logger
R's logging package handed to new maintainers and rebuilt from the inside out
A side-by-side editorial comparison of discretetests and Dragonfly — release velocity, themes, recent moves, and the top alternatives to consider.
Four years of releases land in four minutes as the changelog gets backfilled to GitHub
DiscreteTests computes exact p-values and their full support for discrete statistical tests, returning results as a class that carries the observations, the test statistic and the null distribution. All four visible releases were tagged within four minutes of each other on a single day in May 2026, in reverse order — 0.4.1 carries the earliest timestamp and 0.4.0 the latest — so this is a historical changelog being pushed to GitHub at once, not four days of work.
Dragonfly spends its releases on correctness in the parts of Redis compatibility nobody tests.
The visible window is almost entirely correctness and internals work on the 1.39/1.40 lines. The newest release consolidates two tiering knobs into a single tiered_max_pending_bytes and exports it to metrics. Behind it sit a memory-accounting backport, a cluster fix that journals FLUSHSLOTS from inside the flush transaction, and a HyperLogLog resync with valkey upstream that pulled in a CVE fix and exposed three Dragonfly-local bugs.
DiscreteTests computes exact p-values and their full support for discrete statistical tests, returning results as a class that carries the observations, the test statistic and the null distribution. All four visible releases were tagged within four minutes of each other on a single day in May 2026, in reverse order — 0.4.1 carries the earliest timestamp and 0.4.0 the latest — so this is a historical changelog being pushed to GitHub at once, not four days of work.
Read by version rather than by timestamp, the package grew from binomial, Fisher, McNemar and Poisson tests into rank-based and resampling territory: Wilcoxon signed rank and rank sum, Mann-Whitney U, a conditional homogeneity test, then sign tests and a general permutation test with several built-in statistics. Alongside that the result class was extended to hold ragged observation lists and computational detail, the print output was rewritten on cli, and the old dot-separated function names were made defunct.
The permutation test with pluggable statistics is the most open-ended addition here, and adding further built-in statistics to it is the cheapest way to extend the package from where it now stands.
The visible window is almost entirely correctness and internals work on the 1.39/1.40 lines. The newest release consolidates two tiering knobs into a single tiered_max_pending_bytes and exports it to metrics. Behind it sit a memory-accounting backport, a cluster fix that journals FLUSHSLOTS from inside the flush transaction, and a HyperLogLog resync with valkey upstream that pulled in a CVE fix and exposed three Dragonfly-local bugs.
Two threads run through this window. One is upstream parity — the hyperloglog sync explicitly re-forks from valkey and keeps upstream formatting so future syncs stay cheap, which is a maintenance-cost decision as much as a correctness one. The other is the RESP IO loop v2 migration, visible in a pub/sub ordering fix gated on the v1 loop and in tests being skipped on the v1 leg of the regression matrix. Dragonfly is hardening a drop-in Redis replacement in the places where drop-in claims usually break: cluster operations, pub/sub ordering under pipelining, and memory accounting.
Expect continued RESP IO loop v2 work and more upstream syncs on forked Redis internals. The tiering flag consolidation plus its metrics export suggests tiered storage tuning is the next area to get operator-facing surface.
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 discretetests or Dragonfly.
R's logging package handed to new maintainers and rebuilt from the inside out
Four dormant years end with a test fix, not a feature
A comorbidity-coding package relicenses to BSD and grows past the American coding systems
Three kill-switch explainers to one point release — Unleash's feed is content-led, not release-led.
Jackett's daily release treadmill now spends commits undoing tracker removals it made days earlier.
Kinsta keeps moving MyKinsta's control panel into its API, month by month.
See all discretetests alternatives → · See all Dragonfly alternatives →
Latest ship moves from both products, interleaved chronologically. ⚡ = editorial spark.
They serve adjacent needs but don't currently overlap on shipped themes. Dragonfly 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. Dragonfly 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 discretetests alternatives in Infra & APIs are ranked by recent ship velocity. Browse the "discretetests alternatives" section above for the current picks, or visit /alternatives/discretetests for the full list with editorial commentary on each.
Top Dragonfly alternatives in Infra & APIs are ranked by recent ship velocity. Browse the "Dragonfly alternatives" section above for the current picks, or visit /alternatives/dragonfly for the full list with editorial commentary on each.