FLS11 — the cache CANNOT leak across users (3.3), proven at the mechanism level. The
server's slots are full-width and SHARED (no FLS fingerprint segment); per-user narrowing
happens at read time on every hit and every miss. So: the writer's cold query writes a
slot; the reader's identical query is served WITHOUT a new slot write (shared slot — the
exact surface a leak would use) and STILL comes back without the denied column.
FLS11 — the cache CANNOT leak across users (3.3), proven at the mechanism level. The server's slots are full-width and SHARED (no FLS fingerprint segment); per-user narrowing happens at read time on every hit and every miss. So: the writer's cold query writes a slot; the reader's identical query is served WITHOUT a new slot write (shared slot — the exact surface a leak would use) and STILL comes back without the denied column.