5 Commits
Author SHA1 Message Date
jason.woltje d243f6b25f gizmoduck: row 307 run logs: capture.mjs makes each fixture's data root before sessionStore (base session.mjs:78 needs it), BUILD.md expected red adds claude-history test B (duckworth-0287), steps 1-6 results on 1f6db1589 and 1f00d4f91 with skips named and M1-M4 killed (sage-3455, sage-3465) 2026-10-23 09:37:50 -05:00
jason.woltje 869670d1b1 gizmoduck: row 307 BUILD.md: commits, old-test edits against mos-1459, new tests and the empty-session correction to test B, expected red on 1f6db1589, run plan with mos-1275 limits, mutants M1-M4 2026-10-23 09:37:50 -05:00
jason.woltje a9da046f83 gizmoduck: row 307 (#1814, Vikunja 598) new tests, red first: row 270's ledgers per reason read the same count and the next collection goes through the harness module, core refuses an answer out of contract, pi answers what base scanPi read, claude-code answers harness-unsupported, a module without compactions refuses missing-export; capture.mjs checks the literals against base 1e629c892 2026-10-23 09:37:49 -05:00
jason.woltje 4a71b56abc gizmoduck: row 307 design r1 notes: seat suite whole and compactions-launch T3 run-not-skip in the gate (Duckworth C1, C2), launcher closure comment on bypassing call() (N1), line refs :301 and :302 (N2) (duckworth-0280) 2026-10-23 09:37:49 -05:00
jason.woltje 5208894196 gizmoduck: row 307 (#1814, Vikunja 598) design: harness export compactions({ store, cursors, budget, streams, first }), interfaceVersion stays 1, core keeps the ledger, readLedger reads the reason, ledger compatibility table, four old-test edits for Mos (sage-3440) 2026-10-23 09:37:49 -05:00