Block a user
just release-plan leaves the tree dirty, so just release always aborts
Documentation has drifted from the code: four README examples no longer compile
just release-plan leaves the tree dirty, so just release always aborts
Documentation has drifted from the code: four README examples no longer compile
Allow drift to vary per competitor:
Member::with_drift_scale
Allow drift to vary per competitor:
Member::with_drift_scale
ConvergenceReport.slices_skipped is public, always zero, and nothing will ever populate it
ConvergenceReport.slices_skipped is public, always zero, and nothing will ever populate it
Ingestion discards all convergence state, so appending one event costs a full re-converge
Ingestion discards all convergence state, so appending one event costs a full re-converge
Spiked. The premise of this issue is wrong — the re-convergence is necessary work, not waste. Recommending we close it.
I filed this off the observation that appending one event costs a full…
Cross-slice parallelism: per-slice dirty bits to skip clean slices
Ingestion discards all convergence state, so appending one event costs a full re-converge
Cross-slice parallelism: per-slice dirty bits to skip clean slices
Closing: measured, and this approach cannot pay for itself. The design is also unsound in three independent ways, but the measurement alone settles it.
Work is on branch dirty-slice-skipping…
Test suite has no coverage for degenerate inputs, release-mode behaviour, or non-finite results
Three of the four remaining items done — 2745fbb. Staying open for the fourth, which needs a decision rather than work.
Property-based tests — tests/properties.rs, proptest as a…
Per-slice
SkillStore is dense over the global index space — memory scales with the largest competitor index, not slice size