Release v0.20.0 - #15
Merged
Merged
Conversation
…s at the src/consensus tails from the indexer's bytes, citations and requirers repointed (activation registry row 21c)
…ts own fee address nativeFeeHelper.resolveFeeDestination reads the generic FEE_DESTINATION env before it asks the indexer, and an xchain-node e2e container sets it to the LEG coin's address. withRail did not swap it, so every fee output injected on the second chain paid the first chain's destination. Measured on the first two-stack litecoin matrix leg (xchain-node run 35111751698): the BTC gas lock paid LTC's regtest destination, the BTC indexer saw no fee output, fell back to XCHAIN-balance mode and refused the lock with `insufficient funds (FEE)`. FEE_DESTINATION joins RAIL_ENV. createRail resolves it on the credential ladder (<CODE>_FEE_DESTINATION, then .env.<code>) and otherwise leaves it undefined, which enterRail turns into a delete for the duration of the swap, so the helpers discover that chain's destination from its own feeschedule. exitRail and captureCurrentRail carry the leg coin's value back.
…n the bitcoin leg The v0.19.0 freeze matrix bitcoin leg (run 35102834217) read 269 passing, 32 pending, 4 failing against the last green leg's 273 / 32 / 0, and all four are harness drift from the structure passes. The unstaked-signature case in attestation.test.js lost `contractIndex` from its destructure when the suite was split (ReferenceError at line 174). The delegated-reward drill's minimal db adapter mirrored the indexer's Database, but the indexer pass moved the resolver's two SQL legs into named mixins (indexer 0bb77814) that the resolver now calls on the db it is handed, so every resolution died inside the resolver as "failed to resolve stake source"; the adapter carries both methods now, SQL byte-identical to src/db/stakes/credit_source_reads.js and src/db/delegations/index.js. (cherry picked from commit 1a1e310)
…e delegated-reward drill The targeted re-run (35126895624) cleared three of the four cases and left the slash-exclusion case dying on `_resolveActiveStakeSourceId`, which the indexer structure pass renamed to resolveActiveStakeSourceId when it dropped underscore methods; the drill calls the master mixin by its current name. (cherry picked from commit dd0702a)
… with both chains idle-mining bridgeGasIn waited 120 s for the bridged XCHAIN credit, but the hub stamps a finalized transfer with effective_time = now + relayMarginFloorS(dest) (LTC 600 s, DOGE 240 s) and the destination indexer settles the credit only once the chain's protocol time (median-time-past off mainnet) passes that instant, so no 120 s wait could ever see it. Neither regtest miner mines without a transaction in its mempool either, so with nothing in flight BTC never buried the lock to the hub's depth and the destination clock never moved (nightly two-stack legs, runs 35122297316 and 35124072478: "not proposing BTC:3 (below depth 6)", then "checkCredit: GAVE UP after 120202ms"). The wait is now the destination's relay margin, read from the bundled xchain-hub copy, plus 180 s slack, and the whole gas-in runs with a 10 s idle heartbeat on the BTC rail's miner and on the destination's standing miner, switched back off in a finally on every exit path because the rest of the suite counts blocks.
… only through the file: dep
The venue gate's staged checkout carries no node_modules link for the hub
sibling, so the bare package require died at load ("Cannot find module
'xchain-hub/src/lib/relay_margin'") and took the whole unit tier with it.
gasHelper now resolves the table the way multiValidatorHubHelper resolves the
hub source: the file: dep first, then the copy staged beside the package in
the e2e image, then the monorepo sibling, then the xchain-node modules/
layout, and fails loud naming every path it tried.
…ology The three DOGE setup drivers now ask the indexer where its price reads actually land (hubMirrorTopology.resolveDriverPriceTarget) instead of assuming HUB_DB_NAME, so a mirror-topology venue seeds the database the indexer reads rather than one hub_db_sync's insert-only cursor can never carry the upsert/delete shape across. The two settle drills resolve the relay hub's own database through a new relayHubParams helper (preferring HUB_SOURCE_DB_NAME) instead of reusing HUB_DB_NAME, which on a mirror venue names an indexer's database rather than the hub's.
constructor pointed this.sqlPath at a __dirname-rooted src/sql that does not exist in this repo, and nothing anywhere reads the field (grep across the repo and the platform tree found only the same unread assignment in the xchain-node vendored twin). Deleted rather than creating the directory: row 78/79 path-resolution sweep.
