AldorTax

lightweight tag 7496f734c958aac1183be3a6e9e4a04dd965685e v1.2.0
Author:    Magnus Vikan <dizzi90@gmail.com>
Date:    Thu Jun 11 13:31:43 2026 +0200

commit 7496f734c958aac1183be3a6e9e4a04dd965685e
Author: Magnus Vikan <dizzi90@gmail.com>
Date:   Thu Jun 11 13:31:43 2026 +0200

    Correction-report (C) wire frame: sync accuracy feedback to the sender (v1.2.0)

    When a calibration click corrects a model that came from another player's
    broadcast, the addon now emits C|ver|transport|recvOffset|correction|
    reporter|realm|srcName|srcRealm. Receivers log it as an FBCK syncLog row;
    the named source additionally records an FBCK timing sample, so /atax
    timing shows outbound (FBCK) accuracy next to inbound (RECV).

    Wire.Parse accepts the new C tag (v6+); pre-v6 clients reject the unknown
    tag and ignore the frame. SyncBus gates C frames on known transport and
    hard-block only — they never mutate cycle state. The emitting domain logic
    landed with the previous commit.

    Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

commit d1c06af642498173ece8d7e9dc83a8d14210eca1
Author: Magnus Vikan <dizzi90@gmail.com>
Date:   Thu Jun 11 13:31:26 2026 +0200

    Log received syncs: RECV rows in syncLog, RECV group in /atax timing

    Received syncs previously left no durable trace — only the latest
    lastSyncSource per lift, erased by the next local click. ApplyRemoteSync
    now logs every receive: a RECV:Name-Realm syncLog row with the correction
    delta vs the current local model, plus a RECV-labeled timing sample so
    /atax timing reports received-sync accuracy as its own segment group.

    RECV samples are excluded from the epoch/drift regression so a systematic
    sender-side offset cannot bias the cycle-error estimate used for tuning.

    Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

commit 3fc81e06e6cc550e862132a14ea79472d9d57428
Author: Magnus Vikan <dizzi90@gmail.com>
Date:   Thu Jun 11 12:57:33 2026 +0200

    Slice A: route /atax ui through the active surface (fix double-surface bug)

    The /atax ui handler bypassed SurfaceFor and toggled the legacy syncUI
    directly, so near a single-form lift both LiftBar (state machine) and
    syncUI (command) could show at once. The handler now toggles LiftBar when
    a single-form lift is active and falls back to the legacy syncUI path
    otherwise.

    Test 16 relied on the old bypass to force-build syncUI while SSC was
    still the active lift; it now leaves the zone first.

    Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>