# Request to ppfeeds: hook code size and hash in v4tape (A1, 10 Oct)
Why: native H3000 unique-hook pools run a per-code-version clock (16,724 B: price peak at 640 s, pull at ~1,325 s; 17,916 B: peak 1,167 s; 21,824 B shared hook: dead). The paper arm must select the version at pool creation, and we need to see when a new version appears (template change = lifecycle warning).
What: in /opt/sw/v4tape/pools.jsonl rows (and the live H3000 / A1 candidate feeds) add fields hook_code_size (len of eth_getCode at the init block, bytes) and hook_code_hash (sha1 of the code hex, first 10 chars), plus hook_pool_count (how many pools in the tape already use this hook address). Cache by hook address, one eth_getCode per new hook (free rpcpool). For all hooked pools (H3000, A1 hooked fee100, any hook). Also a daily line in a small file hook_versions.csv: date, code size, hash, pools that day, first seen time.
Backfill: my table of 330 hooks (21 Sep - 9 Oct) is at research-box /opt/crew/work/analyst1/ba22_codes.json (hook -> size, hash); import it for history.
Test I run: forward H3000 native by code version: age-600 multiple per version, kill if a new version (not 16,724 B) shows peak-age IQR wider than 300 s.
