logaritmisk
  • Joined on 2023-10-06
logaritmisk commented on issue logaritmisk/trueskill-tt#19 2026-08-27 15:12:48 +00:00
HistoryBuilder::online(true) is non-functional — Skill.online is never written

Fixed on main (bf9d964..69ddebe). Design doc: docs/superpowers/specs/2026-08-27-filtered-estimates-design.md.

Both acceptance criteria are met — the setting is gone from the public…

logaritmisk opened issue logaritmisk/trueskill-tt#31 2026-08-27 14:59:47 +00:00
Skill's compile-time divergence tripwire is only half-enforced
logaritmisk opened issue logaritmisk/trueskill-tt#30 2026-08-27 14:48:59 +00:00
Spec's tolerance-not-bit-identity caveat for filtered estimates is conservative
logaritmisk opened issue logaritmisk/trueskill-tt#29 2026-08-27 14:48:55 +00:00
log_evidence takes &mut self but mutates nothing
logaritmisk opened issue logaritmisk/trueskill-tt#28 2026-08-27 14:48:53 +00:00
log_evidence_internal's forward flag is only a filtering quantity before convergence
logaritmisk commented on issue logaritmisk/trueskill-tt#19 2026-08-27 13:29:18 +00:00
HistoryBuilder::online(true) is non-functional — Skill.online is never written

A second consumer for the same machinery turned up in ustat, which bears on the implement-vs-remove choice above.

**The use case: a filtered learning…

logaritmisk opened issue logaritmisk/trueskill-tt#27 2026-08-27 13:20:41 +00:00
converge() panics on a History with no events (usize underflow in iteration())
logaritmisk pushed to main at logaritmisk/xy 2026-08-10 12:50:45 +00:00
7f605f361e test(uptime): assert on the wire value, not the rendered table
logaritmisk pushed to main at logaritmisk/xy 2026-08-10 12:07:06 +00:00
29df988d9d feat(cli): render uptime in human units
3dcda14ba5 test(harness): put each restart key on its own line
44304d0625 fix(protocol): reject one-line KDL blocks with multiple keys
938cbcbc08 test(wait-for): pin the gate end to end
516d7d0b86 fix(daemon): keep serving while a gated server waits
Compare 35 commits »
logaritmisk closed issue logaritmisk/trueskill-tt#7 2026-08-04 20:13:51 +00:00
README Todo list and CLAUDE.md architecture section are stale post-redesign
logaritmisk closed issue logaritmisk/trueskill-tt#24 2026-08-04 20:13:50 +00:00
No CI — nothing verifies builds, tests, feature combinations, or determinism
logaritmisk closed issue logaritmisk/trueskill-tt#22 2026-08-04 20:13:43 +00:00
Convergence configuration is inconsistent: two different iteration defaults, and converged: false for trivially-converged graphs
logaritmisk closed issue logaritmisk/trueskill-tt#16 2026-08-04 20:13:43 +00:00
learning_curves() is O(n²) because KeyTable::key is a linear scan
logaritmisk closed issue logaritmisk/trueskill-tt#15 2026-08-04 20:13:42 +00:00
Hot-path Gaussian arithmetic round-trips through sqrt; convergence deltas computed in moment space
logaritmisk closed issue logaritmisk/trueskill-tt#14 2026-08-04 20:13:40 +00:00
Adding events one at a time is quadratic — 130× slower than batching at n=2000
logaritmisk closed issue logaritmisk/trueskill-tt#13 2026-08-04 20:13:35 +00:00
rayon path creates multiple aliasing &mut SkillStore through a usize-laundered pointer (UB)
logaritmisk closed issue logaritmisk/trueskill-tt#12 2026-08-04 20:13:34 +00:00
Evidence accumulates as a linear product, so log-evidence underflows to -inf on large games
logaritmisk closed issue logaritmisk/trueskill-tt#11 2026-08-04 20:13:33 +00:00
History::converge() panics on an empty history (usize underflow)
logaritmisk closed issue logaritmisk/trueskill-tt#9 2026-08-04 20:13:32 +00:00
quality() and predict_quality() panic for anything other than exactly 2 rating groups
logaritmisk closed issue logaritmisk/trueskill-tt#8 2026-08-04 20:13:31 +00:00
Draws with the default p_draw = 0.0 silently produce NaN skills and report converged: true