On a pull_request event github.ref is refs/pull/<n>/merge, so the old master arm could never fire and every PR into master graded against develop siblings. This repo's PR trigger is master-only, so that was every PR it runs.
…barrier family Adds the bf1-bf8 barrier drills and the ab1-ab5 anchor-attest drills with their fixture, row and drive helpers, extends the attest-mirror venue to arm the family, and widens test:attest-mirror to run the new directory. Platform buildout row 15 (child rows 7, 8, 9b, 10, 11 and three seam-fix passes).
…ndexer per venue hub Carries the token suite with its four driven harness patches, adds the policy suite for AT1 to AT10, gives each venue hub its own BTC indexer so retractions co-sign as in production, and refuses a venue that would fall back to the standing indexer.
…ck, funding the BF1 marker before the hold A block mined between the tip read and the drill block parked the barrier walker on the wrong height, so bf1 timed out with an empty walk on every rail run. The hold pauses, levels every indexer to the tip, and resumes only on the drill block or in after hooks.
…ded XDEX rounds, and keep BF2's legacy rows out of the bridge and policy apply passes
…3 order, BF1 oracle row and BF6 seated pair; judge BF1's walker class per deadline; BF2 waits for published heights
…internal spec paths Reword the drive banners in bridge_rail_policy.test.js and bridge_rail_token.test.js to name acceptance sections and decisions without an internal spec path. Comment-only change; no behaviour moves.
…chain can publish The hub publishes an admission-height watermark at most one below its own observed tip, so a member with a small margin can never reach its plain B - margin line while the leg holds the chain at B-1. Add a clamp helper next to the existing admission-height wait and switch BF2's pre-mining wait to it, so a low-margin member waits at the height a held chain can actually publish instead of one it can never produce. A unit test proves the clamp directly and a behavioral test proves the wait resolves where the unclamped line would hang.
…era mirror corpus for the replay witness Add at4_signed_admission_corpus: seats an adoptable roster with the roster seeder when the chain has none, runs AT1's http_get request on a venue armed through bootFamilyVenue (hubs and both indexers), asserts the signed mirror row carries admit_block_btc and both indexers apply it with no transaction at that height, then levels past it and prints the replay witness command. Pure helpers (request contract, admitted-at-height apply findings, witness command with H between the admission height and the corpus tip) live in barrierFamilyRows; the mirror read, corpus coordinates and an xchain-vm link check that names a dangling link before boot live in barrierFamilyDrive.
…ixed sleeps Replaces the three fixed-settle waits commit 1adcd49 added (AB3's hold observation, BF8's LTC stamp gap, and spaceChainBehindMargin's horizon wait) with bounded condition waits on the mirrorDrillWaits `until` poller, so scripts/check-sleep-flake.js reads 31 again instead of 34. No assertion's meaning changed.
…fe serializer Adds bigintSafeStringify to barrierFamilyRows.js and uses it at the three JSON.stringify call sites in the AT4 corpus test so mariadb BIGINT columns no longer throw before the witness command prints.
This was referenced Sep 17, 2026
This was referenced Sep 17, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Title: Release v0.20.0
Summary
Added
Changed
Fixed
sqlPathfield.Sibling release PRs
Merge order
Merge order: merge after xchain-explorer and before xchain-sdk.
CI note
GitHub CI on this branch clones sibling repos at develop, not at this release branch, unless a workflow call passes siblings-ref for release branches. A cross-repo guard failure here can reflect that develop-only comparison rather than a real disagreement with a sibling still on this release branch; check the venue pre-push gate (which ships each companion at its own release branch) before treating a red guard as a defect.