Files
BFM-decomp/config
Drew T 3251a06003 fix(config): RESTORE overlays.mk — deleted AGAIN by wave ei's gate commit
Second registry wipe today. commit:2911 ('ox wave ei overlays — 1 banked') committed
config/overlays.mk as a zero-line file, exactly as commit:2863 did this morning, and the
consequence was identical: no overlay can build, so every gate rejects every draft. Waves
ei/ej/ek/el/em gated ~1,100 drafts and banked 2 between them, with 675 backlog rows reading
'match_one MATCH but the whole-binary gate rejected' — the local oracle proving the drafts
were byte-correct while the tree could not build them.

WHY config_sane() DID NOT FIRE. I added it at 13:0x. The gater process has been running since
midnight and executes the code it loaded at startup — Python does not reload, and a
long-running lane never re-invokes its own module. The runbook already states this ('tool code
next invocation, lane args/env only on a fresh shell') and I applied it to the drafter and the
maintenance lane today while leaving the gater untouched. A guard that is not running is not a
guard.

Restored from commit:2911^ (5,083 lines, 141 binaries). The gater is restarted in the same
change so the protection is actually live.
2026-08-25 17:43:55 -06:00
..