Skip to content

Open questions backlog

CategoryVerificationStatusReviewed / PartialLast reviewed2026-09-12

This backlog tracks unresolved questions after the coordinated BitcoinII Core v31.1.0 documentation, architecture, wallet, mining, compatibility, node, release, and Developers / Source Atlas audits.

Most high-priority source-location/currentness questions are now resolved. The remaining work is primarily about executing controlled tests, gathering longer runtime evidence, validating third-party implementations, measuring live-network behavior, and turning provisional operational guidance into reproducible runbooks.

How can users authenticate all current v31.1.0 binaries more strongly?

Section titled “How can users authenticate all current v31.1.0 binaries more strongly?”

Known: Six assets are recorded; GitHub reports SHA-256 metadata for all six; the lightweight tag points to a GitHub-verified commit; MoreBC2 independently matched the Windows Qt archive hash.

Open:

  • independently hash the other five current assets;
  • determine whether a maintainer-signed checksum manifest or detached asset signatures are published through a canonical route;
  • document any trusted release-key distribution process;
  • determine whether reproducible-build evidence exists or can be produced.

Priority: High

What does a full/long-duration v31 node run establish?

Section titled “What does a full/long-duration v31 node run establish?”

Known: Bounded Windows Qt/server-mode mainnet startup, peer discovery, advancing IBD, cookie RPC, shutdown/restart, and disposable-wallet isolation succeeded.

Open:

  • complete IBD;
  • observe long-duration operation/reconnection;
  • deliberately exercise supported pruning/index combinations;
  • test a current headless-daemon deployment path if an appropriate release artifact becomes available;
  • document production-oriented fee-estimation and monitoring behavior.

Priority: High

Can MoreBC2 independently reproduce controlled candidate-time / target transitions?

Section titled “Can MoreBC2 independently reproduce controlled candidate-time / target transitions?”

Known: Current source review establishes activation, rolling baseline, short-horizon response, timestamp safeguards, emergency recovery, mining-template interaction, and header-sync history requirements.

Open:

  • deterministic candidate-time / nBits vectors;
  • abrupt hashrate-arrival/departure scenarios;
  • emergency-recovery threshold examples;
  • execution/mapping of relevant upstream tests;
  • empirical post-activation block timing/difficulty analysis.

See Research: difficulty adjustment notes.

Which third-party signing implementations correctly support the BC2 replay domain?

Section titled “Which third-party signing implementations correctly support the BC2 replay domain?”

Known: Mainnet activation 57750, domain 0x01324342, and the relevant wallet/raw/PSBT/mempool/block/cache paths are source-traced. Ordinary v31 PSBT signing/finalization was exercised on isolated regtest.

Open:

  • deterministic mainnet-domain sighash/signature vectors;
  • external/hardware signer qualification;
  • third-party wallet/library audit;
  • activation-boundary mempool/signing tests.

Do executed boundary tests match the source-reviewed rule paths?

Section titled “Do executed boundary tests match the source-reviewed rule paths?”

Known: Activation and the principal output/Taproot witness restrictions plus block-connection enforcement are mapped.

Open: execute the located tests and record representative activation-boundary accept/reject vectors.

How does current v31 behave under controlled competing-header/fork scenarios?

Section titled “How does current v31 behave under controlled competing-header/fork scenarios?”

Known: Two-phase sync, fork anchoring, bounded private ShockWave history, and exact-work validation are source-reviewed.

Open:

  • competing-branch/recovery runtime scenarios;
  • malformed/insufficient branch-history cases;
  • activation-boundary scenarios;
  • executed-test mapping.

What is the practical BC2 reorg/settlement risk model?

Section titled “What is the practical BC2 reorg/settlement risk model?”

Known: Chain selection is accumulated-work based. Reorg disconnect/undo/reconnect/mempool-repair behavior is source-reviewed. MoreBC2 uses 50 confirmations as a provisional normal-deposit exchange baseline.

Open:

  • empirical reorg-depth/frequency history;
  • controlled v31 reorg simulation;
  • a worked cumulative-chainwork monitoring example;
  • operational incident handling when a credited deposit is disconnected;
  • value/risk-based escalation thresholds for large or unusual deposits.

