now happens at the C API and build-system layers, not in array semantics.
Other DevOps products tracked by Sparkpulse, ranked by recent ship velocity. Each card links to a full editorial trajectory and lets you pivot into a head-to-head comparison with either Betaflight or NumPy.
Undertow 2.4.0 clears three CVEs and finally lands long-open HTTP/2 and timeout requests
OceanBase is rebuilding itself as a RAG backend without giving up the HTAP story
QGroundControl rebuilt its flight UI around touch screens, then went quiet for ten months
Pelican now ships roughly once a year, and 4.12 is theme housekeeping
WildFly's quarterly train is really a stability ladder, and OIDC keeps climbing it
Jakarta REST implementation in pure maintenance across two parallel branches.
See all Betaflight alternatives → · See all NumPy alternatives →
Latest ship moves from both products, interleaved chronologically. ⚡ = editorial spark.
They serve adjacent needs but don't currently overlap on shipped themes. Betaflight is currently shipping more aggressively (velocity 6.3 vs 2.5), with 1 editorial sparks in the last 30 days against 0. See the at-a-glance table above for a side-by-side breakdown of velocity, recent sparks, and editorial themes.
Sparkpulse doesn't pick a winner — we score release velocity, not feature parity. Betaflight is currently shipping more aggressively (velocity 6.3 vs 2.5), with 1 editorial sparks in the last 30 days against 0. For your specific use case, the alternatives sections above list other DevOps products to evaluate alongside.
Top Betaflight alternatives in DevOps are ranked by recent ship velocity. Browse the "Betaflight alternatives" section above for the current picks, or visit /alternatives/betaflight for the full list with editorial commentary on each.
Top NumPy alternatives in DevOps are ranked by recent ship velocity. Browse the "NumPy alternatives" section above for the current picks, or visit /alternatives/numpy for the full list with editorial commentary on each.