authentik
authentik 2026.8 ships: Actors, domain-joined Agents, and a push past browser-mediated SSO
A side-by-side editorial comparison of BayLum and RabbitMQ — release velocity, themes, recent moves, and the top alternatives to consider.
Bayesian luminescence dating that finally replaced its folder-structure input format.
BayLum runs Bayesian age models for luminescence and combined OSL/C-14 dating on top of JAGS. The 2024 release rebuilt the front end: a single create_DataFile() replaces the separate single-grain and multi-grain generators, reads BIN/BINX and XSYG directly, and takes a YAML config in place of the old prescribed folder layout. Since then the work has been CRAN compliance and documentation.
Two parallel trains, and the 'maintenance' label is now hiding real feature work
RabbitMQ is maintaining 4.2.x and 4.3.x side by side, cutting matching patches into both on the same day, and both trains raised their floor to Erlang/OTP 27. Most of the window is correctness work in the Raft-backed subsystems — quorum queues losing metrics after a restart, leaders committing log entries too optimistically, classic queue index paths accumulating slashes, topic bindings with empty routing keys matching everything. The newest 4.3.5, however, is labelled a maintenance release while carrying encrypted management-UI login tokens, a new authentication logging category, ETag support on the definitions endpoint, and a self-deleting Shovel TTL.
BayLum runs Bayesian age models for luminescence and combined OSL/C-14 dating on top of JAGS. The 2024 release rebuilt the front end: a single create_DataFile() replaces the separate single-grain and multi-grain generators, reads BIN/BINX and XSYG directly, and takes a YAML config in place of the old prescribed folder layout. Since then the work has been CRAN compliance and documentation.
Two long-running threads have converged: making JAGS runs survivable (parallel methods, halved MCMC memory, injectable custom models) and making the inputs survivable (YAML config, consistency checks, auto-detected sample names). With the deprecated generators on their way out, the next phase is removal rather than addition. Release cadence is roughly annual and slowing.
The deprecated Generate_DataFile(), Generate_DataFile_MG() and LT_RegenDose() are the obvious next casualties; a release that drops them would be the first breaking change since the YAML rework.
RabbitMQ is maintaining 4.2.x and 4.3.x side by side, cutting matching patches into both on the same day, and both trains raised their floor to Erlang/OTP 27. Most of the window is correctness work in the Raft-backed subsystems — quorum queues losing metrics after a restart, leaders committing log entries too optimistically, classic queue index paths accumulating slashes, topic bindings with empty routing keys matching everything. The newest 4.3.5, however, is labelled a maintenance release while carrying encrypted management-UI login tokens, a new authentication logging category, ETag support on the definitions endpoint, and a self-deleting Shovel TTL.
The bug pattern remains the tell: nearly every fix is in quorum queues, Khepri or Raft, which is where RabbitMQ moved its metadata and durability story after 4.3.0 removed Mnesia and partition-handling strategies outright. Layered on top is a steady tightening of the operational perimeter — protocol parsers rejecting malformed input strictly across AMQP 1.0, MQTT 5.0 and STOMP, pre-authentication frame limits on stream connections, HTTP API endpoints validating node membership, and headers that stop disclosing supported methods. Feature work is arriving inside patch releases rather than waiting for a minor.
Expect the 4.2.x train to slow toward end-of-life while 4.3.x patches keep absorbing both Khepri edge cases and security-surface work. The encrypted login token, currently opt-in behind a shared cluster secret, is the kind of setting that gets promoted to a default once rolling-upgrade friction is behind it.
Other Infra & APIs 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 BayLum or RabbitMQ.
authentik 2026.8 ships: Actors, domain-joined Agents, and a push past browser-mediated SSO
Rancher's public feed is a build-tag stream: three branches bumped the same Go image on one afternoon
Buildkite keeps converting hand-rolled agent workarounds into first-class CI primitives.
Cursor's agents stop waiting to be asked - they subscribe, and they hold a goal until it's done.
Nexus does the diagnosis; the agent is now reaching into the status page too.
Warp turned its quarter of software-factory essays into infrastructure you can buy.
See all BayLum alternatives → · See all RabbitMQ alternatives →
Latest ship moves from both products, interleaved chronologically. ⚡ = editorial spark.
They serve adjacent needs but don't currently overlap on shipped themes. RabbitMQ is currently shipping more aggressively (velocity 5.0 vs 0.0), with 0 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. RabbitMQ is currently shipping more aggressively (velocity 5.0 vs 0.0), with 0 editorial sparks in the last 30 days against 0. For your specific use case, the alternatives sections above list other Infra & APIs products to evaluate alongside.
Top BayLum alternatives in Infra & APIs are ranked by recent ship velocity. Browse the "BayLum alternatives" section above for the current picks, or visit /alternatives/baylum for the full list with editorial commentary on each.
Top RabbitMQ alternatives in Infra & APIs are ranked by recent ship velocity. Browse the "RabbitMQ alternatives" section above for the current picks, or visit /alternatives/rabbitmq for the full list with editorial commentary on each.