How should the provisional 50-confirmation baseline be maintained?

Section titled “How should the provisional 50-confirmation baseline be maintained?”

Known: CoinEx exposes 2/6; NonKYC exposes 50; NestEx exposes BC2 50; Biconomy’s exact current count remains publicly unverified.

Open:

  • periodic direct policy rechecks;
  • maintainer/community feedback if available;
  • chainwork-aware examples;
  • large-value escalation policy;
  • reorg incident playbook.

The question is no longer “what number?” but how to operationalize and revisit the provisional baseline responsibly.

Which current pool endpoints work end-to-end with v31 BC2?

Section titled “Which current pool endpoints work end-to-end with v31 BC2?”

Known: Published pool modes, fees, and endpoint families have been recorded. Core mining/template source is reviewed. generatetoaddress was exercised in isolated regtest.

Open:

  • subscribe/authorize;
  • job receipt/update behavior;
  • share difficulty and extranonce behavior;
  • accepted/rejected shares;
  • candidate-time / nBits compatibility;
  • block attribution;
  • payout accounting/timing/reliability.

Which mining RPCs have current release-binary runtime evidence?

Section titled “Which mining RPCs have current release-binary runtime evidence?”

Known: generatetoaddress was executed in the v31 regtest PSBT workflow.

Open: getmininginfo, getnetworkhashps, getblocktemplate, submitblock, and submitheader remain source-reviewed but not current-release runtime-qualified.

Is there a verified public valid-BC2 transaction submission path?

Section titled “Is there a verified public valid-BC2 transaction submission path?”

Known: Local zero-peer regtest sendrawtransaction worked. Three Mempool-style /api/tx routes rejected malformed 00 payloads; Official Explorer candidate routes returned 403.

Open: successful valid public submission/propagation remains unverified.

This should only be tested if needed, in a deliberate disposable workflow. Existing user wallets must not be used.

Which services are genuinely independent and reliable enough for production redundancy?

Section titled “Which services are genuinely independent and reliable enough for production redundancy?”

Known: Current point-in-time REST/WebSocket/Electrum behavior is documented; Official Explorer is materially distinct from the Mempool-style services.

Open:

  • operator/backend independence;
  • long-term uptime/reliability;
  • behavior under lag/stress;
  • wallet-level Electrum compatibility;
  • public transaction broadcast behavior.

Which third-party wallets/signers are safe for current v31 BC2?

Section titled “Which third-party wallets/signers are safe for current v31 BC2?”

Open: build a small compatibility matrix using disposable/watch-only fixtures where possible, with replay-domain support treated as a hard signing requirement. Do not use existing user wallets merely to generate documentation evidence.

What production architecture should a serious service use?

Section titled “What production architecture should a serious service use?”

Open:

  • wallet-based vs non-wallet deposit tracking;
  • txindex/indexing recommendations;
  • withdrawal signing/broadcast/monitoring;
  • hot/cold separation;
  • backup/recovery and key-management procedures;
  • reorg/deposit rollback handling;
  • chainwork/risk alerting.

MoreBC2’s current exchange pages are integration guidance, not a certified custody design.

What upstream BitcoinII contact route should integration providers use?

Section titled “What upstream BitcoinII contact route should integration providers use?”

GitHub Issues and a verified Core commit-author email exist, but MoreBC2 has not established a canonical upstream technical/security/integration contact process. Do not invent one.

What still needs recurring recheck rather than one-time source work?

Section titled “What still needs recurring recheck rather than one-time source work?”
  • current release inventory and metadata;
  • exchanges and their deposit/withdrawal policy;
  • explorers/APIs/Electrum;
  • pools and Stratum endpoints;
  • third-party wallets;
  • current Core release/source baseline;
  • public-site deployment and link integrity.

The large Source Atlas/Architecture currentness question is now substantially resolved; future work should be triggered by new releases or evidence gaps rather than blanket re-audits.

Status: Reviewed / Partial
Primary evidence checked: Completed v31 source/architecture audits, September runtime records, current release evidence, public infrastructure evidence, exchange evidence, and current mining/wallet/operator documentation
Notes: This backlog intentionally contains unresolved work. Source questions already answered by the current audit sequence have been removed or rewritten as test/runtime questions.