igneum-network/igneum
1
0
Fork 0
Code Issues Pull requests Releases Activity

Testnet go checklist: the object table on the 0.3.24 line (genesis 01294fd3..., digest b2e856ed..., 18 decimals, the switches as carried), the lineage, the runbook values; the cut's commit the shipper's

Browse source 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
This commit is contained in:
igneum-labs 2026-10-07 21:06:26 +00:00
parent 9c90fde741
commit 2186d8f278
1 changed files with 21 additions and 3 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

24
docs/plans/testnet-go.md
View file

@ -44,9 +44,27 @@ genesis payload on the final message, the scale factor tested known-failed first
release-0.3.22-node object (34a2dbaa, digest `87d103b6...`, genesis `52a3e6a9...`). Two orders ride with it: the testnet genesis carries
the FINAL message (the 5 October payload that `87617621...` carried, never the 4 October "proposed, not final" text that `52a3e6a9...`
hashes: a public node never starts on a proposal; the digest moves and the re-arm absorbs it), and the base unit is 18 by the project lead's word of 21:35 BST (the
8-decimal scratch form is no longer needed); nothing on a seed before the node lane names the cut and the build-server lane's dry run reads clean. This section's object table is rewritten against that object once the node lane names the cut; the gate rows 9g, 9h and
9j re-run on it; row 9c reads "re-arms on each cut". Everything below this paragraph in this section is the 7 October RE-CUT of this
lane, void as a go object by the ruling and kept as the record of what was built and gated.
8-decimal scratch form is no longer needed); nothing on a seed before the node lane names the cut and the build-server lane's dry run reads clean. The carry (22:1x UK): this lane's re-cut merged onto the 0.3.24 line as fork branch `testnet-genesis-3-node` 34892a36, staged by the node
lane as `tn-on-0324` 28d3dcab on c8f9b383 (clean), to land as 0.3.24's second commit after the v5 object commit (the shipper names it).
| Field | Value on the 0.3.24 line (confirmed from the staging binary by the node lane, 22:2x UK) |
|---|---|
| Network | `igneum-testnet-1`, chain id 4462 (never moves), address prefix `igneumtest`, ports 26810 / 26811 / 28810 / 26890 |
| Genesis | hash `01294fd322704dc28fbef0e7a5ef86d6ee260ac5efaf88891cdba661b5fd58ac`, merkle `bcd0e8fb1099aeb3cc1370b50e6dec9f615e204374b2097f25b2438e7d22f49f`: the FINAL 5 October message on the 16-byte subsidy payload of 18 decimals, timestamp 2026-10-05T00:00:00Z, bits `0x1d100000` |
| Consensus digest | `b2e856ed281a41171b594e7a542c719b68f508d34b2f46c004f5157d3df04342` (pinned by `igneum_testnet_identity`; the v5 object commit changes Devnet 3's params only, so it does not move; re-read from the cut's binary at the go) |
| Base unit, emission | 18 decimals (the project lead, 21:35 BST); `EmissionSchedule::TESTNET_1` at the unit |
| From genesis | difficulty v2 and v3, proving v0, finality v3, the DAA-second rule, the leave item (delay 3,600), the signing bonus 0 at 1,000 bps (no burn), the per-block subsidy, class v3 and v4 under the Devnet 3 shape's one-day signal window (byte 7 stamped from genesis), the ladder at rung 0 (window 86,400; 27/35/53 admissible, 88/173/267 not), the era VDF (scheme 0, the reference T), proving v1 with the fresh rule (the Devnet 3 shape), fees v1, consensus proof verification under the manifest's ids (shard `0x2b1a81cb...`, aggregator `0x474678f3...`), `sig_scheme` 0 with its switch, the W5 succession, the cache rung (512 MiB inadmissible) behind its switch |
| Held at never | the fork gate (window 600), the peer directory, the pool split, the class v5 floor (0 only if its Devnet 3 crossing reads clean, by the shipper's cut), exec restart |
| Pairing | igneum-pow at the class v5 freeze commit 1c420786 |
| Where | fork `testnet-genesis-3-node` 34892a36 (both node mirrors), staging `tn-on-0324` 28d3dcab (build-1), repository `testnet-genesis-2` (both repo mirrors) |
Suites on 34892a36 (bounded, build-2 and build-1): consensus-core 168, exec 47, p2p-flows 38, pow 19, kaspad 2; consensus 134 one test at a
time with the pre-existing m20 red (9f); the parallel lib run aborts in the test-order race class the node lane owns, noted, not chased
tonight. The digest lineage: `b7d8c915...` (1c19441d) to `87d103b6...` (the 0.3.22 interim re-arm) to `b2e856ed...` (this). The runbook's
values: `--genesis 01294fd3...`, `--digest b2e856ed...`, the seed-class binaries of the cut's commit (the build-server lane's build from the
named commit, the sha asserted), no override, `--wipe-genesis`. The gate rows 9g, 9h and 9j re-run on it; row 9c reads "re-arms on each
cut". Everything below this paragraph in this section is the 7 October RE-CUT of this lane on the 0.3.18 line, the source of the carry,
kept as the record of what was built and gated.
### The 7 October re-cut (void by the ruling; the record)