Repository navigation
feat(t27os): the Queen band draws the recorded log, labelled 'source: recorded log, not live' (Refs #12806) - #13009
Conversation
PR DashboardGenerated at: 2026-10-31 14:22:02 UTC
Summary
Seal Status
|
PR DashboardGenerated at: 2026-10-31 16:25:22 UTC
Summary
Seal Status
|
PR DashboardGenerated at: 2026-10-31 16:30:01 UTC
Summary
Seal Status
|
PR DashboardGenerated at: 2026-10-31 18:54:14 UTC
Summary
Seal Status
|
PR DashboardGenerated at: 2026-11-01 03:33:14 UTC
Summary
Seal Status
|
PR DashboardGenerated at: 2026-11-01 09:27:00 UTC
Summary
Seal Status
|
…urce: recorded log, not live' Rebased onto current master; conflict in glass_window.t27 resolved (Queen band keeps QGP_LABEL_RECORDED, [32]u8 buffer, recorded-log tests; master's key-age provider test preserved). Part of PR #13009.
5714005 to
ad0ec68
Compare
PR DashboardGenerated at: 2026-11-07 18:24:36 UTC
Summary
Seal Status
|
… (#13293) Master e599065 (#13009, ad0ec68) changed specs/t27os/glass_window.t27 without re-saving its seal, so Seal Coverage and the Spec Guards seal-currency step listed it on master and on every open PR. Re-saved with t27c seal --save from a T27 OS shell-cell pane, using a t27c built from cb1c1d6, the last bootstrap/ change on master. Tests 103/103. The four gen hashes equal the current= prefixes of master Spec Guards run 42402028424. No spec edit, no baseline entry, no bless. Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
What
Minimal visual slice of gap G2 (specs/t27f/live_actors_loop.t27), owner Pareto cut 2026-10-31.
After the merge of master (the lane's band, tap job, first-frame gate and startup hunks landed there
through #13017), what this PR still adds is one thing: a ring that begins with TAP_DEMO or TAP_REPLAY
(a recorded log) carries exactly this label where "DEMO REPLAY" stood:
Honesty rule of specs/t27os/live_petal_plan.t27: a recorded log is never drawn as live, and a live
ring never carries this label.
Scope
specs/t27os/glass_window.t27(+46/-5):QGP_LABEL_RECORDED(30 bytes),the label buffer widened 16 -> 32, and two Section Q tests,
the_recorded_log_label_is_exact_and_fits_its_bufferandonly_a_marked_ring_earns_the_recorded_log_label.State, stated plainly
fix(t27os): the window no longer traps at startup (key-age words overlapped the orbit's spans) #13016 (same spec, same seal) and chore(seals): re-seal t27os FrameGate, GlassWindow, QueenDash (Refs #12972) #13026; each merge of master brings master's GlassWindow seal, which
this slice's label change makes stale. The seal was re-saved in a T27 OS shell-cell pane with
t27c-bc19b54e6: tests 103/103 pass,
seal --verifyreports all hashes MATCH, no home path. The deltaagainst master is two files:
glass_window.t27and its seal.t27c check specs/t27os/glass_window.t27gives Typecheck OK (0 errors, 69 warnings).2-3 s after launch at
brk #0x1(Overflow) in_main, before the first frame; so did build 7652. Thatcrash was master's own and is fixed by #13016,
now merged here. No build of this head has been run, so the band is still unseen.
gitleaks,own-language,gate-topologyandspec-guardsdie compiling committedgen/c/...with-Wincompatible-pointer-typesas an error: #13054(
ubuntu-latestmoved to ubuntu-26.04). The same checks are red on unrelated PRs (spec(t27os): ladder_serve, the t27-native serve loop of the Ladder (Refs #13039) #13048, chore(ci)(deps): bump actions/create-github-app-token from 2 to 3 #13052, spec(t27f,physics): research_graph SPEC row + executable heat bench (#13045); renderer not connected #13053).spec-guardsstops at its first step, so its seal-currency step did not run; locallyseal --verifyof the GlassWindow seal says all hashes MATCH.Corpus ratchetnames a new conflictedtype
ParsedCommand(defined inspecs/tri/collections/namespace.t27and a port spec, neither touchedhere) and is red on spec(t27f,physics): research_graph SPEC row + executable heat bench (#13045); renderer not connected #13053 too.
Check L1 TRACEABILITYfails intools/l2_regen_check.py --listin CI(exit 1 after 0.2 s); the same tool exits 0 here against fresh master, and the check is red on spec(t27os): ladder_serve, the t27-native serve loop of the Ladder (Refs #13039) #13048 and
spec(t27f,physics): research_graph SPEC row + executable heat bench (#13045); renderer not connected #13053 as well. No fix pushed: the repair is in CI config or the C emitter, not in this slice.
tri landcannot run on this Mac ("tri: cannot run 'land' -- t27c is not built.",exit 2). CI is the check for the rest.
Draft. Auto-merge off. Not a merge request: the owner decides when this leaves draft.
Refs #12806, #12827, #12780, #12779
🤖 Generated with Claude Code