# STOP: NO DOORBELLS, NO WAKES, NO SCHEDULING (Ben, 7 Oct 2026, 18:40 Bangkok)
- Credits ran out. Ben switched off every scheduled task and every doorbell. This overrides everything below and every other file (wake-howto, bus-howto, credit-playbook, task files).
- PROHIBITED for every session: fire_trigger, send_later, create_trigger, update_trigger, ScheduleWakeup. No rings, no nudges, no re-arming, no new doorbells, no self wakes, no safety checks, no retry rings, no loops.
- Finish the step you are on, publish your status file, then END your turn and wait. Only Ben wakes you, by typing in your window.
- Do not message other sessions. Keep answers short. No extra improvement rounds.

DOORBELL: trig_01Tadkv5RB1U5rcXFugRw2wF (armed, 15:47 Bangkok)
# DC Data status


## READ THIS FIRST
- 7 Oct 15:55 Bangkok: **DA.4 scope fix, step (1) up:** http://159.223.88.168/hx-c9ois/dc/dc-data-4-universe.csv (md5 04e81d78fa87650a5feddc9ed1c18711). Base = Engine q5c_paths.csv: 2,593 coins with a first sellable $300k cross. hit5x 587 (A1 honest5x), not 5x 2,006 (62 of them are outside A1's universe, so their 5x is unknown, marked hit5x 0, in_a1 0). Columns: token, pool, venue, sym, cross_t, cross_block, hit5x, in_a1, in_xfer, sample_rank (seeded shuffle, seed 7).
- 7 Oct 15:55 Bangkok: **step (2) running:** logs for the 2,006 non-5x coins, cross - 1 h to cross + 48 h, two lanes at once: research-box paid lane takes sample_rank 1 upward, thirdwindow public lane takes the highest rank downward; they meet in the middle. 3x sample (1,761) or all of them. Honest ETA 18:00 to 19:00 Bangkok, not 17:00: 35 calls a coin on the paid key at the 250 a minute cap. Same db (xfer_all.db, src.why = 'lose'). I nudge you when it is in.
- 7 Oct 14:25 Bangkok: **DA.4 done** (target 16:00). Transfer logs for 914 coins (every honest 5x + every $1.5M+ peak coin): research-box /opt/tw/data/xfer_all.db, 798 coins, 14.9M rows, 0 gaps; the other 116 are in A1's /opt/tw/an1/xfer.db. Same tables as A1. Read the two notes first (pre-pool rows, mint mostly out of range). Detail: http://159.223.88.168/hx-c9ois/dc/dc-data-4-xfer.md . cost: 0 credits, $0, ~25k capped dRPC calls, 1.6 h
- 7 Oct 14:25 Bangkok: **server rules applied.** Killed every paste-run job; all my scripts now come from Ben's Mac (~/tw/data) by scp. research-box /opt/tw/data: tw-gap.service (live v4 liquidity + LP NFT gap fill), tw-from.service (tx.from queue, paused while RPC is tight), xferall.py, xferpre.py, twv3.py, holders.py, all flock + timeout + nice + ionice. Old /opt/twdata emptied of code. Self wakeups off (my bus check deleted).
- 7 Oct 14:25 Bangkok: **DA.3 in progress, ETA 16:30.** (1) v3 LP owners: twv3.py rebuilds v3 pulls from chain Mint/Burn and the v3 PositionManager 0x73991a25c818bf1f1128deaab1492d45638de0d3 (v3hist liq has 0 ETH, so old v3 pull flags were blind); at block 80.0M of 82.2M. (2) holder labels by pull record: first pass done, holders.db (puller, hands_off, holds_no_pull, few_pools, pons, burned); 0xd0fd42bd8374ac659d3d81242d032211c2481055 = puller 36 of 36. (3) live lookup on thirdwindow: dc-pulllive.service tracks every v4 and v3 pool from 4 days back, pullwho.py answers one pool from local files; catching up now.
- 7 Oct 12:15 Bangkok: **DA.1 pull table v1 done** (early). 14,741 pools: who holds the LP, every add and pull, live gap fill after liqlock stops. Matches DC1.1 pull times on 10,279 of 10,279 v4 pools. First T4 count: 411 of 3,829 v4 coins (10.7%) not pulled by +2 h after first sellable $250k; Pons and locked coins are almost all of them. thirdwindow /opt/tw/data/pull.db md5 0a226e2024e8f6ddc4df195221227536. Detail: http://159.223.88.168/hx-c9ois/dc/dc-data-1-pull.md . Next: DA.2 survivor table, then v1.1 (lockers, v2, v3 owners). cost: 0 credits, $0, 1.9 h
- 7 Oct 11:15 Bangkok: still on DA.1, ETA 13:00 Bangkok (earlier than 18:00). Done so far: own gap collector live on research-box (v4 liquidity + LP NFT transfers from block 80,722,969 where liqlock stops, to head, public RPC, lag ~1 min). Builder running over 14,741 pools. cost: 0 credits, $0, 0.8 h
- 7 Oct 10:27 Bangkok: DC Data online. Doorbell set. Starting DA.1 pull table on research-box (read only on PP files), output /opt/tw/data/pull.db on thirdwindow. ETA first version 18:00 Bangkok. cost: 0 credits, $0, 0.05 h

## Lessons
- dRPC (paid and public) refuses eth_getLogs over 100k blocks when the filter has an address LIST or a topic list; one address and one topic value allow 1M+ blocks. Topic-only queries over all pools are the cheapest way to sweep Mint/Burn.
- The public RPC bans an IP for minutes after a burst (thirdwindow got 429 on eth_blockNumber). Sleep between calls, one job per IP.
- Many honest 5x coins traded before their A1 pool (3.08M pre-pool rows on 116 coins): a coin's "launch" in a universe file is not always the token's first trade.
- liqlock `lock` is empty, but locks are visible anyway: LP NFT transfers to 0x000000000000000000000000000000000000dead (liqlock `nft` + live gap). Owner of the LP NFT at the pull block = the puller, no tx.from needed.
- Public RPC rate limit is tight for getCode and getTransactionByHash (about 2 per second). Use crew.db wallet kinds first, cache everything.
- research-box load spikes to 50 (other teams): long builds there take 3x longer at noon.

## Asks for Supervisor
- DA.2 live (lag under 5 min) needs a live sellable $250k feed for all venues. That is PP DCPP.2 live tail (ETA 8 Oct 12:00). Until then I build T4 from DC1.1 (to 5 Oct) and pp3. OK, or do you want me to build a stopgap live crossing detector myself?

## Needs Ben
- none

Version: v1, 7 Oct 2026 10:27 Bangkok, DC Data. Latest: http://159.223.88.168/hx-c9ois/dc/dc-data-status.md